Demand forecasting

I'm Bob Artner from TechRepublic and our topic today is demand forecasting, and by demand forecasting we mean the ability in the near term to forecast demand on your cost, on your sales, on staffing any of the things you need to do to run your organization. And the point I want to make is, it's very important to choose the right statistics and the right time period when you're getting your bases for your forecast, otherwise your numbers will be wrong and your forecast will be worthless.

Here's what I mean. My first job in management was working in restaurants over 20 years ago, and one of the most tedious jobs restaurant managers had to do was forecast sales. We would look at the last four weeks of sales activity, so this would be this year's sales, and say, "Okay, what is this downward trend here, so if I'm forecasting what the next four weeks are going to be, I would say, okay, that looks like the tread line, looks like it's negative, so I must get a forecast more or the same." Just kind of extrapolate that tread line forward. You know what? It often was wrong and I'd, you know, we'd end up ordering too little food, or too much food, or too much staff, or not enough staff because the recent past was not a good indicator of what the immediate future was going to be. And here's why: The restaurant business is cyclical.

What you need to do is look at the last four weeks from that same period because what if you're seeing a different trend altogether. If you look at last year's numbers, and this year's numbers, and you look at the difference, that gives you the opportunity to do a better forecast here. Let's say that even though this is declining we're still averaging 9% more this year over last, what's called "year over year." Then if we want to forecast this future thing here we would say, okay what did we do these same four weeks last year. Well, you know, maybe schools out, maybe there's vacations going on or whatever, but you know last year these four weeks were actually pretty good. Instead of trending a decline if we're 9% over, and we're trying to look at what the future is going to be, it's more than likely that the trend for the near-term is going to be in a completely the opposite direction than the trend from the recent four weeks, and if we get that right then you'd be able to have enough staff on hand, you'll be able to control your costs by having enough food on hand, and not have to go out and order more, or turn away sales because you don't have the food to support it. That's what I mean by saying it's important to get the right time period and the right metrics.

Now one caveat to this real quick: What if I work in a business that's not cyclical? What if I work in a, let's say, a Web site and I'm forecasting traffic, and I've only been around for 18 months and my traffic pattern looks like this, over the 18 month period. Well, it's not going to be very useful for me to say, okay, here's the four weeks from last year you can see that in a period of rapid growth or relatively new product innovation, you might have to go back and look at the recent past as a better indicator.

So, the key takeaway here, know the kind of business you're in, you know what the significant metrics are and use that when you're doing demand forecasting.