Want a more accurate sales forecast? Start with past behaviours

Everybody gets it: if you understand data about your customers, you understand what will get your consumers to buy, buy again and buy more. It’s an equation so well understood that period-piece TV shows like “Mad Men” use it as a plot point. Sterling Cooper Draper Pryce would not have devoted significant office space to an enormous IBM machine if it didn’t generate insight they could use.

Even 50 years ago, businesses understood that, on the buy side of the equation, using data – and the right technology to analyse it – is vital. In the early 1970s, we used mainframes; today, more elegant analysis of CRM data the application of predictive analytics is the norm. We follow every step in the path to purchase to make predictions and develop strategies that We wouldn’t dream of using guesswork to design a marketing strategy or media buy.

But while data-driven predictive analytics is well entrenched in the buy side, when it comes to people on the sales side of the transaction, we’re still using voodoo to make predictions and formulate incentives.

Sales forecasting and – more specifically – forecasting individual sales performance remains an imprecise and subjective activity. Typically, it’s performed by sales managers using educated guesses – perhaps some historical data, reports from reps tempered by reps’ past forecasting behaviours, an eye on current market behaviour, and a host of other fuzzy factors. Forecast accuracy depends largely on the sales manager’s ability to read these tea leaves and use his or her intuition to arrive at next year’s goal number. This isn’t because imprecision is acceptable – it’s because forecasting is hard, and guesswork has long been the only tool available.

As a result, our standards are astonishingly low. A recent study by CallidusCloud found that forecasting accuracy of less than 80% is deemed acceptable by four in 10 marketers. And about one in 10 said that accuracy below 60% was acceptable. Where else in a business strategy would we (or shareholders) tolerate that margin of error?

A key problem is that, unlike with our customers, we haven’t had a sense of which behaviors and signals we need to measure in order to predict the performance of a sales professional or team. Nor have we had a proven model for analysing those signals.

But that’s changing.

Digital footrprint

As in just about every area of business, predictive analytics is now becoming a tool for sales forecasting. New projection models are emerging specifically for the salesforce, targeting behaviors of individual salespersons that have shown to objectively predict future performance. Fortune 500 companies are beginning to implement these models to develop a more rigorous approach to assessing and incentivising salesforce and channel activities.

So, what exactly is it they measure?

Interestingly, much like the 'path to purchase' of a consumer is revealing, the digital footprint of a salesperson (how they use sales tools) coupled with historical data on incentives and commission structures, can unlock accurate forecasting on future efforts – i.e. whether or not sales goals will be met.

Many large companies already have considerable knowledge about their sales teams’ performance captured in their compensation systems, and companies using automated compensation management can easily extract and visualize this data.

In order to translate that into a forecast, however, the next step is to pair compensation data with the 'digital footprints' of sales reps and create individual profiles of each of them. Translating this into the language of customer profiling, we might think of this as an ideal sales 'persona'.

Telltale footprints might include the individual’s use of sales technology, training, coaching and sales enablement resources. Tracking these kinds of activities against sales performance provides an indication of how similar or additional activities could predict and improve project outcomes for corporate sales efforts, sales teams and individual sales reps.

Analysing and visualising the data efficiently requires capable technology, of course. But such a predictive platform, used in conjunction with a flexible compensation plan, can inform a range of decisions. The right tools can be used not only to predict, but also to influence individual sales rep behaviour. Current commissions plans may drive deals, for example, but perhaps modifying the plan to include customer experience metrics or other factors can drive even higher performance over the long term. The right analytics can reveal these sorts of opportunities using data instead of guesswork.

Combining a sales person’s performance history with their measured activity promises to improve forecast accuracy and, through gap analysis, provide insight for decisions on everything from territory management to sales training programs.

New predictive modeling techniques will enable businesses to project the effects of changing behaviour on bottom-line sales. They will also provide the confidence to change how sales reps are motivated without conducting a risky experiment. Tomorrow’s performance-driven salesforce will be built on data-driven methodologies.

Giles House is SVP at CallidusCloud.