I’d like to claim that I am not easily annoyed….but that would be untrue. And one of the things that are guaranteed to irritate me is glib statements from people who claim to be experts – particularly if it is part of an act to get people to buy something.
In the realm of forecasting, one of the statements that I often hear is that rolling forecasts are ‘best practice’ – which is another thing that irritates me because it usually is just code for ‘everyone else is doing it’. Just like getting a tattoo is ‘best practice’. Yeah, right.
When I challenge people to explain what they mean when they use the term ‘rolling forecasts’, they usually say something like ‘it involves reforecasting more frequently’, which is wrong. Don’t misunderstand me – forecasting more frequently is very definitely ‘A Good Thing’, but that is not what a rolling forecast is, and those proffering this definition have demonstrated that they don’t really understand why they forecast in the first place.
Put simply, a rolling forecast is a forecast that has a constant horizon. So if every time you forecast, you looked 18 months out or just six months out, this would be a rolling forecast. Contrast this with forecasts to the end of the financial year made in January, March, June and September. These are not rolling forecasts because their horizons are not constant, being 12, 9, 6 and 3 months, respectively.
Why is this important?
To answer this question, we need to go back a step.
In the last blog, I described how forecasts help you make sure that your plans are appropriate. That is that they will help take you from where you currently are to where you want to be. Plans are potential future decisions - decisions to commit to action and provide the necessary resources.
What is critical is that this commitment is made at the right time.
If you commit to action too soon, you are potentially vulnerable because you have lost the flexibility to respond to changes in the world. And for obvious reasons, if you commit too late, you are also vulnerable. So it is critically important that the visibility of the future that your forecast horizon is linked to your ability to respond – to the time lag between deciding to act and the results of the action being manifest.
To refer to the sailing analogy I used in the last blog - if you are the captain of an oil tanker and it takes you a mile to change course, then you need to be able to ‘forecast’ a mile ahead, probably using radar. Any shorter forecast horizon is dangerous, any longer is unnecessary. If, on the other hand, you are in a very manoeuvrable speedboat, you don’t need radar at all because you can see far enough ahead unaided.
So you need a rolling forecast horizon to make sure that you are able to make the right decisions at the right time.
This is not a matter of opinion.
The design of your rolling forecast process is an inevitable logical consequence of the decision-making architecture of your business. This is well known to supply chain managers where understanding the ‘cumulative lead time’ for every product is critically important to ensure that you can maintain supply.
On the subject of timing, I said earlier that forecasting frequently was ‘A Good Thing’. But is this always the case? If you currently forecast on a quarterly basis, should you forecast monthly instead? Or weekly, or daily?
Those readers who have to spend hours in meetings in unproductive arguments about things like this will be pleased to know that there is a ‘right answer’ to this as well, rather than it being a simple matter of opinion.
Let’s return again to the idea of forecasts as information in support of decision-making. You need to update your forecast only when:
The piece of (forecast) information is important for the decisions that you need to take
It is likely that the information has changed materially (i.e. in a way that might change your decision) since the last time you produced the forecast.
My guess is that this will mean that you will end up forecasting some things far more frequently than you currently do and other things far less frequently. In fact, I’d wager that the only thing that you are NOT likely to do is to carry on forecasting everything at the same frequency at the same level of detail as you currently do.
So the art of good forecasting requires a firm grasp of time, particularly the rate at which things change in the marketplace compared to how quickly you can respond. Ideally, you would be able to react so quickly that forecasting would not be required at all. Still, this is a pipe dream for most businesses, so the ability to anticipate has to compensate for the inability to respond.
Perhaps I shouldn’t get irritated by the widespread failure to understand perhaps the most basic principle of forecasting. After all, it wasn’t that long ago – at a time when I designed and ran forecast processes for a living - that I was equally as ignorant. And there is another important topic where the depth of ignorance is equally as great.
Can you guess what it is?