Sales Forecasting Inaccuracys Last Refuge in the Big Data Era

One of the critical tasks for sales managers is forecasting. Forecasts drive a host of decisions in the organization, from investment in production to allocation of service personnel to the guidance offered to shareholders.

But, for some reason, it's increasingly acceptable for forecasts to be not particularly accurate. In a survey of sales and marketing pros I conducted for CallidusCloud, only about 12 percent of the respondents claimed their organizations' sales forecasting was “very accurate,” while about one in three said it was “inaccurate” or “somewhat inaccurate.”

The same survey showed that 32 percent of respondents felt that a forecast that was less than 80 percent accurate was acceptable. That number swelled to 42 percent among sales pros.

Those results should concern anyone in business. How do you anticipate delivering what customers need, define customer experiences, or justify your budget to your boss if you have no idea of what you're planning for?

Is there any other part of the business where 20 percent inaccuracy is acceptable? If HR hired 20 percent more people than were needed; if manufacturing produced 20 percent less product than was needed; if accounting was inaccurate 20 percent of the time; if professional services failed to meet customer expectations in one out of five engagements -- heads would roll.

But in sales, missing the mark by a fifth is acceptable. It shouldn't be.

Human-Driven Confusion

There are good reasons for inaccuracy, of course. Many sales organizations still rely on sales managers' intuition about their sales staffs to hone forecasts. Perhaps one rep is a people pleaser and habitually moves leads down the funnel just to keep his manager happy. Maybe another one is a rampant optimist and sees every deal as on the verge of closing. Perhaps a third is trying to manipulate quota expectations and holds back her reporting of deals' progress.

All these human factors can take the mathematical exercise of forecasting -- looking what's in the funnel and at what stage, and making an informed guess that gives greater weight to deals closer to closing -- and throw it into utter chaos. The reality of the funnel may be completely at odds with the version of the funnel reported by the sales staff.

Sales managers' intuition is important in this scenario, but what happens when you promote a new sales manager who may not yet have history with his sales team? In this case, his first forecast may be wildly off the mark. The repercussions on the business are significant -- but so is the impact on the sales manager's confidence and reputation. The learning curve is steep and fraught with peril.

And judging by the responses in the survey, this condition of human-driven confusion is seen as acceptable by a world of sales managers resigned to the prevailing conditions. In fact, given the nearly infinite variables that managers face, getting within 20 percent of accuracy is a pretty neat trick.

Taking the Historical Perspective

However, even as sales managers wrestle with the ongoing struggle of forecasting, we're constantly reminded that we're in a new era of Big Data -- and not just big data, but predictive analytics.

Businesses are starting to harness the data they're collecting that illustrates the history of things -- ranging from insurance actuarial numbers to refrigerators' performance -- and using that data to anticipate future activities and customer needs. In other words, they are forecasting an aspect of the business in order to serve up what the customer needs and wants before he or she even recognizes it.

If we're already doing this for customers, why can't we do it for our own internal sales forecasting?

We can -- if we know which data is important for the task. We already collect a lot of data about individual sales behavior via compensation management software. Examining performance over the last several quarters can reveal patterns in sales behavior for each sales rep in the organization. This becomes a great real-life substitute for what the sales rep has been reporting in the past.

For instance, if a sales rep reports a funnel worth $500,000 this quarter, but his past performance has shown him to miss forecasts by an average of 20 percent for the last four quarters, it would be reasonable to factor that in and reduce his contribution to $400,000. The same can be done for reps who underestimate.

New sales reps without a history could present a monkey wrench -- unless you use an aggregate number of reps in their first, second or third quarter with your company as a substitute for the new rep's historical performance.

Using such data takes the errors introduced by personality and politics during the forecasting process and sidelines them for a reality anchored in history. Doing so doesn't take all the guesswork out of forecasting, but it does shift the sales manager's attention to more important things, like anticipating changes to the forecast caused by new product introductions, new sales technology and new sales techniques.

Instead of being forced to focus on the personality quirks of individual reps, managers can focus on the large-scale impacts of major changes. They should also gain more time to coach their reps about selling skills instead of dwelling on the procedural nitty-gritty of forecasting.

The big-data era isn't about collecting more data, in most cases -- it's about using the data we already collect the right way. Turning forecasting from a guessing game into a data-driven exercise that's based on past results is one example of that -- and it's an example that can help your entire business operate more efficiently and predictably. 

Creative Commons Creative Commons Attribution 2.0 Generic LicenseTitle image by  kudumomo