In our first blog in this series, we provided an overview of the myriad changes to sales and marketing caused by artificial intelligence and machine learning. With new channels and methods, the world of selling is changing—and with it, the role of the salesperson. As such, these key areas of your sales compensation program must be reconsidered as well:

  1. Pay level and mix: Pay level is determined by both demand for talent and the skill level required. Some could argue that fewer people will be needed for certain aspects of the sales process. This could be due to the replacement of some activities with a machine or other channels, higher salesperson productivity due to AI, or both. This could put downward pressure on demand for some roles and, therefore, pay levels.

    On the flip side, many sales roles are being asked to be much more    sophisticated in what they do and how they do it. They are coordinating with multiple channels (or orchestrating the channels, in many cases), which requires a much higher (and potentially different) level of skill than one-on-one selling. This will put upward pressure on pay levels for the necessary skills. While it’s fair to conclude that the salesperson of the future will require new skills and competencies to succeed, the net impact of these two competing forces remains to be seen.

    As for pay mix, this is largely determined based on the influence of the salesperson. If we believe that some roles have less of an impact on the final sale, a case could be made for a lower percentage of their pay at risk. However, we have seen sales rep influence be reduced in other industries (pharmaceuticals, for instance), with no reduction in the percentage of pay at risk. In these cases, companies believe that there’s a first-mover disadvantage, as they believe their best talent will go elsewhere where they can make more incentives (because these salespeople believe they still do have a major impact on results). We do not anticipate a major change in pay mix in the near future due to AI.
  2. Metrics: AI is not meant to replace the salesperson or stop them from thinking and exercising their judgment. In fact, the best applications that we have seen to date are aimed at enhancing the salesperson function by presenting prioritized recommendations, simply and in the most relevant manner. Given that salespeople are getting much more direction about their sales activities and in some cases pricing, should we consider paying on activities, not results? Historically, best practice suggests the opposite: pay on results, not activities.

    Consider the following four scenarios of a salesperson adhering to the recommended actions, and the results those actions drive:



Does not achieve results

  Achieves results

Adheres to      recommended actions

Model incorrect or poor execution quality of actions

  Same payout using either metric

Does not adhere to recommended actions

Same payout using either metric

 Model incorrect or rep wins without best action due to skill, luck or will


If adhering to the recommended actions ends in the desired result (upper right and bottom left boxes in the chart), then paying on adherence or paying on results will provide roughly the same payout. But what if the model is wrong—no model is perfect—yet the salesperson adheres to it and gets a lower result than they otherwise might have (upper right)? We would be penalizing them for not achieving the desired results despite giving them increased direction on how to get the results. Or what if the opposite happens, where a salesperson goes off script but still achieves the desired results? Would we penalize someone who delivered what was expected even if they did not adhere to the suggestion engine?

For now, companies that have adopted AI have stuck to results-based metrics for their sales compensation plan. Their logic is that results pay the bills, and that the AI-enhanced methods should help salespeople achieve their results. But as models continue to get better, you can imagine a world where paying for desired activity becomes at least a part of the incentive plan.


3. Plan design: This is an area where we believe AI could impact salespeople the most in the long term. Just as AI has determined the best way to sell to a customer, it could also be used to know how to best motivate a salesperson. For example, while money may be the primary motivator for many salespeople, each individual is different. There are four ways in which people are motivated:


Control: Our need for choices and to be the master of our destiny

Affiliation: Our need for social contact and cooperation

Reward and recognition: Our need to be acknowledged and appreciated

Excellence: Our need for accomplishment and growth

Machine learning could determine the optimal motivators for an individual based on his relative profile and recommend a certain sales compensation plan at the beginning of the year (but still give him his choice of plan out of multiple options). Taken even further, the plan could offer point-in-time incentives based on what has proven to work with salespeople of a similar profile. It could differentiate not only based on salesperson-inferred preferences but also by tenure and geography, or any other factor proven to influence the effectiveness of a particular sales incentive plan.

Or it could be as simple as providing timely reminders based on salesperson preferences and current performance level. In a simple example, insurance agents were notified via an incentive app if they were within striking distance of achieving the next level of incentive. Using a test and control group, the app proved to have a significant impact on results.


There’s no question that AI and machine learning are impacting sales in many ways. As a consequence, forward-thinking sales compensation leaders are considering how these changes may impact the way that we motivate the sales organization. In the third and final part of this series, we will look at how AI will change the way we think about sales goals.