What is the point of forecasting? It is a fair question, given that forecasts are rarely 100% correct. Another, equally pertinent question is what makes a good forecast, from a business perspective? In an interesting piece in the Financial Times, the excellent Tim Harford points out that the basic problem with forecasts is that you don’t know which ones are right and which ones are wrong except with hindsight. The internet is full of people making predictions - on the stockmarket, sporting outcomes, election results etc. - some of which will inevitably turn out to have been correct. However, they are of no use unless you know in advance which ones to believe. So, the question remains: what is the point of forecasting for a business? And, knowing the answer to that question, what then makes for a good forecast?
Businesses have to forecast because they have to make decisions about the future. It is only by knowing what they are likely to sell that they can take steps to source/make those items and support their customers. And it is only by knowing what their cash flows are likely to look like that they can put in place the necessary funding arrangements to support them. For these purposes, a forecast does not have to be perfect, it just has to be in the right ballpark - at least, more so than the alternative approach of crossing your fingers and hoping for the best.
From this perspective, the forecasting needs of different organisations will be very different: some capital-light businesses, particularly those in the services arena, may well be able to react very quickly to whatever opportunities and challenges the market throws at them. They may not need such rigorous forecasts as businesses that need to make significant capital allocation decisions, or place substantial orders for long lead-time products. However, all businesses would benefit from knowing what is around the corner. Forewarned is forearmed, and surely, the point of a forecast is to help an organisation to take decisions which better prepare it for the future.
In a business context, forecasts are not typically “right” or “wrong”, but they can be more or less accurate. What matters is not necessarily exactly how accurate a forecast is, but how useful. And, as outlined above, a useful forecast is one that helps an organisation to prepare for the future. Indeed, even with the most accurate forecasting model in the world, an organisation would be unwise to ignore other possibilities. In an uncertain world, it always makes sense for management to hedge their bets and prepare for a range of outcomes, if only to enable them to sleep soundly at night.
So, what makes for a useful forecast? To be useful, a forecast should be:
As discussed, a forecast does not have to be perfectly accurate to be useful. However, it needs to be sufficiently accurate to support the purpose for which it was created and should at least be directionally correct. It is also useful to have a good idea of how accurate it is likely to be - or to create a set of forecasts under different scenarios. That will help a company to prepare for the best and worst cases, as well as the most likely outcome.
A prediction is of no use without a degree of specificity. Predicting that a coin toss will come up ‘heads’, without specifying which particular toss of the coin that prediction applies to helps no one. In his article, Tim Harford cites the example of the expert who predicted a global pandemic at a conference in 2019. Devastatingly accurate with hindsight, and of huge global significance, it still failed to be of much practical use as it was not specific about the nature of the pandemic, or likely source or timing. After all, people had been predicting global pandemics for years (and still are).
Closely allied to the specificity of a forecast is the level of detail it provides. If specificity is about knowing exactly what a given forecast applies to, or when it will take effect, detail is about knowing the different elements that make up that forecast at a lower level. For a business, this is vital, as they will typically need a fair amount of detail in order to plan effectively. Breaking down a business forecast into different elements at a granular level helps individual departments to make meaningful plans, and allows parts to be updated independently of each other. This is a prerequisite for the last of our requirements: that a forecast should be up to date - meaning that it needs to be dynamic or easily updatable.
Forecasts can be created over the short, medium or long term and, understandably, we expect recent short-term forecasts to be more accurate (and specific and detailed) than those made over longer time periods. For a business, the forecasts they make before the start of a financial year will need updating as the year progresses, in the light of new information about the marketplace and the company’s performance. This is hard to do for many companies, who throw everything into a big budgeting exercise before the year starts, but cannot follow that up with regular reforecasts to a similar standard.
Key to making a forecast easy to update is creating and recording the detail of what it comprises, as discussed above. If the original assumptions and calculations have been stored, then it is a straightforward task to input revised estimates based on more recent data, and recalculate the outputs.
Forecasts are an essential, if frequently burdensome, element of the business cycle. As Tim Harford says in his article, some forecasts are right and some forecasts are wrong, and by the time you know which is which, it is too late. In a business context, this perhaps needs refining: business forecasts are not typically binary predictions that are either “wrong” or “right”. However, they can be more or less accurate, and from a directional perspective they can be certainly be correct or incorrect (predicting a growing market when it is actually declining, for example).
What matters for a forecast is whether it is useful, in the sense that it helps a company to plan for the future. For this it must be reasonably accurate, specific, detailed and up to date. This, in turn, demands it be based on sound mathematical models, grounded in good, clean data. Automatic feeds of live data are useful to keep it current, and to assist with reforecasting and, of course, a good computer system is vital to bring these elements together with the requisite level of detail and to provide the ability to interrogate it.
Compas has been developed over many years to meet exactly this challenge. Our consultancy-based approach works with clients to design a forecasting approach that works for them, and to help them identify data sources and construct feeds of high quality data. The simple and intuitive user interface can be customised to create a set of powerful dashboards and reports, and the whole system lives in the cloud to minimise costs and the burden on the IT department.
To understand more about how Compas can help you to build accurate, detailed and dynamic forecasts, contact us now for a free demonstration.