DEALING WITH IT
So now you've learned that most of us must forecast, history alone will not get you a good one, errors can be cut by using intelligent input, and when variances occur, they should be used to uncover potential business problems and opportunities and deal with them quickly. It REALLY doesn't come down to just using the right model, does it? But, how do you spend less time, get better results, minimize error, make tough calls on dubious input, handle new customers and products, and know whether or not you should freeze the forecast? Continue reading, and I'll tell you the things that have worked for me.
Freezing the Forecast
Let's take the easy one first. I have never been able to think of a single good reason to freeze the forecast, although it is done quite frequently. The most common reasons I hear relate to significant changes within the total lead time and MPS fluctuations that go up one period and down the next.
For starters, a forecast if not an execution plan—the MPS is! One of the reasons you have an MPS is stability for execution (purchase, planning, and manufacturing). If you want to freeze something, freeze the MPS—that's why you have one! All you do by freezing the forecast is to eliminate the possibility that there could be any proactive response to the change within the lead time. You may be able to expedite more components in, or maybe with lot sizing you already have enough. You may be able to add production capacity through overtime, subcontracting, or simply diverting from less needed product. But if you don't pass the new forecast to the MPS, you will never know if a proactive execution response is even possible—you've just blindfolded the entire organization. Even with the worst-case scenario, putting the new, higher or
lower forecast into a frozen MPS will enable the lump or drop to be seen, so the additional components can be ordered now, or unneeded components rescheduled out. When you freeze the forecast, you have no opportunity to prevent the problem.
Notice that I said prevent the problem. Whether or not you freeze the forecast, if you do not change the circumstances that are occurring that caused the need for the forecast change, that event will still happen and you will then certainly have a problem. For instance, if you are getting a new account, and you freeze the forecast because you cannot build the additional product in less than full lead time, you will still get the orders from the new customer! Only now, by the time you receive them, they will be a surprise to the entire supply chain, and will result in back orders that you now must scramble to fill, with the additional side benefit of making the new customer wonder why they switched vendors in the first place. The same holds true with individual item forecasts that may go up one month and down the next—if
the circumstances that caused the need for the change do not change and you freeze the forecast, you will still get more orders one month and fewer the next. You haven't prevented a thing except the ability of the organization to react before it is too late.
About Those New Customers
If you have been in the forecasting field for a while, you know that getting a forecast from a new customer is not always the manna from heaven that it is purported to be. I'll share some techniques I've learned that clear up the decision process. There are specific steps you can follow after you answer three questions. First, do they have experience with the line and this is simply a vendor change for them, or is it an entirely new line? Second, are there any differences between your company and the new customer in marketing approach that would cause differences in product mix or seasonality? Lastly, is it a forecast of their intended purchases or sales to their own customers?
Now for a simple test for any forecast you get from a customer. Take all their item forecasts in a product line or two, a few dozen items, or however many items resources allow, and add them up on a spreadsheet. Then use the spreadsheet to calculate the percentage of contribution to the mix for each item. Select the same items from your records, and do the same thing. If the percentage contribution to the mix for the customer forecast and your own experience are close, you can use it pretty confidently. But what if they're different?
If they have experience with the line and they have market differences, determine whether those differences explain the situation. For instance, if I were an automotive replacement parts manufacturer that served all typed of vehicles, and my new customer was one that serviced just heavy trucks, that would explain the difference. If I was a home air conditioning manufacturer serving the entire nation and the new customer was a chain that served only Arizona, it would explain higher volumes on heavy-duty units. If, after discussing the outcome with the sales and marketing folks, the differences can be explained, use the forecast. If they can't be explained, I would use it for the initial order only, because that's what they think they need, and then throw it away. Believe it or not, there ARE companies with forecasts worse than yours.
If it is a new line for a new customer, I strongly recommend that you give them a forecast, rather than the other way around. Your company has experience, and who better to determine a valid stocking plan? If marketing plans and circumstances are the same, the same percentage mix should be used to present to the new account. Likewise, if there are major differences, work with sales and marketing to take those into account. This approach will help your new customer sell more, since the correct merchandise will be stocked, and prevent waste for both of you, since you will not be rushing to build merchandise that will only be returned, older and dustier, years later.