• Sales and Operations Planning Across the Supply Chain
• How to Resurrect a Failing System
• The Truth about Capacity Planning and Scheduling Systems
• A Master Scheduler's Dozen - 12 Keys to Master Scheduling Success
• Keeping the Promise of Available to Promise

Sales and Operations Planning Across the Supply Chain

Jack Gips, President
Jack Gips Inc.

For many years, sales and operations planning has proven to be one of the processes most critical to the success of manufacturing systems. It is management's means of driving the sales and manufacturing organizations toward their goals for customer service, financial results, resource utilization, human resources, and inventories. It is a monthly forum for discussing the state of the business and for comparing the operating plans with the annual business plans. It is a mechanism for bringing together the right people and the right information to make good business decisions on a timely basis.

In recent years more and more companies have begun to view their businesses in a different way. The supply chain view considers not only manufacturing, but the distribution and sales organizations it feeds and the purchasing and other internal manufacturing organizations that deliver materials and goods to it. A complete supply chain view would also extend out to the external customers and to the suppliers of the suppliers. For purposes of our discussion here, we will limit ourselves to a discussion of the internal supply chain starting with the primary suppliers and ending with our sales organization's performance for our customers.

Prior to adopting the supply chain view, many management teams made independent decisions and applied independent measurements to sales organizations and manufacturing plants. The annual planning exercise and informal management decision processes were the means of bringing all the sites into alignment. The importance of the supply chain view rests in its ability to give management a way to blend all these internal goals and decisions into the best set of drivers for the whole enterprise. It surfaces conflicting objectives and allows management to trade off performance in one node of the supply chain for the benefit of the entire chain. It brings to light redundant inventories and activities to more efficiently position inventories and resources. It leads to shorter cycle times and increased flexibility for the customers.

Sales and operations planning across the supply chain requires great coordination. All sales organizations must deliver their demand plans at the same time for the distribution centers and/or manufacturing plants to start their planning activities. Manufacturing feedback to the sales groups must arrive quickly to ensure they make valid promises and allocation decisions to satisfy the customers. Executive management's role is to provide strategy and direction at the start and approve the plans at the end of the sales and operations planning flow. The entire flow must take place within the period of a month and mesh with the master scheduling processes at each of the supply sites.

Supply Chain Example
For purposes of our discussion we will assume a global supply chain consisting of fifteen sales groups fed by three distribution centers which are in turn supplied by ten manufacturing plants. Assume that the sales groups have local warehouses and maintain inventories of the finished goods that are sold in their locations. The distribution centers carry the main supplies of these inventories and are the consolidation point for all of the demands that are sent to the manufacturing plants. Some goods can be sold into multiple markets while some are made specifically for individual markets.

Supply Chain Questions
In order to determine the sales and operations planning flow and responsibilities, the day-to-day planning responsibilities must be defined. The following questions must be answered:

• Who creates the sales plans that drive replenishments?
• Who maintains the on hand inventory balances at each level?
• Who is responsible for the levels of inventory at each site?
• Who plans the replenishments for each level?
• Are finished goods shipped to the sales groups from distribution centers, manufacturing plants, or both?
• Do the distribution centers sell any products directly to external customers or always ship to a sales affiliate?
• Who determines the shipping schedules to the sales affiliates and distribution centers?
• Are products made only at one plant or can they be sourced from several plants?
• How are safety stocks determined and are they maintained at each level?
• How is customer service measured at each level and who is responsible for it?

The answers to these questions will vary from company to company. In our example we will assume that the sales organizations are responsible for the forecasts and sales plans. The distribution center planners receive all the demand statements and plan the replenishments to both the sales groups and to the distribution centers from manufacturing.

Supply Chain Issues
Consider some of the issues these questions raise.

At the top of the S&OP structure there must be a sales plan to drive the process. This may start out as a pure sales forecast, but it should also take into account management's goals and budgets for the sales organizations and actions like promotions, price changes, and discounts to make them happen. Should manufacturing or distribution people set these plans or second guess and change them?

Where in the supply chain should sales and operations plans be created? Should S&OP meetings be held at each manufacturing site monthly, or should there be S&OP meetings at the distribution sites as well? Should there also be S&OP meetings at the sales affiliates? If so, what would they discuss?

What about the timing of the meetings? If one feeds information to another should their timing be coordinated in a single global sales and operations plan calendar?

Should critical resource issues at one site be considered in the initial planning of a site that is their customer in the supply chain before demands are passed down the chain? Or, should all demands be sent down and feedback returned once the supplier site reviews its resources against the total set of demands?

Should the executive sales and operations planning meetings occur at the start of the supply chain process to drive it, or at the end to accumulate the results and review them?

Some companies initially want to keep sales and operations planning as a completely independent activity. They want the S&OP output at one level of the supply chain to become direct input to the S&OP the next level down. This ignores the detailed decisions, policies, and inventory adjustments that are made between the levels. It also ignores the exaggerating affects of lot sizing, lead-times, and safety stocks at each level. It may not seem like a big assumption, but by the time you get from the top of the supply chain to the bottom, requirements and supply plans can be very different in both quantity and timing. It has been demonstrated in many cases that driving the sites at the bottom of the supply chain directly with the forecasts at the top does not lead to good supply chain performance. For that reason, planning should follow the pattern of S&OP to MPS to MRP (if required) at one node of the supply chain to S&OP to MPS to MRP (if required) at the next node. Figure 1 shows a Supply Chain S&OP Flow.

FIGURE 1

Guidelines for Supply Chain S&OP
Here are some guidelines for the supply chain process:

• The sales and marketing organizations should bear the ultimate responsibility for providing an accurate and timely demand statement. They are closest to the customers and the competition, and have the best knowledge to ensure the accuracy. They know their own local market issues and decide on actions to satisfy their own goals and measurements.
• All challenges to the demand statement they provide should be made before it is "officialized" in the affiliate demand meetings. Some companies have created demand management organizations, or forecasting councils, to challenge and validate the demand statements within the sales organizations before they are passed to the suppliers.
• Once the official demand statements are generated, no one else should adjust them at any point in the supply chain. Only executive management and the sales/marketing people should change these plans.
• The sales and marketing organizations should be measured on and held accountable for the accuracy of the demand plans.
• The demand statements must be completed and consolidated early in the month (first or second week) to support the needs of the rest of the supply chain.
• If the planners who plan the replenishments to the sales affiliates' inventories are part of the sales organizations and located at the affiliate sites, then a full S&OP process including supply planning meetings is warranted. These meetings should set the supply plans that drive the affiliate master schedules. These master schedules are, in turn, sent to the distribution centers as demands on their inventories. Other things that might be considered in these meetings are the local financial plans, inventory levels and warehouse labor and space requirements to support these plans.
• If these planners reside at the distribution centers, then the affiliate processes may be limited to demand planning activities only and not require a full S&OP process. If this is the case, a distribution requirements planning activity or (affiliate master schedule) should be created to plan material shipments from distribution and arrivals at the affiliates. If there is a standard shipping arrangement (like one delivery by ship per month), this should be accounted for in the planning of this master schedule. This schedule should be visible at all times to the affiliates even though it is planned for them by planners at a central distribution point.
• Central distribution centers are excellent points in the supply chain to consolidate demands and schedules, apply controls to manage capacities and inventories, and stabilize the orders that drive the manufacturing sites. Customer service and manufacturing performance can be greatly affected by the planning that is done at this level. If the planners use good judgments in reacting to exception messages calling for schedule change, respect time fences and capacity constraints, and manage their inventories well, the entire supply chain will perform better.
• A distribution center S&OP should consolidate the requirements from the affiliate master schedules and treat these as the demands on the DCs. This will account for lot sizing, safety stocks, and container loads that determine the shipments to the affiliates and make them different from the forecasts. This demand is the input to the S&OP meetings and a supply plan is the output.
• If a single distribution center is the single source of demand for a product family that is produced in a dedicated work center, that resource can be considered in the DC's resource requirements planning and S&OP. If demands come from more than one distribution center, or the product family is produced in multiple plants or in a work center that is shared by multiple families, it is better to do the resource planning at the plants and feed back problems to the DCs.
• The supply plan should drive the DC master schedules that call for replenishments from the plants. These master schedules may be delivered to the DCs, or the shipments may go directly to the sales affiliates if the products are unique to a particular market. These master schedules become the demands that are consolidated for sales and operations planning at the plants. They are also the demands that consume the production forecasts in the plants' master schedules.
• S&OP at the plants should not be focused on challenging the demands. It should be heavily concentrated on creating production plans to satisfy the demands and resolving resource constraints. If these constraints cannot be resolved, immediate feedback should be provided to the DC planners to adjust the demands.
• The production plans resulting from the plant S&OP meetings should be disaggregated from the family level to the end item or SKU level to drive the plant master schedules.
• If the plants are customers of other plants, or if there is another level of plants in the supply chain that supply them with components or raw materials, another level of S&OP and master scheduling may follow.

Management's Role
Executive management plays a critical role in this supply chain process. They obviously have to review the aggregate results of all these plans and agree that they satisfy the goals of the business. If not, they should drive adjustments to the plans through the supply chain in the next round of planning. This is a variation of a role with which most executive management teams are comfortable. It is common for them to review last month's actual results against their objectives and business plans, and react to unacceptable situations. The difference in the supply chain S&OP approach is that they can now look at projections and take actions in advance to avoid these unacceptable results. It is the difference between being proactive and reactive managers.

The best management teams take this proactive stance a step farther. Instead of reviewing the projections of the supply chain planning at the end of the process, they become the drivers and influence it at the beginning. They begin to do their planning earlier. They will inform the organization that they will be asking for increased year-end or end of the quarter sales volumes sooner than the month before they are required. They set policies for adjusting inventory levels before they become too high rather than after. They relax measurements on one part of the supply chain to get better performance from the rest of the chain. They sort out measurements that cause behaviors that are detrimental to the overall performance of the chain.

For example, there are businesses in which every plant is held to the same standards of plant utilization and efficiency. This may cause a raw material site to produce large quantities infrequently to maximize their efficiencies and lower their costs per unit. The impact on the rest of the supply chain may be slow turning inventories, larger than necessary lot sizes, and inflexibilities for their customers. The next tier of plants may have difficulty achieving their own efficiencies because of the way their supplies are delivered. The net result is that, instead of the customers determining what the plants should produce, the raw material suppliers determine it.

The most important management issue is whether they see their role as one of making decisions, driving the business, and measuring performance to plan, or one of reviewing results and reacting to them after the fact?

Synchronizing Supply Chain S&OP
This type of communications flow causes the need for highly synchronized planning and a supply chain planning calendar to time the S&OP activities. A typical supply chain calendar may look as follows:

FIGURE 2

It is important to separate sales and operations planning decisions from master scheduling decisions. Most of the time S&OP decisions are family level decisions and relate to longer term resource constraints. Master scheduling decisions are more likely to be item level decisions or decisions on individual orders. They may be tied to the availability of specific materials or shorter term capacity constraints in individual work centers. Inside the time fences, it is a good idea to keep changes to S&OP's production plans and supply plans to a minimum. It should take a conscious decision in an S&OP meeting to change these plans. Master schedule changes inside the time fences can be based on individual customer demands or manufacturing problems. Master schedulers make these decisions whenever it is possible to satisfy the customers. Sometimes this causes a trade-off of one schedule for another, so one change can actually result in several master schedule changes.

There are two key measurements on this planning that affect the supply chain. One is the master schedule summary which tests whether the sum of the master schedules for a family equals the production plan or supply plan for that family over a specific period of time. This tells you whether all the daily master scheduling decisions are causing a violation of the S&OP plans. The second measurement is master schedule stability. This tests the amount of schedule change that is being passed down the supply chain.

S&OP Meeting Topics
There are a number of sales and operations planning issues that cross the levels of the supply chain and should be communicated through this information flow. Some examples are:

• New product launches and product changes
• Major process changes and start-ups
• Products with seasonal demands
• Price changes
• Promotions, one time deals, discounts
• Supplier capabilities and limitations
• Allocations of limited production items to customers
• Inventory reduction programs
• Plant sourcing and outsourcing decisions

In some of these cases, executive management should provide directions to the supply chain. In other cases the decisions should be worked out within the supply chain S&OP processes. Whenever decisions are made that affect multiple nodes of the supply chain, they should not be made unilaterally at one node without consideration for the others and the supply chain as a whole.

The ultimate measures of supply chain performance such as customer service and financial results to plan are dependent on the quality of the plans that are created from the top down in this S&OP flow. They also depend on everyone's execution of the plans from the bottom up. Measurements such as production plan achievement and master schedule conformance are critical to ensuring delivery to the plans at all levels. If the goals are well coordinated and everyone treats them seriously, the teamwork in the supply chain will bring excellent results to these very complex businesses.

The supply chain view adds new perspective to the way management views and manages a business. Sales and operations planning is management's tool to manage the business. The combination of these makes the decisions, information, measurements, and systems work together to achieve better performance than ever before.

BIOGRAPHY

Jack Gips is President of Jack Gips, Inc., a firm that provides high quality consulting and education to manufacturing companies.

Jack has spent 33 years in Manufacturing, as both a practitioner and a consultant. He has helped companies in many diverse industries improve their performance in manufacturing.

Jack has a B.S. and an M.B.A. degree from Case Western Reserve University. He served as Chairman of the 1977 APICS International Conference. He has served as a member of the APICS Curricula & Certification Council, Subcommittee for Production Activity Control, and the JIT Subcommittee. He is certified at the fellow level and speaks at numerous professional society meetings.

He has authored numerous articles and was Editor of the "Capacity Planning" chapter in The Production and Inventory Control Handbook.

How to Resurrect a Failing System

(Manuscript for upcoming presentation at the APICS Congress for Progress Conference in Baltimore)

Jack Gips, President
Jack Gips Inc.

There are thousands of manufacturing systems running on the world’s computers today. Since the 1970’s, a large percentage of the systems have been built on standard software purchased from a wide variety of vendors. Over these years, the technology has changed and improved dramatically and the software has become increasingly sophisticated. So, it would be logical to assume that most of these systems in place would be successful in operation. In fact, there are many more successful systems operating today than there have been at any time in the past. This is a tribute to both the quality of the software and the knowledge of what it takes to be successful that has developed through time. Unfortunately, there are also a great number of systems that run on our computers, but whose output is ignored in actual practice. This paper is dedicated to the companies trying to deal with these ineffective systems.

What determines success and failure?

There are many symptoms that can make success or failure evident.

1. The bottom line for any system is whether the people in the organization trust it enough to make daily decisions directly from the system’s information and tools.

	Are they processing a lot of data in, but using only a little of the information coming out?
	Are there many non-system tools and reports in place that substitute for the formal system’s tools and reports?
	Do key people have their own unique sources of information to make decisions?
	Is the data accurate enough for the system to make valid recommendations?

2. Are many of the modules that were originally purchased actually not in use?
3. Is the system used to "track and transact" but not to "plan and execute"?
4. Does Management use the sales and operations planning tools to make their business decisions?
5. Are the scheduling and capacity planning tools used in daily practice?
6. Do the users like working with the system and believe that it really helps them, or do they feel like "slaves to the system"?

How did it get this way?

A system can be a rousing success or a dismal failure right from the start.

1. If its beginnings were a "software only" implementation, with little attention paid to the data, policies, practices, measurements, and preparation of the users, it was headed in the wrong direction from day one.
2. If it was installed in six months under the pressures of a Y2K or similar deadline, chances are that the implementation was incomplete.
3. If the initial focus was on software modifications to support the current way of operating, the users may have automated non-working processes. This is often the result when "software selection" is the first step taken in implementation.

Other systems start off successfully, but fade over time because the structure to maintain and sustain them was not established.

1. If a strong training and education program is not in practice, people moving in and out of key positions may not receive the knowledge to keep the concepts intact.
2. If there is no organization in place to protect the integrity of the concept, the users may ask for changes that violate its original intentions.
3. If they ask for improvements and the IT organization does not respond, they may build a new set of informal systems using spreadsheets to get what they want.

If these are the symptoms, how can we determine the problems?

The best way to determine the real problems is to assess and measure the system’s performance. The standards for measurement have been well established over the years. There are many checklists available to evaluate the performance of manufacturing systems. Most of these focus on many of the same practices, measures, and goals. The best of these checklists evaluate processes like sales and operations planning, master scheduling, and purchasing through interviews, reviews of reports and on-line screens, and watching the actual practices. They also apply the results of basic measurements against their goals. These measurements should address data accuracy (like inventory accuracy), quality of planning (like master schedule stability), and quality of execution (like on-time delivery of supply orders). The measurements must be at the diagnostic level and not the bottom line measures (like inventory turnover) if you wish to identify the real problems.

The existence of these measurements is, in itself, one measure of the status of the current system. If the measurements are already collected, even if they show poor results, it is an indication that many things were done well in the initial implementation. If they do not exist, it raises a big question about the start.

The measurements lead to the improvement project.

Once you review the processes and take the measurements, the major tasks to improve the system’s performance can be defined. If inventory, bill of materials, or routing accuracy is below par, there will be a task to attain the required levels. If plans are past due, overloaded, unstable, or inconsistent at different levels of the planning structure, there will be task teams assigned to do root cause analysis, design the solutions, and implement them with the users. Schedules that are valid but not followed, output hours that do not meet capacity plans, and schedules and orders that are not delivered on time also lead to major tasks. The checklists and measurements determine the targets of these task teams and later become proof that the problems are getting resolved and the system is becoming successful.

Why does it have to be "a project"?

Why not take the measurements and just make it the users’ jobs to reach the goals as part of their daily activities? In most cases when a system goes bad, even the good parts are dragged down by the rest. It is not unusual for people to give up on a failing system. It typically takes a concerted effort to create an atmosphere of change and make it happen. Sometimes these projects turn out to be as comprehensive as the initial implementation minus the selection and implementation of the software. So we make it a project to give it the structure, importance, and emphasis that it needs to get everyone involved and accomplish the goals in a reasonable timeframe.

How is the project organized?

This is a big project that is worthy of a full project organization. A Management steering committee is needed because the project requires their attention. The steering committee’s role is to:

	Set priorities
	Provide the core team with clear direction
	Provide resources to ensure the program is a success
	Hold the core team accountable for accomplishing the implementation on time and within budget
	Review core team proposals on a regular basis and judge the feasibility of the plans in relation to all other operational activities , as well as to the vision of the results
	Ensure all levels of management maintain a high level of commitment and enthusiasm throughout the lifetime of the project
	Gain sponsorship from all other functional groups , divisions , vendors and customers
	Mentor specific teams

There is a core team or project team that is responsible for the overall activities of the project. Their role is to:

	Develop and implement the project plan
	Track key implementation issues through each phase of the change
	Use a set of ongoing measurements to determine the effectiveness of the implementation
	Identify implementation problems
	Report progress to Steering Committee
	Define education and training requirements
	Review and present policies, procedures and performance measures for approval
	Coordinate task team activities and assure consistency between teams
	Participate as team members/facilitators
	Assure user participation and cultural change

The task teams are managed by the project leader and the core team. Their job is to do the data gathering, root cause analysis, solution design, implementation of the solution with the users, handoff to the users, and follow-up to assure success. A typical set of task teams may include teams for:

	Inventory Accuracy
	Bills of Material Structure and Accuracy
	Demand Management
	Sales and Operations Planning
	Master Scheduling/Materials Planning
	Purchasing Data Accuracy
	Routings/Time Standards/Lead-time Accuracy
	Shop Floor Control/Capacity Planning
	Product Changes/New Product Introductions
	MRP II Performance Measurements
	MRP II Education/Training

The initial measurements discussed above usually determine which teams are necessary for a given company.

The task teams follow a basic task structure to fulfill their roles:

1. They are given team education that covers

	ERP concepts related to the project
	Their specific team charter
	The current status of their part of the system based on the measurements

2. They are given the high level project plan and are asked to develop a detailed implementation plan for their task including sub-tasks for Logic-Data-Policy-Process-Performance-People issues
3. They design and develop the systems and process solutions for their task
4. They gain design approval from

	Core team
	Steering committee

5. They develop the supporting policies and procedures
6. They provide education and training for the users
7. They implement the systems, policies and procedures
8. They implement missing performance measures and automate manual ones
9. They handoff their task and results to the users
10. They work with the users to validate that the new design is working and to achieve Class A results

The core team and task team are also responsible for creating an infrastructure to sustain and improve the system for the long term. This may involve:

1. On-going performance measures and a measurement management process

	Regular performance measurement meetings
	Annual audits

2. Setting up a system integrity group to review...

	System changes and additions
	Growth of informal systems

3. An on-going education and training process that assures a high level of knowledge among the users into the future including...

	Education and training policy
	Matrix of positions and training topics that relate to them
	Training documentation and measurements
	Development of training materials
	Qualification of instructors
	Exposure to new ideas

What are the critical success factors?

The greatest obstacle to bringing a system back from the brink is the attitudes of the people in the organization. If they dislike the system because of its past performance, or if they believe it can never be made to work, it will be a great struggle to convert their thinking. The idea of resurrecting it must be sold rather than delivered by edict. If they like the informal systems they have authored to supplant the system, they must be convinced to give them up. If they take a passive "wait and see" stance, they must be convinced to join the effort as active participants. Without the support of the users there is no chance to succeed.

Another obstacle is the temptation to make this another "software" project. The whole idea of this project is to make an existing system succeed, not to buy a new system. Certainly, there may be software tasks to fill the gaps or modify some features of the system. This will be especially true if the original system has been modified drastically because of its past performance or the lack of understanding of the organization. However, the focus must not be primarily on the software. It must be on the data, policies, practices, performance measurements, and people because these are the areas of opportunity.

A final obstacle is the approach to the performance measurements themselves. Some organizations feel threatened by these measurements. This happens because measurements may have been misused in the past. If they were used to cast blame on individuals rather than to constructively improve performance of the business, there may be a fear of the same happening again. If so, Management has to convince the organization by word and deed that this is not the case.

The measurements are also the true test of commitment in the improvement process. When the new practices are in place, the scores will not necessarily be at their goals. Then it is time for the users to dig in and raise the scores through hard work and additional changes to practices. If they are not committed, they will challenge the measurements and the goals. It will be important for the steering committee, the core team, and the task teams to hold their ground and help them move to the next level. The measures are relatively standard and there are many other companies that achieve these goals. There is no reason to lower the goals, loosen the tolerances, or change the measurements. After all, the objective is not to get a certification. It is to become a better business by developing a successful system.

Successful results lead to more successful results

Once the project is over, the view of the system is quite different from the original view. The organization has learned to trust and use the information from the system. The informal replacement reports and data are gone. The users are committed to maintaining timely and accurate information in the system because they understand the importance of doing so. There is a structure in place to continue the measurements and assure that performance does not deteriorate in the future. There is usually a new attitude toward formal systems and greater ownership among the users.

In addition to the improved performance on the diagnostic measurements, the bottom line measures will have improved dramatically as well. If customer service was poor, it will have improved. If it was already good, it will still be so, but operating costs and inventory turns will have improved. If there was chaos in the organization and an inability to predict results, there will now be order and predictability.

The best indicator of the results is the large number of companies in which these improvements were accomplished by one plant or division with such remarkable results that the others have decided that it is worth their efforts to do the same. Good results that spread up and down the supply chain can have compounded benefits for any business.

The Truth about Capacity Planning and Scheduling Systems

Jack Gips
President, Jack Gips, Inc.

INTRODUCTION

The emotional debates wage on. Optimal (finite) scheduling zealots demonstrate the amazing calculation capabilities of their software and how it automatically does things that traditional scheduling modules can not. Members of the traditional sect ridicule the optimizers for their high cost, over use of precision, and constant need for recalculation to arrive at the "perfect" plan. Unfortunately, these debates are focused on the negatives and do little to help practitioners gain positive ground on this area of historical weakness.

The strengths of the traditional scheduling systems are their top down approach and their ties to management's philosophies on the relationships between customer service, revenue driving, and internal capacity utilization. Their weaknesses are tied to their lack of good visibility into the capacity relationships between work centers, their inability to simulate changes before they are made, and their lack of tools to realign schedules.

The strengths and weaknesses of many of the finite scheduling systems are in many ways the exact opposites of those of the traditional systems. Their calculation and 'drill down' capabilities are superb. Their simulation and schedules provide outstanding visibility. However, their ties to management's thinking are tenuous and their focus on detail at the lowest levels limits the value that they add to our decision processes.

Instead of bickering over the negatives and arguing over which is the best, why not combine the best of both and take a step forward in scheduling and capacity planning?
We finally have the software capabilities to create schedules and capacity plans that users can trust. Why not apply those capabilities to support the users in their decision making?

CAPACITY PLANNING FUNDAMENTALS

Our history of capacity planning is filled with examples of companies that installed software with the right intentions, but whose mind sets and informal policies were so different from the rules that guided the software that they were never free to properly utilize the tools. Agreements on the limits of capacity have usually taken a back seat to the need to service the customers, and rightfully so. Simply loading customer orders into a schedule, however, provides no guarantee of delivery. This guarantee can only come from assuring the availability of material and capacity and then producing the products on schedule. Disregarding capacity limits leads to poor customer service, higher manufacturing costs, higher inventories, and less flexibility. The assumption that capacity is so elastic that it can be stretched to satisfy almost any level of demand has gotten numerous companies into trouble. The perceived solution to these problems often is overly simplistic . . . If we just had a better set of tools; everyone would be willing to use them. So, instead of addressing the policy and rules conflicts, the endless search for the perfect logic becomes the direction. Some people even believe that the right software will force management and others to change their thinking and conform to the system's rules.

When discussing capacity planning, it is usually better to begin with fundamentals and leave the software solutions to the end. Here are some basic issues that determine the effective use of the capacity planning tools. How well the software helps us deal with these issues determines its success.

Is there a plan? Do we make every effort to validate the plan? Is it based on a complete and accurate statement of customer demand? Does it consider the availability of materials and capacities? Is it doable? Do we make every effort to achieve the plan? Do we measure performance in meeting the capacity plans, material plans, and customer promises? Is our actual performance against the plan taken into consideration when we create tomorrow's plan, thereby validating its integrity?

When planning capacity, is the objective to meet the plans or change the plans quickly and conveniently? There has been a great deal of attention paid to creating systems that automatically and instantaneously change plans. If the same efforts were made to measure and manage excellence in meeting the plans, the results would undoubtedly be more beneficial.

When it is necessary to change plans, is capacity considered before the changes are approved and promised to the customers? Is there an implicit assumption that if material is available, capacity will be made available somehow? Is each decision made on it own merits or is the overall impact on capacity reviewed?

Do people make the decisions to change or is it done automatically by the computer? In most companies, compromises must be made between maintaining perfectly precise schedules and maintaining schedule stability by avoiding unnecessary changes.

What is the value of maintaining precise schedules when they are based on imprecise data and assumptions? Lead-times, time standards, queues, demonstrated and finite capacity numbers, loss factors, productivity ratios, and routing/work center relationships are all based on estimates, averages, and historical trends. They are all used to plan future schedules and capacity plans. They are all subject to variations, changes in performance levels, and choices made by manufacturing people. Should the schedules be changed every time there is a variation? Or, should these schedules be targets for manufacturing and changed only when the tolerances set around them are exceeded? What is the value of a plan that continuously changes when it is missed? Should it result in a new plan or actions to meet the original one?

If the original plans are set based on one set of guidelines (i.e., demand statements and customer service goals), but intentionally violated based on a second set of guidelines (manufacturing objectives to optimize productivity), have you set yourself up to fail? Why not determine the real guidelines and priorities before planning and then stick closely to them during execution?

Does your business try to accommodate last minute customer orders? If so, do you reserve capacity and material for these so that you will not have to violate capacity plans or displace other customers? Is the compromise between reserving capacity, safety stocking inventories, and providing quick response to the customers well understood and backed by sound policies?

Is there agreement that schedules and capacity plans that are overloaded, front loaded, and scheduled past due are not valid? Will actions be taken to avoid these conditions and to fix them if they occur? Is there a structured approach to these actions? Are there rules that define what actions to take to adjust capacity levels and who makes them? Is it clear when to change the capacity and when to change the plans?

Is the system's data accurate and trustworthy? Is there a good definition of the capacity available in each work center? Are the time standards complete and set consistently? Is shop reporting timely and accurate?

Are manufacturing's issues factored into the plan? If you overlap operations, perform simultaneous operations, campaign batches, sequence orders a certain way, run items with similar setups together, etc., does the scheduling account for these and create initial plans that do not have to be violated and realigned once they reach manufacturing?

Are there good performance measurements on the delivery of hours of work into the work centers and output of hours from the work centers compared to the plans? Performance in capacity can only be improved if there are measurements in place to identify problems and fix them. If actions are taken to adjust capacities, are there measurements that review whether those actions provided the predicted results? If hours are delivered on time, then component orders can be delivered on time, and products can be delivered on time.

Is there a structured process for planning capacities and reviewing the performance measurements? Are there regular meetings to discuss work center capacities and loads? Is there a formal way to evaluate schedule changes and the actions necessary to meet them?

It is important to recognize that our systems built with sound logic have never been able to override management's decisions even when they seemed irrational. It is therefore critical to deal with the philosophy issues above before determining the approach to capacity planning and the tools that should be implemented. Many companies have proven that buying the tools to force the philosophy on management results in unused tools and wasted time and dollars.

TRADITIONAL SYSTEMS

The capacity planning tools that have been offered as part of MRP II systems over the past thirty years are Resource Requirements Planning, Capacity Requirements Planning, and Input/ Output Reporting. Resource Requirements Planning is linked with Sales and Operations Planning. It is Management's tool to assure there is enough capacity available to support their proposed production plans. It ties capacity to product families for critical work centers. Its purpose is to identify the actions necessary to get load and capacity reasonably in balance with each other and prevent "infinite" loading. Once capacity is in the "ballpark", the production plans drive the master schedules and material requirement plans.

Capacity Requirements Planning is a more detailed planning tool. It is used to validate the manufacturing schedules. It considers specific items' schedules, nets out existing inventories and work in process already completed, and accounts for lead-time offsets. It employs assumptions based on history and current measurements for lead-times, time standards, productivity factors, queues, etc. It matches capacity requirements with planned capacity availability, usually in graphical form, to highlight time periods of over or underloading that require smoothing or actions to raise or lower capacity. It provides information about the detailed orders that create the capacity requirements in specific periods for realignment.

Input/Output Reporting measures the flow of capacity requirements into a work center and the flow of output from the work center. These are compared to the plans for input and output to separate capacity problems from other problems that cause missed schedules. Proper use of Input/Output Reporting leads to quick reactions to capacity problems to avoid late deliveries. It also measures whether actions taken achieve their expected results.

Basic Concepts

Any approach to capacity planning has to be based on a general philosophy supported by consistent concepts. Some of the basic concepts behind the traditional approach to capacity planning are:

• The planning has to be driven from the top. Management's high level sales and operations plans must assure the availability of resources to give manufacturing a chance to succeed. Resource Requirements Planning tools give management the information they need to assure resources are in the "ballpark".
• The company's objective is to satisfy its customers' demands as often as possible. If customer demands exceed available capacity, the first course of action is to find ways to expand capacity. If these demands cannot fit within the limits of the available capacity, or if adding capacity is impractical, the customer demands will be prioritized and realigned to fit.
• Plan the master schedules to support the forecasts and customer demands. Then make the component schedules support the master schedules.
• Schedule to customer needs first, and then adjust the schedules to optimize manufacturing.
• Schedules are targets that must be met within a tolerance. It is not necessary to react every time a schedule is not precisely met. It is desirable to stabilize the schedules and react only when tolerances around the targets are exceeded.
• Create schedules using standards. Measure actual performance compared to these standards. React to variances that exceed the tolerances. The key performance measurements to monitor are:
  • Input/Output vs. plan
  • Actual queues vs. planned queues
  • Actual delivery vs. scheduled dates
• Allow the computer to schedule orders outside the time fences and lead-times. Planners must manage the schedules inside these lead-times to assure validity and stability. It is not necessary to smooth the capacity plans outside these lead-times. It is only necessary to be in the "ballpark".

Weaknesses

The major weaknesses of the traditional approach are:

• Realignment of individual schedules to maintain schedule validity and smooth short-term capacity plans is usually a manual operation. In companies with complex scheduling relationships or many orders, this can become overwhelming.
• When the plan is to meet the due dates by expediting and re-prioritizing individual operations, there is a need to compress queues to maintain schedule validity. Only a few of the traditional systems provide automated help in queue compression.
• The logic to optimize manufacturing efficiency and combine similar setups is missing from most traditional systems. They require complicated modifications to enable sequencing.
• Although most have developed logic to plan overlapping and parallel operations, the logic to plan campaigns is another modification that is generally not available.
• There is little automated assistance to do capacity smoothing even for short-term schedules.
• There are limitations in planning for multiple constraints such as equipment, labor, and tooling simultaneously.
• When manual changes are proposed to deal with capacity problems in one work center, there is little capability to see the impact of these changes on other related work centers until after the changes are made.
• Traditional systems do not prevent overloading in individual time periods. Resource planning prevents overloading in total over extended periods. Although this approach does not allow "infinite" scheduling, it does create "lumpy" schedules that must be smoothed.

OPTIMAL (FINITE) SCHEDULERS

These are, in many ways, the exact opposites to the traditional methods of scheduling and planning capacity. Their logic is designed to eliminate schedule delinquency and prevent overloading. The earliest versions were strictly finite schedulers and were automatically calculated with little human intervention once their rules were set. They often failed because they had to be shut off to override their decisions.

Today's new technologies have paved the way for the development of "optimal" schedulers. These systems can use finite scheduling logic, capacity requirements planning logic, or both applied in different time frames.

They allow human intervention in the form of manual overrides to individual orders, the ability to freeze schedules in specified parts of the horizon, or changes to the scheme of priorities.

They are very flexible in their rules and constraints. They can simulate results before changes are made. They provide graphical views of the schedules and capacity plans, display the relationships between orders and loads and between work centers. "Drag and drop" graphical capabilities make their schedules easy to manipulate.

Many of these have also been designed to handle the sequencing, overlapping operations, parallel operations, campaigning, etc., that have been missing from many of the earlier systems.

Basic Concepts

Despite all the desirable features described above, optimal schedulers will fail to satisfy their users if they are applied without a philosophy supported by consistent concepts. Some of these are listed below:

• "Finite" scheduling loads manufacturing orders into the work centers in a pre-determined priority sequence until the load hours in any time period reach a "finite" level of capacity. Once this level is reached, additional orders must be loaded into the next time period with available capacity. It "forward schedules" from the current date until all orders are scheduled. This approach eliminates past due schedules and prevents overloading work centers in any time periods.
• "Optimal" scheduling may start with a finite method, but allows a planner to override the decisions of the finite scheduler when necessary. It allows adjustments to the "finite" capacities, the priority scheme for loading orders, and even the time periods and work centers in which the finite logic is applied.
• The focus is on capacity and the optimization of manufacturing. Deliveries are promised only when there are enough materials, lead-times and capacities available to do so.
• Many of the assumptions and averages used in the traditional methods are replaced by calculated numbers based on current schedules. Lead-times and queues, for example, are calculated based on the predicted loads and priorities each order will face as it moves through the plant.
• More emphasis is placed on manufacturing's considerations like sequencing. The objective is to create a set of schedules that manufacturing really will follow.
• If the capacity available in a work center changes, it is important to realign the orders to maintain the validity of their dates and to utilize capacity well.
• Schedule stability is secondary in importance to schedule accuracy. Frequent realignment is key to maintaining this accuracy.
• The computer should do as much of the scheduling as possible. A great amount of manual intervention significantly reduces the value of the tools.

Weaknesses

The major weaknesses of the optimal schedulers are:

• In many situations, capacity is not "finite". It is extremely flexible. Management is often more inclined to stretch capacity than to risk disappointing a customer with a promise date later than the requested date. Under these conditions, it is difficult to treat precisely calculated schedules as real objectives.
• It is difficult to determine the real priority schemes to apply to the orders. In many companies, there are a number of people who set and override these priorities on a case-by-case basis.
• Like the traditional methods, these systems rely on assumptions such as time standards, productivity factors, available capacities, etc. Even the calculations that replace the lead-time and queue assumptions are based on these factors.
• These systems calculate their plans very precisely. If their data is not accurate or if the plans are not followed, they re-calculate another set of precise plans that are different. Plans that change every day are not really plans at all. This approach may be useful for realigning very short-term schedules, but it is questionable when applied longer term.
• These systems schedule using calculated queue times based on orders predicted to be in the queues. Unreliabilities in manufacturing, such as equipment downtime or absenteeism, cannot be absorbed when queues are planned this way with no tolerances. The system reacts by rescheduling, making schedules unstable and changing delivery dates.
• The ability to manually intervene is both an advantage and a weakness. If there is too much intervention, these tools cannot serve their purpose. This makes users carry the burden of scheduling at a very detailed level. It violates the basic concepts. Some companies have even converted their "finite" schedulers into "infinite" schedulers by overriding their finite capacities with larger numbers to accommodate more customers.
• The rules and data that must be maintained in the "optimal" schedulers can easily become too complex and confusing to the users. Once this happens, they are likely to set aside these "sophisticated" tools for ones they can understand that require less effort.

CONCLUSION

The dilemma facing most users is how to take advantage of the capabilities of the tools and technologies available to us. Installing capacity planning systems is not the goal. Using them to make better decisions is. We believe it has to start with an internal discussion about the company's philosophy related to capacities and customer service. There must be agreement on the rules and policies from Management down through the manufacturing people who assign orders and people to work centers. Management, instead of being the primary violators of the capacity plans, must become the leaders and enforcers of the new philosophy.

There must also be agreement on the data to be used and responsibilities for keeping them timely and accurate. Responsibilities for setting plans, overriding them, and executing them must be clarified. Performance measurements must be generated to assure conformance to these agreements.

Once all of these agreements are made, then it is time to determine the best tools to fit the company's needs. Between the traditional and optimal methods, there are certainly enough good tools to support these needs today. A combination of the two applied to different products or different periods in the planning horizon may certainly be appropriate. The key is to first understand and adopt the philosophy, and then apply only those tools that best support the philosophy.

A Master Scheduler's Dozen - 12 Keys to Master Scheduling Success

Jack Gips
President, Jack Gips Inc.

The master schedules are the main links between most companies' long-term plans and short-term schedules. They are the point at which forecasts meet actual customer orders, where manufacturing budgets meet customer service requirements, where management bravado meets capacity limitations. They are the few driving schedules that lead to the many detailed activities of production, purchasing, and all their support services. If the master schedules are well managed, the component schedules, and work center capacities will usually be under control.

When you structure your master planning system, you make some key decisions that determine how well it will work in the future. We will describe 12 key features that, when properly applied and linked together, create a master scheduling process that fits an individual company well. Attendees will learn about the 12 features, the implications of ignoring them, and what happens if they are used properly.

Master scheduling has long been the cornerstone of the decision making, responsiveness, and management of our manufacturing systems. In the earliest manufacturing systems, we attempted to manage all the detailed parts. This led to heavy expediting, poor inventory control, and customer service problems. We learned over time that the best results came from managing the master schedules and reacting by exception to the need to manage individual component parts. We also learned that our scheduling systems must support the logic master schedulers want to use. They must be set up to drive the activities of master scheduling, not to create obstacles for the planners.

The issues that follow are some of the keys to effective master scheduling.

• Production Plan - Master Schedule Link
  Sales and Operations Planning is the process for creating a game plan at a family level to reach the "best" blend of customer service levels, inventories, capacities, financial performance, and human resources in the overall results of a plant, division, or company. The outcome of Sales and Operations Planning is a formal sales plan and a formal production plan. The importance of S&OP is now well recognized and many companies have adopted it as a key monthly planning activity.

Problem
Once the S&OP plans are created, the key is to link them directly to the master schedules. This is the crucial point at which many manufacturing companies and software vendors have a missing link. Instead of driving their master schedules with production forecasts disaggregated from their S&OP production plans, they load their end item sales forecasts into the top lines of their master schedules. If the production plans are exactly equal to the sales forecasts, this approach works well. When S&OP decisions are made to level production, raise or lower inventories, or build anticipation inventories for seasons, promotions, or shutdown periods, the production plans should drive the master schedules.

Implications
Loading the sales forecasts causes the advice of the system to plan master schedules that ignore the S&OP decisions. The exception messages tell the planners to move orders to dates that match the customer orders rather than the production decisions that were made in the Sales and Operations Planning meetings. To plan according to the decisions that were made, the planners have to ignore the messages and firm plan MPS orders against the system's advice. Now the system is fighting the needs of the planners rather than supporting them. This can result in master schedules that violate the S&OP decisions, extreme effort on the part of master schedulers to use the system, and switching off the exception messages.

Solution
Create the system links and scheduling practices to drive master schedules with the production plans. Make the master scheduling system work for you. This requires a place to store the production plans (usually in the forecasting module or at the family level in an S&OP module) and a method for disaggregating them and loading into the master schedules.

• Time Fence Rules
  Good planning at the master schedule level means adherence to time fence policies. Time fences are among our most misunderstood concepts. They are supposed to represent the time it takes to produce a given product including the purchase of raw materials and components, the manufacture of components and subassemblies, and final production or assembly. These time fences can be reduced by shrinking the lead-times of the items on their critical paths or by stocking them in significant quantities. Edicts or wishful thinking cannot shorten them.

The purpose of maintaining time fences is to give warning when scheduling decisions are needed that will affect manufacturing or purchase orders already in motion. They set the boundaries between orders that need schedulers' judgements and those that the computer can automatically plan. The rules that are generally applied are that master schedules inside the time fences are firmed and manually planned based on exception messages. Orders outside the time fences can be left to the computer based on the planning policies that have been loaded. This allows the master schedulers to concentrate on the most important orders that have the greatest impact on manufacturing and the outside suppliers.

Problem
Many companies ignore time fences, artificially shorten them, or take shortcuts in calculating them. It is common for companies to understate their time fences because long fences imply inflexibility for sales and customers. This does not fix the problem. It simply removes the warning and creates a false sense of capability and loss of control of the outcome.

Some of the software has rules that do not respect the boundaries. It is not unusual to see automatic scheduling inside the time fences or system-planned orders inside of firm planned orders.

Implications
These problems lead to master schedules that are not valid because the computer logic is very literal. It will schedule the impossible, drive delinquent component orders, or expedite orders to replenish small quantities of safety stock. Master schedulers spend inordinate amounts of time undoing what the system has created. Customer service and bottom line results are not dependable.

Solution
Lead-times should be measured for manufacturing and purchasing. Time fences should be calculated from these lead-times and adjusted based whether actions have been taken to make long lead-time items readily available. Each master scheduled item should have its own, independently calculated time fence. Orders must be firmed to cross into the period inside the fences. The system must not automatically plan orders inside the time fences or earlier than orders that have been firmed or released.

• Ability to Reject Exception Messages
  One important feature of our manufacturing systems has been the capability to plan by exception. Items that have been planned and require no further changes need not be reviewed. Those that have possible availability changes are identified to the planners by exception messages from the system for review.

Problem
In most systems if a planner accepts a recommendation by taking the recommended action, the message does not reappear in the next planning cycle. If a message is ignored or no action is taken, it reappears. The problem comes when a planner reviews an item and rejects the message or intentionally decides to take a different action than the system recommends. Many systems treat a rejected message as one that has been ignored. They repeat the message continually until another change occurs or the order closes.

Implications
This means the planners receive messages they have already handled in addition to new messages. It becomes difficult to sort the new messages from the old ones and the pile grows larger. This can result in switching off the exception messages and reviewing all the items instead of just the exceptions.

Solution
There are two common methods of dealing with this problem. You can run net change MRP instead of regeneration. Most net change logic involves the creation of an activity file that determines which items are reviewed. Removing the item from the activity file eliminates the message from the next planning cycle. It will take a new change to put the same item in the activity file for review again. The problem with this is that net change does not work properly in many of the software packages. Although it is offered as a feature during the software selection process, many vendors recommend that you do not actually use it. They suggest that regeneration is safer and that it plans differently than net change.

The second method is to create a suspense file that stores the rejected messages. When MRP regenerates, the system reviews the new exception messages against those in the suspense file. If they are identical, the messages are suppressed.

• Forecast Consumption
  In most of today's master scheduling systems, demands consume forecasts over a period of time. Since they are usually the result of monthly plans broken into daily or weekly increments, it is important to avoid reacting when short-term demands do not exactly match these incremental forecasts. Forecast consumption serves the purpose of stabilizing master schedules by assuming that forecasts are more likely to be accurate over a month or two rather than on a specific day or in a specific week.

Problem
Most software consumes all items over a globally set forecast interval. There is no allowance for different items' forecasts to be consumed over different periods of time. Some items may have better accuracy over a one-month interval and others over a two-month period. In some cases demands that exceed the forecasts are consumed into the future, but demands that are below the forecasts are dropped immediately.

Implications
The basic idea is to ensure that forecasts are consumed over the period of time that represents the most accurate forecast even though the master schedules may be scheduled to a precise date. If this does not happen the demand will change too frequently and, in turn, drive excessive master schedule changes. Consumption rules should work in both undersell and oversell conditions because push out reschedules are not necessarily less costly than pull-ups.

Solution
The interval for forecast consumption should be set by item rather than globally. Forecast accuracy should be measured for each item at one, two, and three-month intervals to determine which yields the highest levels of accuracy. System logic should be set up to consume both over and undersell conditions over this interval.

• Replenish Safety Stocks at the Time Fences
  Safety stocks are planned to cover for demand and supply variability. In most companies it is acceptable to drop below safety stock levels to some degree. If a company performs so well that they never drop below the safety levels, then there is no need for safety stocks in the first place. If only a small portion of the planned safety stock has been utilized, there is no need to panic and replenish by expediting except in unusual cases.

Problem
Standard logic in many systems treats safety stocks as requirements in the first time period of the planning horizon. This creates exception messages to expedite very short-term orders when projected inventories drop below safety stock levels.

Implications
Planners must reject these messages and determine where to place the replenishment orders to avoid expedite situations and schedule instability.

Solution
Safety stock replenishment orders should be planned at the time fences and allowed to progress at full lead-time. Planners should manually expedite exceptions such as critical items or items whose safety stocks have been depleted by unacceptable percentages. This can be accomplished via system logic or by temporarily reducing the safety stock planning factors and setting up safety stock replenishment requirements at the time fence.

• Demand Fences

Problem
There are some situations in which companies wish to forecast the sale of end items to drive manufactured and purchased components to stock, but do not want to convert the product to its finished, packaged, or final assembled configuration until a customer order is in hand. They may do this to keep their inventories in the most flexible position or in a more stable state until the last moment. Standard master schedules either drive products to stock against a forecast or wait for an order to be entered before buying or making anything other than safety stocks.

Implications
Customers may not be willing to wait the full lead-times for delivery. Converting products to finished goods too soon may make the company less flexible for its customers while raising total inventories.

Solution
Setting a demand fence at the lead-time of the finishing operations allows you to bring all components to stock based on a forecast, but only convert the finished goods for which you have orders. Some companies also use demand fences to bring in long lead-time raw materials, but not convert them to manufactured components until they have customer orders.

• Two-Level Master Schedules

Problem
If there are so many final configurations that it is impossible to forecast accurately at that level, it may be better to forecast at a semi-finished level that has fewer choices that can be converted to many different end items. This been recognized as an issue in assemble-to-order products for a long time and has been the source of much of the logic behind product configurators. There are also many companies that offer a wide variety of end items made from a smaller number of semi-finished items that do not need a configurator because they are not assemblies. They may offer already configured end items in a catalog, but not wish to stock them. They also want to keep inventories of the semi-finished items flexible and minimized.

Implications
Stocking all the catalog items would require significant safety stocks to achieve high levels of customer service. Total inventories would be high because the semi-finished items would be included in all of the finished item inventories. Producing only to order would not satisfy the customers' expected lead-times. Customer service would be negatively affected by the company's ability to forecast each of the end items.

Solution
Two-level master scheduling allows a company to forecast at the semi-finished level, but still retain already configured end items. The end items are initially planned for no demand and no supply orders. The semi-finished items are planned to satisfy their forecasts with lot sized master scheduled orders. When orders are entered for the end items, master schedules are created to drive the finishing processes. These master schedules drive component demands to the semi-finished master schedule level and consume the semi-finished forecasts. Available-to-promise information is captured by checking the semi-finished ATP and adding on the finishing lead-times.

• ATP for Line Items
  Customer service people are important users of ATP information along with the master schedulers. The way in which ATP is presented to these users often determines their ability and willingness to use the data.

Problem
Most systems present the information to everyone the same way . . . at the bottom of individual items' master schedule reports. Many customer service organizations take large numbers of orders for multiple line items. They need ATP in a form that helps them deal with these orders in a quick and efficient manner.

Implications
It takes too long for many of these people to look up ATP for items they are entering one at a time. If they cannot do this while the customer is on the telephone or at their computers, there will be little chance of negotiating deliveries before promises are given. The value of ATP will be wasted and customer service levels will be affected.

Solution
It is possible to move the ATP calculation to the customer service representatives' desks. Their order entry screens can automatically check ATP for each line item as orders are entered and provide valid promise dates to the customers immediately. Whole orders can be reviewed to determine if they can be delivered at the customers' requested ship dates, the need for partial deliveries, or the need to check with master schedulers to see if delivery dates can be improved.

• Multiple ATP
  When a manufacturing plant produces products that are sold in several markets it is often beneficial to sort the products that have been planned for each market into separate categories of ATP.

Problem
If two different marketing or sales organizations provide forecasts for an item, and one forecast is accurate while the other is not, the error in one has the potential to create poor customer service in the other.

Implications
Manufacturing may produce enough to satisfy both forecasts, but over-selling in the market with the poor forecast is likely to result in shifting products away from the market with the accurate forecast. If the inventory is treated in a first-come-first-serve manner, the poor forecasters may be rewarded while the good ones suffer. Manufacturing usually ends up with the blame.

Solution
Separate the ATP information for the two markets based on what they have planned. As orders are entered, consume the ATP for the market from which they originate. If one over-sells, ATP will show there is no planned product available, but they can negotiate for more product with the marketing organization that still has ATP quantities. This becomes a non-manufacturing decision. It puts emphasis on improving the forecasts to get better service. It reduces the chances of poor service in the market that is well forecasted.

• MPS Summary

Problem
Production plans created in Sales and Operations Planning drive the master schedules. At the master schedule level, detailed decisions are made about specific situations on specific items. The aggregate result of all these decisions may call for more resources than were originally budgeted in the resource planning during S&OP.

Implications
This opens the door for a disconnect between annual business plans, high level operating plans, and the detailed plans that actually drive sales, manufacturing, and purchasing.

Solution
A simple check on the variance between production plans and master schedules is to aggregate the master schedules into monthly quantities by family and compare to the family production plans that were created in Sales and Operations Planning. If the sum of the master schedules varies by more than an acceptable tolerance in a specific month or cumulatively over time, the master schedules should be brought back in line or the production plans changed to accommodate the difference.

• MPS Measurements

Problem
The master schedules are the key to creating high quality schedules at all levels. They must be valid in terms of lead-time, capacity, and stability. On-time delivery of the master schedules leads to excellent customer service and conformance to budgeted costs. It is not good enough to just practice master scheduling and make MPS decisions in the framework of a system. To get these benefits, it is important to measure performance in planning and executing the plans. Unfortunately, many companies practice master scheduling, but do not apply good measurements to their process.

Implications
It is not unusual for companies to report high levels of customer service. After all, there is an extreme focus on this measurement at all levels of the organization and even personal involvement in some customer relationships. There is a difference in the means of attaining this service from company to company. Some do it by planning well and executing their plans. Others do it by carrying high inventories to cover for poor planning and execution. Still others do it with heavy expediting and overtime costs. The difference between the first approach and the others is often a result of their measurements. When the focus is entirely on the final customer service measurement and not on the factors that contribute to excellent performance and eventually lead to excellent service, the result can be chaotic and costly.

Solution
Apply a few key measurements to the master schedules and you can achieve excellent service at a reasonable cost. These measurements are:

• MPS Stability - The number of MPS orders that change compared to the total number of MPS orders.
• MPS Summary - Does the sum of the master schedules for a family equal the production plan for that family? (Have we diverged from the resource plan?)
• MPS Lead-times - What percent of the master schedules are scheduled at less than lead-times (time fences)?
• MPS Execution - What percent of MPS orders are delivered on time and at full quantity?

• Master Scheduling Campaigns

Problem
Master schedulers in those companies that campaign products for process reasons often spend a great deal of time aligning and realigning campaigns, and orders or batches within campaigns, to deal with the needs of their customers and to utilize their capacities. It is difficult in many systems to make these moves without a lot of manual effort. Campaign planning often requires a more distant look into the future than planning individual orders. This often causes these planners to firm orders far out in the horizon, which also increases the maintenance burden. Some of the advanced planning and execution systems do a good job of campaign planning, but many of the traditional ERP systems do not.

Implications
Without some system logic to support campaign planning, the effort to manage the long horizons and many detailed orders may discourage planners from using the formal master scheduling system. It is not unusual to find that campaigned products are managed on spreadsheets, thereby depriving other system users of the information about the campaigns. This weakens their ability to use system information for forecasting, budgeting, and to track status of these orders through manufacturing.

Solution
Many companies have added logic in their master scheduling systems to handle campaigns, trains, product clusters, super-batches, etc. This logic separates the orders for the individual products or batches, but allows schedulers to plan and reschedule the whole campaign as if it were a single order. By incorporating this logic into the standard planning logic, the system can be trained to create planned orders for campaigns instead of firm planned orders. The computer can manage these when they are beyond the time fences. If this is unacceptable for some reason, or if the time fences are very long, at least the whole campaign, instead of individual orders, can be managed by the planners.

Keeping the Promise of Available to Promise

Jack Gips
President, Jack Gips Inc.

Master scheduling is the meeting point for sales forecasts, order entry, and manufacturing planning in most manufacturing companies. One feature contained in the master scheduling modules of many MRP II and ERP software packages is a calculation that is descriptively named "available to promise". Its purpose is

• to provide information for promising valid delivery dates to customers,
• to reserve or allocate products for customer orders in advance of delivery,
• to provide warnings that supplies are getting low so remaining products can be allocated to satisfy the greatest number of customers or the most important ones,
• to prevent unusually large demands from reducing inventories unexpectedly to levels that cause stockouts for customers who have placed steady demands and/or provided good projections of their needs,
• to provide sales organizations with information about what is available based on previously planned product schedules so they can sell what has been planned rather than take orders at random levels,
• to differentiate quantities of products that have been produced to satisfy the forecasts of different markets, customers, or sales organizations and assure that each gets its fair share.

A-T-P must be applied differently to products that are make-to-stock, assemble-to-order, and make-to-order. In all cases it compares current and future supplies against existing customer orders to determine what supplies are not yet allocated. When a new order is entered or reviewed, the customer's request is matched with the unallocated supplies to determine when it can be satisfied.

On make-to-stock products, A-T-P can be used to determine the following:

• if products are available at the time of order entry,
• if all line items can be shipped immediately or at the customer's requested date,
• when every item on an order will be available for a single shipment, or
• when partial shipments can be made.

On assemble-to-order items, A-T-P can be used to determine when a configured product can be promised based on the availability of the options and features that have to be assembled.

Deliveries of make-to-order products can be promised based on A-T-P for critical raw materials, components, or capacities of critical work centers.

Obstacles to the use of A-T-P

In truth, many of the companies whose systems are capable of calculating available to promise information have failed to use it in actual practice. For some, available to promise does not fit the dissociation that exists between sales and manufacturing. Sales may forecast to give manufacturing a driver, but has no obligation to sell into that plan. Customer orders are often accepted and promised with "standard delivery lead-times" or simply entered for immediate delivery without any comparison to the plans or available to promise. Sales people may in fact be rewarded for overselling the forecasts or for selling the dollar volume even though the mix bears no resemblance to the forecast. If this approach continues to exist once a system is installed, available to promise is usually switched off or ignored.

Some companies would like to use their available to promise information, but it is not accurate enough to use or not presented in the right form to use it. If the supplies (on hand inventories or supply orders) cannot be trusted or the demand data contains errors or false demands, available to promise information calculated from these will also be untrustworthy. If many customer promises based on these data are not met, available to promise will not be used for very long.

If available to promise is not presented to the people who enter the orders in an easily useable form, the same will be true. Sales and customer service people usually do not want to see available to promise as one line on a multi-line master scheduling worksheet. They want it to appear on their order entry and quotation screens. If the orders call for multiple line items, they usually want to enter the whole order and review available to promise for the entire order rather than one line at a time.

A-T-P Calculation

Calculating A-T-P is relatively simple. The formula is as follows:

• AVAILABLE TO PROMISE = Total Supply - Total Actual Demand
• where, Total Supply = Quantity On Hand + Quantities On Order
• and, Total Actual Demand = Quantity on Customer Orders

The Quantity On Hand includes any safety stock since safety stock is not allocated to any particular customers and is therefore available to promise. The Total Actual Demand excludes sales forecasts since they are a prediction of future demand and the inventories to cover them are not allocated to any customers. Figure 1 shows the calculation of A T P for a make-to-stock item.

FIGURE 1

There are some problems with the A T P calculation in some of the newer ERP software packages. Several of these packages deduct safety stocks from the on hand quantities before calculating A T P. Their argument is that safety stocks are for emergency use only and not normally available for sale. Since most of these safety stocks have been calculated based on the variability of demand against the forecasts, they are planned to cover those expected variations and not just for emergencies. The users of those packages can often be found either manually adding back the safety stocks to the A T P or ignoring A T P altogether.

Another common problem is the exclusion of system planned orders from the calculation. These systems only consider released and firm planned orders. The argument in this case is that system planned orders change frequently and are therefore not reliable enough to base promises on. The result for companies that take orders now for delivery at customer requested dates in the future is that the calculation runs out of supply before it satisfies all the demand. Manual calculations show that the order can be covered by the planned orders that are being ignored. In fact, the planned orders are the plan. The system will always cover demand if there is demand to be satisfied. They should become firmed as they approach the item's time fence and released at the lead-time. This problem becomes much more serious if the software does not provide for time fences and firming messages when system planned orders cross them. If the system is allowed to plan orders inside the time fence and in the first few weeks of the horizon, available to promise will show very little if any supply to satisfy the customers if these orders are excluded. It will show A T P that is impossible to meet if it includes planned orders when the system is allowed to plan them inside the time fences.

There have also been arguments that supply order deliveries often cannot be trusted because the components to produce them are frequently not available. Or, if a supply order is to be pulled up to an earlier date, it is important to be able to drill down to see the availability of the components before moving it. There has been much ado about the need for "drill down" capabilities in the software. In reality, however, only a small percentage of companies are in a position to use this kind of information. As an example, take a company whose product has 100 components and four levels in its bill of materials. A customer requests a quantity greater than is currently planned. We drill down to look at all these components and find twenty of them would be short if the next order is pulled up to satisfy this request. This is to be expected since the order is greater than the original plan. What do we do now? Call ten suppliers and check on availability? Analyze the work centers for capacity and the raw materials for the manufactured items to see if it can be done? Do we do all this before promising delivery to the customer? And what if this scenario happens many times every day? Is all this detail really better information than we can receive from A T P? Do we have time to analyze it?

It may be practical to drill down to the details in companies whose products have small bills of material and whose processes have a few short paths. We learned a long time ago that it was not practical in many companies to manage our plants at the detailed component levels. It was one of the reasons we emphasized the need to manage our master schedules and drive the decisions made there to the component levels. Having the technology to manipulate the details does not necessarily make it the right thing to do.

A-T-P Scenarios

When a customer order is ready to be entered, there are a number of situations and decisions that A T P can prompt. The actions taken and decisions made may vary by company or product. One factor may be the degree of importance of the item to the
customer (such as a life saving drug). The fierceness of the competition or the importance of controlling costs may also come into play. Available to promise
does not make products available. It simply tells you if they are planned to be available when a customer requests them. If not, there are usually two options . . . change the plan and make them available or promise delivery at a different time when they will be. Anything else is wishful thinking.

Scenario #1 - The customer asks for 100 units of an item at a particular date and there are 100 or more available.

Action - Promise the delivery as requested.

Scenario #2 - The customer asks for 100 units of an item at a particular date and only 50 units are available to promise.

Possible Actions - Ask the customer if a partial delivery at the requested date and a second delivery when A-T-P shows more available will be acceptable. If this is not acceptable, contact the master scheduler to see if there is a way to adjust the schedules to satisfy the customer.

Scenario #3 - The customer asks for 100 units of an item at a particular date and the system shows that there are exactly 100 available to promise. However this is a new customer and the 100 have been planned based on forecasts from important existing customers. After checking with the master scheduler, it is determined that if they are given to the new customer, the replenishment cannot be available for several weeks.

Possible Actions

• Give them to the new customer and take the heat from the rest.
• Call the existing customers and check on their needs. Negotiate with both to split the quantity until more can be made available.
• Buy the product from another customer who has it or from a competitor to satisfy both customers.
• Save the product for the customer who gave you the forecast and turn down the new business.

Scenario #4 - The US marketing organization has worked diligently to provide a reasonably accurate forecast for the domestic demand. The International marketers provide a very inaccurate forecast and sometimes forget to forecast at all. They assume that the plant can make whatever they need. So the manufacturing planners forecast International demand based on history. There are 100 units available to promise. They were produced to satisfy a US forecast of 80 units and an International forecast of 20. An International order comes in for 75 units. A replenishment order cannot be available for several weeks.

Possible Actions

• Give them to the International customer and hope US marketing understands the situation.
• Call the US marketing and ask them if they are willing to share. Negotiate with both to split the quantity until more can be made available.
• Have International marketing call the US marketing and ask them if they are willing to share. Let them negotiate and keep manufacturing out of the middle.
• Save the product for the US customer who gave you the good forecast and tell International that it is important to provide a good forecast if they want product availability at short notice.

Scenario #5 - There are 100 units in inventory in the finished goods warehouse. A customer requests 65 but the system shows only 50 are available to promise. This is because another customer previously ordered 50 but requested delivery three weeks later. The next supply order is not due until week 5.

Possible Actions

• Ship 65 on the new customer order and assume that manufacturing will be able to respond.
• Call the master scheduler before shipping and assure that manufacturing can respond. Then ship the 65 on the new order.
• Try to make a partial shipment on the new customer order and deliver the remainder in week 5.

The changing business model

The scenarios above are based on the traditional customer service model with a customer on the phone talking to a customer service representative (CSR) during normal business hours. However this model is changing quickly, especially in a make to stock environment. Industry consolidation and information technology is mutating customer service processes at the speed of light. Three trends will force us to reconsider how our ATP 'toolkit' should be deployed to engage and inform customers in the future.

The first trend is customer segmentation. No longer does one set of service rules apply to every customer. The level of responsiveness required of your business may vary based on the industry your customer serves, or the importance of that customer to your business strategy. For example, if your customer serves one of the mass merchandisers like Home Depot or Wal-Mart rest assured that you will be required to adjust to variations in their demand with very little lead-time. If you want the business, the mass merchandisers and the mass merchandisers' customers set the rules of the game. Your only decision is whether to play or not. However, you will not likely plan on offering Wal-Mart type lead-times to all of your other customers. Customer service is becoming an increasingly strategic discipline, counterbalancing the other critical procurement variables, price and quality. For each customer segment the business leaders, not operations, must define the level of service and service features that will give your business competitive advantage. Over-serving with no clear goal is a waste of shareholder value. So is under-serving if there is market advantage to leverage. Differentiating customer service strategy by customer or customer group will impact the way you manage many factors: standard lead-times, inventory, capacity and logistics. Available-to-promise is a means of managing these differentiated strategies. For example, you could create a group of mega customers with their own dedicated available-to-promise to assure that they get preferred access to planned supplies over non-members of that group. Within that group, however, the mega customers who place their demands first will get priority access to that supply over their fellow mega customers. Likely business rules will cause whoever provides the best forecast to earn the right to a preferred level of service over their short lead-time peers.

The second trend is asynchronous communication enabled by advances in information technology. Much service activity that used to be conducted by telephone and fax is migrating to other communication channels like true EDI and Internet. The term 'true EDI' implies real data interchange between supply partners' information systems not 'mock EDI' that needs to be re-keyed. Customer limited access to their order information through the Internet, often referred to as an Extranet, is evolving to include computer to computer communication that is mutating and merging with our traditional definition of EDI. The benefits to both the customer and the supplier are enormous. Customers can access up-to-date information any time of the day or night, a particularly powerful advantage as the sun never sets on many global industry leaders. Suppliers benefit as well, as the costs of serving customers drop dramatically as proportionately fewer staff are required to process orders and distribute information.

Again, there are profound implications for available-to-promise. While the traditional scenario suggests that customer service wants ATP info to appear on order entry and quotation screens, or after multiple order lines have been entered, what are the ramifications if the customer is now in the driver's seat? All of the ATP user issues mentioned earlier are intensified when you visualize your customer accessing the information directly without the filter of a customer service representative deciding whether they can trust, and how they should present, the information.

• Is the inventory accurate?
• Will the supplies really appear when planned?
• Do planned orders provide a sufficient picture of supply into the future?
• Does the demand data contain false demands like safety stocks?
• What are the business rules for dealing with unusual spikes in demand?
• How can you assure that ATP adapts to differentiated service levels among customers?
• How should the information be presented to the customer in a user friendly way?

An additional issue arises as well. What happens if the customers' requested lead-times cannot be met? Likely you web site will need to allow the customers to indicate their preference by noting whether they will accept the alternative promised dates, whether they want to trigger a human intervention to expedite the delayed item, or whether they want to cancel the request for that item.

The third trend, vendor managed inventories (VMI), takes the customer out of the picture altogether for routine supply chain management. In a VMI environment the supplier has access to the customer's inventory and usage data and is responsible for maintaining the inventory level required by the customer. Now ATP technology comes full circle, as the supplier's customer service representative or planner becomes a surrogate buyer. Technology and business rules developed to support trend two, true EDI and Internet commerce, will greatly enhance the efficiency of handling the customer's purchasing activity in house.

Conclusion

Available to promise has proven to be a very effective tool to coordinate the efforts of sales and manufacturing in both make-to-stock and make-to-order environments when it is used correctly. As we look into the future, we see that it will become even more important because it is exactly the type of information required to streamline the supplier-customer interface our high speed world will demand. It is usually discarded quickly if the right conditions are not in place when it is started up. This means that we must get it right the first time. Our inventories must be accurate and our supply orders must be delivered on schedule. We must clearly define the target levels for customer service and the rules for prioritizing our customers and for handling the scenarios. We have to determine the roles of the customers and the suppliers in utilizing the systems and A-T-P if we are to eliminate middlemen and bureaucratic delays. Once this is accomplished, A T P promises to link manufacturing and sales to a single plan, reduce unnecessary schedule changes, improve the quality of promise dates to the customers and on-time delivery to those promises, and simplify our customer relationships. It will help create an environment in which our customers will not be inclined to look elsewhere when they need our products.