WORDS FROM WALLACE
New Year’s Resolutions
Many of you probably have a list of things to do in this coming year that’s as
long as your arm, and don’t need any more ideas from me. For the rest of you,
let me lay out an idea for possibly improving your forecasts and making things
a lot better in your company: eliminating biased forecasts.
Are your forecasts consistently higher than actual sales, or consistently lower?
That means those forecasts are biased, the following being one example:
FORECAST: 100 100 100 100 100 100
100 >
ACTUAL SALES: 80 90 70 70 90 70
CUMULATIVE
FORECAST ERROR: +20 +30 +60 +90 +100 +130
How would you like to be dealing with a forecast like this for products coming
in on a slow boat from China? You would be consistently planning on the high side
and sending those high requirements to the contract manufacturer(s) over there.
In this situation, the finished goods inventory will build. And, as the finished
goods inventory builds, you might cut back production for a while. Sooner or later,
the inventory may drop down and then the tendency will be to again plan to that
high, biased forecast. Hence more inventory build-ups, and more signals for the
manufacturer to slow down, speed up, slow down, and on and on. The same result
can occur, of course, in your domestic plants.
Biased forecasts are the pits; they represent absolutely the worst kind of forecast
error. But why do they happen? Well, my colleague and co-author Bob Stahl has
a strong opinion on that: he says “Biased forecasts are forecasts that are wrong
on purpose.”
When I first heard that, I felt it was a bit harsh but then Bob explained it
to me: “There is always an underlying cause of biased forecasts and it’s almost
always caused by people being led to do the wrong thing.” One example: does the
sales person get pats on the back, and perhaps bonus money, for beating the forecast?
Then he or she is being incented to forecast low. Another example: is the forecast
consistently higher than actual sales, as in the example above? This frequently
results from in imposition of stretch goals that simply aren’t attainable.
Think back to your TQM training. The technique called the 5 Whys says to ask why five times, each time peeling back another layer of the onion and thus getting
to root cause. The 5 Whys is an excellent tool for determining the true root causes of biased forecasts.
If you think you may have a bias problem in your forecasts, you might consider
the following steps:
1. Calculate the cumulative forecast error by comparing actual sales against
the last forecast used for each given period. Use the logic shown above, perhaps
with data from 2005. (In the jargon of our trade, this cumulative error is sometimes
called the running sum of forecast error [RFSE] or the sum of deviations [SOD]).
2. Calculate, as best you can, the dollar impact of the cost of these biased
forecasts: excess inventories, stockouts, overtime, costs of hiring and layoffs
resulting from unnecessary production rate changes, and so forth. This does not
need to be accurate to four decimal places; it is directional, not precise.
3. Do a root cause analysis on these biased forecasts. Utilize the 5 Whys or other techniques that provide focus and rigor.
4. Enlist the aid of a champion, someone on the executive staff who understands
the problem and can see the need to change (sometimes the senior finance executive:
CFO, VP of Finance).
5. Make the case for change to top management, involving the champion directly
in the presentation. When obtained, proceed to fix the problems.
Will it be difficult convincing top management? It may be. Why, if it’s such
a good deal? Answer: because it can involve some very fundamental changes in how
people are evaluated (both formally and informally), plus how they’re compensated
and otherwise rewarded. Or the needed changes might involve how the top management
team itself operates. Much will depend on how well your organization accepts change.
On the other hand, it may be easy. This is an internal improvement project; you
won’t need outside consultants, software, or huge teams of people needing to get
trained. It’s a very low cost initiative. And for many companies, the payoff can
be huge.
Is it worth going for? Seems to me it is. Seems to me it’s a lot better than
just sitting around and griping about “those lousy forecasts” – or second guessing
them – both of which are what I did 30+ years ago when I had a job in industry.
Thanks for listening,
Tom
_______________________________________________________________________
Tips from Tom: Speaking of lousy forecasts, I’d like to raise the issue of managing abnormal demand. This is demand that typically has not been forecasted, is quite large, and
is a surprise. Frequently these are caused by a competitor going on strike, or
perhaps having a plant get flooded.
When these kinds of orders are routinely accepted, the resulting problems – scrambling,
missed shipments to existing customers, missed shipments to the new customer –
are often blamed on lousy forecasts. And that’s wrong. The problem wasn’t with the forecast at all; it was caused by the lack of good
processes to manage abnormal demand. We’ll look more closely at this issue in a future column. If you want to get
started on this problem sooner, drop me an e-mail through TFWallace.com or if you have our book: Sales Forecasting: A New Approach, check pages 90-94.
|
