February 27 by Brion Schweers

Machine Learning influencing Sales Rep Behavior

Earlier this week I was scrolling through my news feed when I came across an article on how introducing machine learning into Incentive Compensation Management (ICM) could increase sales. Since using Behavior mechanics to influence outcomes is a key tenant of our Quote-to-Cash (QTC) philosophy, this post caught my attention. As it turns out, the article was written by one of my colleagues, Sarah Van Caster.

Modifying Sales Activity with Positive and Negative Stimuli

At the core of behavior modification is B.F. Skinner’s theory of operant conditioning. Skinner theorized that behavior could be modified through selection and that selection could be reinforced through positive and negative stimuli. Over time an individual will consistently make the correct selection and the incorrect selection will become extinct.

Understanding this basic principle of human behavior has been the foundation of all successful incentive compensation plans for many years. Sales reps will naturally determine how best to maximize their earnings based on the variables within their compensation plan. When these variables align with the goals of the enterprise, the compensation plan effectively aligns the behavior of the sales team with the goals of the business. For example, if launching a new product is a goal, then offering additional compensation when the new product is sold, is a form of a positive stimuli. Over time Sales will habitually sell the new product and the stimulus can be removed. Conversely, paying Sales less commission when they sell an obsolete product is a negative stimulus that will break their habit of selling that product.

Data Science Meets Human Behavior

As Sarah writes “Data science is a discipline that allows us to analyze the unseen”. She goes on to say that machine learning “allows us to look at large sets of data and surface patterns”. It’s these patterns that we are interested in when it comes to the Configure Price Quote (CPQ) process. With Apttus CPQ we use Machine Learning to provide: Cross-sell / up-sell recommendations, Pricing Intelligence and Quote Scoring. By “surfacing patterns” of similar customers purchasing comparable products we can predict one propensity to buy.

What is interesting is that Sarah’s article applies this same concept to rep behavior. The challenge in the past has always been in determining what degree of positive, or negative stimuli was required to modify the behavior of the sales team. Too much stimuli results in over compensating for the desired outcome, too little will result in failing to meet corporate objectives. The classic approach to optimizing compensation plans was to use what is known as a “Monte Carlo” simulation. This is interesting and machine learning certainly could play a role in analyzing past performance, identifying “behavioral gaps” and recommending changes to a plan. The challenge is that these are annual plans, so the opportunity for course correction would come in the form of SPIFFS and/or Promotions that could stimulate minor activity corrections.
But what if you could introduce stimuli in the day to day activities of the sales team?

How Data Science Optimizes the Deal

One way to do that could be to surface cross-sell/ up-sell opportunities during the quote process that helps the rep sell deeper into the sales catalog. Another might be to provide pricing guidance that helps the reps create quotes for as much as possible to win the deal, instead of as low as possible to still get approval. Reps know that the list price is too high and that the discount approval process is difficult. So, when they ask for an approval they will ask for all they can get on the first try. But what if the system could predict how much the customer would be willing to pay for the product and automatically apply that to the quote? As the sales team begins to realize that the optimized price is in fact the right price, they will begin to lead with that more often. The combination of these two capabilities will inevitably increase their deal size and their margin contribution.

If you add Quote scoring to the mix, now the system is making intelligent recommendations to the rep on how to increase the conversion rate of the quote. This might include adding a product and/or adjusting a price, but it might also include finding a coach or suggesting an executive briefing.

Optimizing the Deal and the Incentive

When CPQ and ICM are combined, there is an opportunity to provide further positive reinforcement by introducing Estimated Compensation. Once the rep has completed creating the quote, and it has been optimized by machine learning, the rep can view the estimated commission associated with this deal that is now larger, has more revenue and has an increase conversion rate.

Sarah goes on to say that if you connect Apttus Machine Learning capabilities with your sales activity tracking tools, the system could identify which activities have a direct correlation to sales. This knowledge could then be used to help sales managers coach their team members and further increase performance. It occurs to me that this might also be a good source of information to incorporate into your gamification efforts.

In summary

The next evolution of Sales Performance Management will include an integration of sales execution tools, incentive compensation tools and machine learning. When these technologies are applied together they can fundamentally improve your outcomes.

Download a complementary copy ($2495 value) of the Forrester Wave Report for CPQ to see how leading CPQ vendors are leveraging Machine Learning today.
Forrester Wave Report CPQ

Trending Blogs