Tuesday, May 20, 2008

Can we forecast marketing performance?

One issue which I've been looking at recently is that of forecast accuracy in marketing. The question I've been asking myself is this - how accurate can a forecast be and if we're not sure of the forecasts accuracy, what should we do - ignore it?

Tim Harford (the Undercover Economist) recently wrote an interesting piece on Economic Forecasting and the long running debate about why economists bother when their forecasting is so often inaccurate (see http://timharford.com/2008/05/why-economic-forecasts-are-so-hard-to-get-right/). The conclusion which is reached by Prof David Hendry is that we fail to forecast accurately because in part we fail to quickly realise when a "structural break" has taken place in the environment being investigated.

This issue is one that applies tenfold in the field of marketing because there are so many potential things which can lead to a structural change in the market. For example, a competitor launches a revolutionary new product; your advertising is run in a recently vandalised environment; a factory burns down. None of these things in inherently predictable but as a brand manager each can be catastrophic to the performance of your portfolio. Does this mean that it's not worth forecasting because the likelihood that the forecast will be "accurate" is low?

Predictably I'm going to say "no". There are many reasons that forecasting is a useful discipline in itself. In the marketing area, these would be something like:

  1. It forces you to consider a wide range of outcomes from each strategy
  2. It's a discipline which helps ensure consistency across the business and gives credibility to your operations
  3. Getting all your forecasting ideas and assumptions down on paper will help you evaluate and learn from past mistakes. Finding out that an idea was a success is great - being able to repeat the logical process that led to the idea is fantastic and will take your marketing to a new level

Will the forecasts be accurate? In the marketing world, the answer is they might be in the short term but in the long terms, it's anybodies guess. Will advanced analytics solve this? Again better analytics may give a small improvement in accuracy over the long term but it has yet to be proved. In the short term, practical quick analysis is more powerful than advanced time consuming analysis.

Forecasting is necessary in all businesses (although many still fail to add any discipline to this process). How accurate these forecasts are does indeed depend on the effort put into them and the predictability of the market in question. Whether the incremental effort that goes into advanced analytics is worth it depends on whether you can find the correct balance between speed of analysis and the degree of "forecastability" in the market.

No comments: