Sales Incentives and Machine Learning: Intelligently Motivate Revenue-Driving Behaviors 

Kamal Ahluwalia | Aug 29, 2017 270 views 1 Comment

Share on LinkedIn

Imagine you are trying to get children to do household chores for the summer. You decide to offer them an incentive of $5 per completed chore. At the beginning of summer, this works great; work is getting done and the kids enjoy the reward. However, after a few weeks, the system has fallen apart. The children only do the chores they enjoy, they neglect other jobs, and the quality of their work has decreased. Should you change the reward amount? Should different chores have different incentives? Should each child receive different rewards? Did this system work better last year? What do your neighbors do to get their kids to do chores?

In this simple example of human behavior, there is a broken incentive system and a lack of data to help determine how things should change. This scenario is not a far comparison to many organizations’ sales compensation systems. Every year, new compensation plans roll out to the sales teams who go out and sell based on how they believe they will be compensated. As of 2017, 90 percent of U.S. companies change their compensation plans on an annual basis. But are these adjustments optimal – are they going to lead to desired business outcomes? Is there a better way to identify which changes need to be made, and when to make them?

Traditionally, sales compensation changes are based more on intuition than data, but often there is some level of analysis that goes into designing and adjusting incentives. However, most reports today are all hindsight and adjustments are made well after there is any potential to change behavior before the end of the quarter or end of the year. What if optimal changes could be identified mid-period and put in place well before it is too late? Better yet, what if the incentive compensation solution itself could identify, recommend and implement plan changes? The most practical way to do this is with machine learning.

The concept of machine learning is not exactly new in the world of technology. It has been around in a variety of forms for decades, but its application to enterprise software is relatively new, particularly with sales incentive management, and its popularity is growing. There are many factors making machine learning realistically applicable in the business world – changes in the economics of cloud computing (cheaper than ever before), cloud storage, proliferation of sensors driving Internet of Things (IoT), pervasive use of mobile devices that consume gigabytes of data in minutes, and freely available algorithms are all major contributors to accelerating machine learning adoption. Add to these the complex problems companies face including managing sales compensation, and the perfect environment is in place for machine learning to dramatically proliferate.


Machine learning is all about applying learned data to prescribe more economically efficient business decisions. Sales incentives apply the disciplines of psychology and behavioral economics to prompt people to make desired decisions. When you combine principles of behavioral economics with the data science of machine learning, you create the potential to optimize your sales incentives and drive powerful business outcomes.

Machine learning allows us to assess large sets of data and surface patterns, identifying when past performance is indicative of future results. For instance, machine learning can accurately forecast what products are most likely to be sold and which customers are most likely to buy. But what if you not only want to understand potential outcomes, what if you want to completely change outcomes?

What is going to motivate your sales team to do what you need them to? The difference between expectations and reality is often referred to as the “behavioral gap” (see chart below). When the behavioral gap is significant, an inflection point is needed to close that gap. The right incentive (an added bonus, President’s Club eligibility, a promotion, etc.) can initiate an inflection point and influence a change in behavior.

The behavior gap depicted above represents the difference between raised expectations (management increasing quota) and the trajectory of current sales performance.

The behavior gap depicted above represents the difference between raised expectations (management increasing quota) and the trajectory of current sales performance.

In the US, studies from Harvard Business Review and other industry publications estimate that companies spend over one trillion dollars annually on incentives. That number is four times the money spent on advertising in the US annually. What that means is that, as a nation, we are deeply invested in motivating our employees, partners and customers. Incentives are most effective when they are intelligent, or data driven. Deloitte University Press published a report stating that when it comes to the relationship between data science and behavioral science, “it is reasonable to anticipate better results when the two approaches are treated as complementary and applied in tandem. Behavioral science principles should be part of the data scientist’s toolkit, and vice versa.”

With Machine learning and behavior mechanics, sales teams can plot out the path from one goal to the next and analyze and implement proper incentives. As an example, let’s say your company is a furniture manufacturer that uses a CPQ tool to manage its complex quoting and pricing processes. One of the major reasons your company invested in the CPQ solution was to curb chronic, costly discounting by the sales team. You are a new sales rep using CPQ to build a quote. What if, mid-quote, your system alerts you that the discount you entered, while within the approved range, may not be ideal. Machine learning ran in the background and identified a different discount used by the top 10% of reps that has had more success. Additionally, you learn that if you choose the prescribed discount, you will earn 40% more commission! Talk about a relevant incentive, based on powerful data.

When applied together, machine learning and sales incentives provide powerful business results by collecting relevant, timely insight and defining incentives that align human behaviors with organizational goals.

Print Friendly, PDF & Email

Recent Editor's Picks:

Categories: BlogCustomer AnalyticsEditor's PickSales Performance


One Response to Sales Incentives and Machine Learning: Intelligently Motivate Revenue-Driving Behaviors

  1. Andrew Rudin August 30, 2017 at 7:12 am (220 comments) #

    Hi Kamal: thanks for this interesting article, and for sharing how AI can influence incentive-pay systems. In the example you cite, I expect the algorithm would require visibility into the opportunity details, including competitive situation, client budget, purchase timeframe, and other important risk information. Absent that, I am not sure how AI can make an informed – let alone, useful – recommendation to the rep about the magnitude of the discount to offer. In fact, if the feedback to the rep is simply the amount of additional commission to be made (presumably by offering a smaller discount), why not just recommend selling at full list?

    Somewhere, the AI system must make an assessment of the vulnerability of the deal by considering demand elasticity. I am not sure how that can be done via AI without examining a broad population of similar accounts. Many times, none exist.

    A second, related issue: I see value in using AI to design incentive compensation systems, but AI is not good for assessing ethical conflict and potential harm to stakeholders – a major issue in designing incentive compensation systems. What safeguards can be implemented to ensure that an algorithmically-generated incentive doesn’t drive dysfunctional behavior?

    I look forward to learning your thoughts on these.

Add Your Comment (All comments are reviewed by moderator, no spam permitted!)