Blog

Check out our up-to-the-minute thoughts and insights into goings-on. Find useful links and ideas updated frequently.

Machine-Learning-Main-Image.jpg

Machine Learning for Optimal Compensation Plans

Advanced predictive analytics is slowly making its way into the broader Sales Performance Management domain. More organisations are investing resources in machine learning techniques and how they can augment sales effectiveness.  This ranges from utilizing historical data for better forecasting, real time feedback to sales reps during quoting and pricing, supervised learning in territory management and a gamut of other aspects of a sales cycle where machine learning functions in the peripheral.

Incentive Compensation, however, still exists predominantly within the realm of behavioral economics and psychology. Decisions on which incentives will drive desired sales behavior across the sales force are based more on intuition than analysis. The manual analysis that an admin can perform is often restricted to a few years of historical data and a limited number of variables. Often, the result is an overly convoluted compensation plan that is difficult for sales reps to understand and fails to achieve the desired sales behaviour.

If you are considering improving your sales performance programs through machine learning processes, consider Intangent’s Current and Future State Analysis strategic service.  Your organization will gain valuable clarity into current pains, future needs and how to bridge the gap.  Learn More Here.

Why Machine Learning?

Why-machine-learning---makes-you-betterApplying machine learning techniques to historical sales and compensation data can help reveal hidden patterns that manual analysis cannot and in turn, help design a more intelligent compensation plan. The right quota, product portfolio, territory assignment, bonuses, commission rates and many other incentive metrics can be derived from the analysis of historical performances. Consider these incentive compensation plan questions:

  • Is a commission rate hike of 1% on a product the best incentive for a desired product revenue growth of 30%? Or will an annual kicker bonus will provide better results?
  • Who are the best reps for a product in a new territory aimed at a specific demographic?

An optimal compensation plan will be a result of an analytical approach to these questions and not an intuitive one. Such a plan not only utilizes historical data to provide optimal incentive metrics but it continues to adapt based on the results over time. The predictive machine learning models put in place for designing an optimal compensation model can also act as trend checkpoints. If the trends in the current year do not fall in line with the desired outcomes, these predictive models can catch it faster and recommend corrective action before it is too late.

The goal of every compensation plan is to drive the desired sales behavior with minimum incentive overhead. Compensation plans based solely on intuition or limited manual analysis tend to over- or undershoot these incentives.  For example, an organisation would not want to increase the commission rate on a product by 3% for a revenue growth of 50% if the same targets could have been met with a 2% commission rate. Also, that same organisation would not want to miss that target by not providing enough incentive for the sales team. Minor suboptimal rates and bonuses in a compensation plan can have a substantial financial impact. The hidden underlying sales behavior of a sales force, an individual’s motivation or a team’s response to subtle comp plan changes can only be unravelled by utilizing the heaps of sales data that organisations accumulate over time but hardly use. Utilizing the power of machine learning to consume and interpret this data can result in deeper insights into sales behaviour.

Where to Start?

The good news is that there are already several resources out there to get started on the path to an optimal compensation plan. Most SPM vendors have some sort of advanced analytics capability. IBM Watson, Xactly Insights and CallidusCloud Thunderbridge are an ideal choice for organisations already using those respective platforms for calculating incentive compensation. More than the choice of analytical software for predictive analytics, it’s the preparation of data for such analysis that is the more daunting task and often dissuades organisations from such exploratory analysis.

Any machine learning algorithm is only as good as the data that it is learning from. Identifying a problem statement, recognizing the pertinent data elements that could help solve the problem, filtering and preparing that data in a form that is usable for analysis are essential precursors to any predictive model. Once these prerequisites are met, the modeling can be performed via a proprietary software or a custom simplistic machine learning algorithm.

Conclusion

Sales teams are the driving force for any organization and an optimal compensation plan that continues to motivate them year after year is the one that learns and adapts. There is an opportunity to fine tune compensation plans based on advanced analysis of sales and compensation data. There is tremendous volume of sales data with hidden information and an array of tools available to mine that information. Using machine learning to optimize every aspect of sales cycle is not merely a theoretical concept anymore. Tangible results are now within every organization’s reach.

If your organization is currently considering augmenting sales processes through machine learning, consider Intangent’s Current and Future State Analysis.  We can help accelerate the achievement of your organization’s long-term sales goals by providing you clarity and visibility into overcoming your sales struggles.  Schedule a meeting with Intangent here.

Subscribe to Our Blog