Why doesn't forecasting excite management (when it should)?
It is often said that ‘no-one is happy with their forecasts’. The statement certainly chimes with my experience of dealing with senior supply chain executives, but I can’t help feeling that it is a strange kind of unhappiness.
It is not the kind of dissatisfaction or frustration that fires people up, giving them the energy and determination to improve things – fast. Instead it is more low energy; disappointment may be a better way of describing it. Almost fatalistic.
The lethargy with which we are faced certainly frustrates those of us who believe we have answers to many of the problems that forecasters face and that improvement is not only possible but also very worthwhile from a business perspective. Perhaps if we understood the cause we could do something to awake the passion needed to tackle what I believe to be the biggest source of unrecognised waste in the supply chain.
My sense is that the causes of the problem are threefold:
Unrealistic Expectations
One source of the problem could be unrealistic expectations.
A few years ago I was presenting to the demand management process lead for a major multinational – an articulate, clever and experienced person – and casually made what I thought to be an uncontentious statement: ‘of course it is not possible to predict the future exactly’, only to be completely floored by the response to my assertion: ‘why not?’ As usual in such circumstances I had not prepared a reply to a question I had never contemplated being asked and so floundered around for a few seconds before coming up with a completely unconvincing reply.
I hope that few readers of this post will share my client’s belief in a completely deterministic universe, but in my experience many people do have completely unrealistic expectation about our ability to predict the future and consequently the level of forecast errors that are achievable. As a result I believe people often make major investments in forecasting software or changes in process only to become rapidly disillusioned when the tried and trusted recipe of IT and accountability fails to make the ‘problem’ (i.e. forecast error) disappear.
For me, what they fail to recognise is that every data series contains noise, and that noise by its very nature is unforecastable; so we should not expect errors to completely disappear. If we were able to see beyond the noise, and stop doing things in the name of ‘improvement’ that made matters worse they would see that the prize is big and eminently achievable.
Not knowing what ‘good’ looks like – and what failing to achieve it costs the business
A consequence of not being able to differentiate between the signal and noise, and consequently the forecastability of a data series, means that very few supply chain professionals have any idea of what constitutes a good level of forecast performance. For sure, traditional metrics like MAPE and Forecast Accuracy and any targeting mechanism that refers to them, are a totally useless guide to performance.
A corollary of not knowing what ‘good’ looks like, is that most supply chain professionals cannot begin to know how far short of good they are and what this failure is costing the business. This is compounded by the fact that few people (including those working in finance) understand how poor forecasting manifests itself in the Profit and Loss account and Balance Sheet and even fewer know how to go about calculating it, even in principle.
If they did what they would discover is that avoidable forecast error – that is the waste associated with the forecast process – typically costs the equivalent of 1% to 3% of the cost of production, which I believe would make it the biggest source of inefficiency in most supply chains. Furthermore somewhere between 30% and 50% of all SKU level forecast destroy value when compared to a simple consumption based replenishment strategy. This means that the ROI for many forecasting improvement projects is zero or worse. All this is shocking but illustrative of the prize if improvement efforts are better directed and successful.
A belief that improvement is difficult, painful and not cost effective.
Most people I speak to in business believe that their level of forecast error is way too high and that this reflects badly on the competence of the people running the process. Also my experience is that every customer failure is automatically assumed to be the fault of demand management rather than unreliable supply, shoddy inventory management practice, forecast overrides coming from sales or S&OP or simply the inevitable consequence of customer service targets being less than 100%. Furthermore many businesses have made major investments in process and software most of which have barely moved the needle.
It is hardly any wonder that supply chain professionals are reluctant to lead the charge for more investment since their experience is that if you put your head above the parapet and raise hopes of victory you will fail and ultimately get shot by your own side. Few people have any experience of more than a few yards of territory gained as the result of much hard work and suffering; in the language of business there are no ‘low hanging fruit’.
While I understand and sympathise with the predicament I believe that the notion that improvement is not possible is a product of a failure to understand the nature of the problem and the tendency to look for software ‘silver bullets’ and process ‘best practice’ neither of which is capable of delivering value on its own. The very fact that 30%-50% of all SKU level forecasts can be easily improved by reverting to a simple naïve forecasts – which almost certainly will not be the best method – demonstrates that it is possible to make improvements quickly and easily providing you are able to target your actions by measuring performance in the correct way.
So why should management be excited about forecasting (when they are not)?
- 1.Many businesses have forecast processes that perform poorly and nearly all have forecasts that destroy value 30%-50% of the time. The return on investment made in software and processes has been misdirected and so falls well short of what was promised and what is possible.
- 2.Huge improvements are possible once forecast performance is measured in an appropriate way. Doing so will allow improvement efforts to be better directed. In many businesses avoidable costs could be halved over time, customer service improved and the ROI of investment in forecasting multiplied many times over. All this will help forecasting, and the people that work in the process, to gain the respect and trust of the rest of the business.
- 3.Much improvement can be made quickly and painlessly. For example, it is easy to improvement the performance of those forecast that are destroying value by doing less.
- 4.Proper measurement will also enable demand manager to demonstrate the value that they add and where external intervention or unnecessary process steps make matters worse, thereby removing one of the barriers to improvement.
Post a comment