How Predictive Opportunity Scoring Can Help You Prioritize Deals

We all succeed at times and fail at times. And then, we look back in retrospect and wonder how things could have gone differently.

But if we could start our endeavors with this hindsight, we could avoid many failures.

Psychologists call it “prospective hindsight,” but for the purpose of this post, we’ll refer to it as retrospection.

In my role, I work with sales teams from large enterprises to growing startups. Using the predictive applications we build, we help them increase their odds of winning so they can sell more.

Predictive opportunity scoring

Using machine-learning techniques, we score sales opportunities to determine the likelihood of winning a deal.

The opportunities are then categorized as

• Very Likely
• Likely
• Less Likely
• Not Likely

Sales reps can then use these scores to prioritize the opportunities they work on.

While how these scores are used by sales managers and reps is interesting, I wanted to shed light on how predictive scoring could be used as a tool for retrospection.

Customer example

XANT was recently working with a large technology enterprise that was keen on seeing what predictive analytics could offer them and requested we build them a predictive model for opportunity scoring.

XANT utilizes a unique temporal database to catalogue every change an opportunity undergoes as time elapses. Using this high-definition sales data captured from CRM and combining it with external buying signals, we developed a predictive model unique to this customer.

Not only did we score their opportunities in the current pipeline, we also scored every version of the opportunities they have closed in the past quarter.

The results were fascinating. By comparing the scores of closed opportunities to the actual outcome of the opportunity (i.e., a win or a loss), we were not only able to prove the accuracy (for the technically minded, this is both looking at the lift in win rates as well as precision/recall measures) of our predictive scoring, but also shed light on how the sales team could have benefitted from appropriately prioritizing the deals based on the scoring.

Based on their average sales cycle of 127 days, we looked at scores 1, 30, 60, 90 and 120 days before the opportunities closed (see Figure A). Interestingly, 98 percent of the opportunities we labeled as “Very Likely” the day before opportunities closed, actually ended up winning and less than 1% of the opportunities classified as “Not Likely” won.

More importantly, the scores of opportunities were accurate even 120 days before they closed.

This is a distinction from other emerging players in the market, who do sales lead/opportunity scoring, and claim “high accuracy” based on scores just before the opportunity closes. The truth is that most sales reps can “predict” an opportunity win/loss the day before it closes.

While this analysis was in retrospect, the actionable insight was looking at this customer’s current pipeline and how these open opportunities were scored. This is because the results seen in the last quarter are indicative of what will follow in this quarter (by making sure the model is not overfit for the closed opportunities) (Figure B).

This company had about 36 percent of their opportunities in the “Not Likely” category. By deprioritizing these deals, the sales team could significantly increase their win rate, but more importantly, they could sell more by spending more time on deals they were losing but still had a decent chance of winning (i.e., Likely and Less Likely deals).

Now the sales team has a “prospective hindsight” in their arsenal to sell more!

Are you interested in providing this “prospective hindsight”or retrospection to your sales team? Do you want to empower your sales organization with predictive and prescriptive insights?

Download this free ebook.