According to research published by CSO Insights, less than half of all forecasted sales opportunities actually result in a sales win. Poor forecast accuracy metrics are a nightmare for most companies, as sales forecasts are one of the primary ways to predict revenue. The root of the sales forecasting problem is not lack of data – most sales teams employ some form of analytics to build their forecasts. The problem is bad data.
The Common Method
While the best sales managers are moving towards forecasting on a deal-by-basis, the vast majority of organizations are still relying on the “weighted pipeline” forecasting system. This involves a sales manager taking each deal in their pipeline, and based on its stage in CRM, assign it a probability to close. For an eight-stage system, the sixth stage would likely be marked as 75% likely to close. The manager then multiplies this probability by the value of the deal. The resulting value is what goes into the forecast.
According to this method, a $100,000 deal in the sixth stage of an eight-stage system would account for $75,000 in the forecast. This is an easy way to get quick look at the pipeline as a whole. But when it comes to generating insight that will influence business decisions, this method is dangerously simple and ignores several fundamental elements of enterprise sales. There are three key places this system gets misleading:
The “weighted pipeline” method assigns the probability a deal will close based on its opportunity stage, so the stage frequently becomes the most scrutinized field in CRM. But there are rarely concrete definitions in place for what each opportunity stage means, and rep optimism or pessimism often skews the accuracy of even the most articulate stage definitions. Determining the stage of a deal is an inherently judgemental process, so no amount of emphasis can guarantee that all reps will mark stages exactly the same. It becomes almost impossible to get good forecast accuracy metrics when basing the analysis on subjective data.
Another problem arises by assigning simple percentages to each deal stage. These percentages are supposed to tell the probability a deal will close once it reaches that point in the sales cycle. But every deal takes a unique path to close, and assuming that all deals have the same probability to close at the same stage neglects to account for huge amounts of data that can add much more accuracy to the probability rating, most of which is right at the company’s fingertips. Factors like the prospect company size and type, the rep working the deal, even the time of the year all complicate whether a deal will close.
The number companies get when they multiply their total pipeline deal value by each deal’s probability rarely comes close to the total revenue that team ends up closing at the end of the quarter. Sales doesn’t not work on a sliding scale. If a deal is lost, it is totally lost. Imagine a rep is working a $100,000 deal, and it’s rated with a 50% chance to close. Putting that deal in the forecast for $50,000 is misleading no matter the outcome. When it closes, won or lost, the forecasted value will either have been overshot or underestimated by $50,000. Obviously this method is meant to be applied across all deals in a company’s pipeline so that the underestimates and overestimates even out. But the number of deals that would need to be included in a calculation of this type for the result to be statistically sound is far greater than that which is in almost any company’s pipeline.
Path to Better Forecast Accuracy Metrics
Forecasts that rely on the “weighted pipeline” method rarely achieve more than 75% accuracy. More frequently, as reported by CSO Insights, they’re in the ballpark of 50%. It doesn’t take long for companies to pick up on this, and the forecasts themselves become an exercise in futility that ultimately get second-guessed or even disregarded entirely. Though it takes longer, the only way to achieve accurate, reliable forecasts is to assess each deal on its own merits.