Sales forecasting is an activity that consumes an enormous amount of management’s time, yet few organizations are happy with the accuracy of their collective forecasting efforts. So how is it that sales forces can expend so much effort on forecasting and still produce revenue predictions that stink? To find out, Vantage Point recently turned its research engine toward sales forecasting in an effort to identify the sources of forecasting failure as well as the best practices that can make sales forecasting more accurate.
Our first observation is that many sales forces are using forecasting models that don’t reflect the way their sales forces actually sell. The predominant forecasting approach is, of course, to build the sales forecast from a pipeline of opportunities that are being actively pursued by the sales force. Each opportunity is slotted into a stage of the company’s sales process, and a percentage is then applied to the deals in each stage to generate a probability-adjusted revenue prediction. In fact, our research shows that 85% of business-to-business sales forces currently forecast using this tried-and-true method. So what’s the problem with that?
We will set aside for the moment the many problems with how this opportunity forecasting method is being executed, and we’ll instead focus on a more fundamental issue that cripples many sales forecasts. The issue is the fact that you should probably be using a forecasting method OTHER than the classic ‘opportunity’ forecasting approach – one that is based on something other than a pipeline of deals with probability-weighted stages.
Our research revealed that there are at least 3 alternative methods of sales forecasting that return better results in certain situations. For instance, sales forecasting should be done at the Account level, if you have a business model where a high number of deals flow from a handful of existing customers. Or perhaps you should be forecasting at the Territory level, if your salespeople cover geographies with dozens or even hundreds of accounts. And perhaps you should be forecasting at the Call level, if your sales team makes a high volume of outbound sales calls on unfamiliar prospects. In fact, in our survey of 62 global sales forces, 74% said that they SHOULD be basing their forecasts on something other than opportunities. But they aren’t – only 36% claim that they ARE using a different forecasting method. Yeah… That could be a problem.
If your forecasts are based on a model that isn’t relevant for your sales force, then it’s easy to see why your forecasts stink. You’re developing your forecasts using data and assumptions that don’t reflect what’s actually happening in the field. It would be like trying to read this article while moving your eyes from right to left rather than left to right… You’d be taking in a jumble of information that your mind can’t process, because you’re following the wrong process. Similarly, we’ve seen a remarkable number of companies try to shoehorn their forecasts into an opportunity-based model with stages and percentages, when it just didn’t make any sense to do so. And not surprisingly, their forecasts didn’t make any sense either.
So if you’re using a traditional opportunity-based forecasting model and suffer from erratic forecasts, you might consider whether you should be using a different method to develop your forecasts – one that more closely reflects the reality of your sales force. Otherwise, you will continue to invest large chunks of time in an activity that will never yield better results. And that stinks worse than the bad forecasts.
Jason – you’re right that preparing a sales forecast based on the prospect’s perceived position in a sales process is a flawed approach. So much so, that I recommend companies not bother trying. It’s only slightly better than throwing fairy dust in the air and making a revenue prediction based on which direction it flies.
Still, many companies cling to this model because they resist giving up a deterministic approach. For example, “When leads enter the ‘demo phase’ of the sales cycle, we count that as 60% likely to buy.” That’s it – off they go calculating the revenue outcome, instead of examining what contributes to the uncertainty, and performing more robust calculations based on those conditions. What a non-deterministic analysis can reveal are dependencies for the stated uncertainties (for example, there is a connection between the buying risks customers perceive and the amount of time the opportunity remains open in the sales cycle). Identifying these risks and figuring out how to reduce the volatility in the worst-case/likely/best-case range will help to ensure better quality forecasts. Most important, in the B2B world of sales forecasting, it’s impossible to predict future events based solely past performance or past events. That reality points to the need to perform probability modeling.
As far as your recommendations for how to aggregate forecasts – we recommend assessing forecasts on the most detailed level possible, as forecast errors have a better chance of canceling one another out (high versus low). This is the same for project estimation. Better quality estimates consider the time (or cost) to complete each step, and then to aggregate these into a total, rather than to offer a single estimate for the project.