One engagement, in particular, had a profound influence on my career. The project involved working with the Asia-based sales force of a global consumer products company. This company practiced “route sales,” which meant reps spent their days visiting mom-and-pop convenience stores, servicing accounts. One thing about the organization surprised me: Its sales managers spent inordinate time listening to the reps complain about their compensation.

The complaints were based on what the reps saw as a myriad of problems. Their quotas were set too high, so they couldn’t possibly reach them. Or their territory was subpar, limiting their ability to sign new accounts. Sometimes the complaints focused on fairness: A rep who was hitting his quotas and making decent money would want a manager to do something about a “lazy” colleague who was earning outsize pay simply because he had a good territory. Imagine any conceivable complaint a salesperson might have about compensation, and I guarantee that sales managers at my client’s company had heard it.

The reps weren’t the only ones obsessed with the compensation system. The company liked to play around with the system’s components to try to find better ways to motivate reps and boost revenue, or to increase the return on the money it spent paying salespeople—a large part of its marketing budget. This company’s sales comp system was fairly basic: Reps earned a salary and a commission of around 1% of sales. The company worried that the system was too focused on outcomes and might over- or under-reward reps for factors outside their control. So it began basing compensation on their effort and behavior, not just on top-line sales. For instance, under the new system, a portion of the payment was based on customer satisfaction surveys, the number of prospective accounts visited (even if they didn’t buy), and the retention of existing accounts.

Primarily because of this consulting assignment, I became so curious about the best ways to compensate salespeople that I began reading academic articles on the subject. Eventually, I pursued a Ph.D. in marketing at Yale, where I studied the theory and practice of how companies can and should manage and pay salespeople—research I now continue at Harvard Business School.

Although there are fewer academics studying sales force compensation and management than researching popular marketing subjects, such as the use of social media or digital advertising, in the past decade, it’s become a fast-moving field. Some of the fundamental theories established in the 1970s and 1980s still apply that academics are testing hypotheses using two methods new to this area of research. An empirical analysis of companies’ sales and pay data and field experiments in which researchers apply various pay structures to different groups of salespeople and then compare the groups’ effort and output.

This new wave of research is already providing evidence that some standard compensation practices probably hurt sales. For instance, the study suggests that caps on commissions, which most large companies use, decrease high-performing reps’ motivation and effort. Likewise, the practice of “ratcheting” quotas (raising a salesperson’s annual quota if he or she exceeded it the previous year) may hurt long-term results. Research-based on field experiments (as opposed to the lab experiments academics have been doing for many years) is also yielding new insight into how the timing and labeling of bonuses can affect salespeople’s motivation.

In this article, I will take readers through the evolution of this research and suggest the best ways to apply it. With luck, this knowledge not only will help companies think about better ways to compensate salespeople but also might mean that their managers spend fewer hours listening to them gripe about unfair pay.

The Dangers of Complex Compensation Systems

The principal-agent theory has long guided researchers studying sales force compensation. This theory, drawn from the field of economics, describes the problem that results from conflicting interests between a principal (a company, for instance) and an agent hired by that principle (an employee). For example, a company wants an employee’s maximum output. Still, a salaried employee may be tempted to slack off and may be able to get away with it if the company can’t observe how hard the employee is working. Most incentive or variable pay schemes—including stock options for the C-suite—are attempts to align the interests of principals and agents. Commission-based plans for salespeople are just one example.

Salespeople were paid by commission for centuries before economists began writing about the principal-agent problem. Companies chose this system for at least three reasons. First, it’s easy to measure the short-term output of a salesperson, unlike that of most workers. Second, field reps have traditionally worked with little (if any) supervision; commission-based pay gives managers some control, making up for their inability to know if a rep is visiting clients or playing golf. Third, studies of personality type show that salespeople typically have a larger appetite for risk than other workers, so a payment plan that offers upside potential appeals to them.

During the 1980s, several important pieces of research influenced firms’ use of commission-based systems. One, by my Harvard colleague Rajiv Lal and several coauthors, explored how the level of uncertainty in an industry’s sales cycle should influence pay systems. They found that the more uncertain a firm’s sales cycle, the more a salesperson’s pay should be based on a fixed salary; the less unpredictable the cycle, the more pay should depend on commission. Consider Boeing, whose salespeople can spend years talking with an airline before it places an order for new 787s. A firm like that would struggle to retain reps if pay depended mostly on commissions. In contrast, industries in which sales happen quickly and frequently (a door-to-door salesperson may have a chance to book revenue every hour) and in which sales correlate more directly with effort and so are less characterized by uncertainty, pay mostly (if not entirely) on commission. This research still drives how companies think about the mix between salaries and commissions.

To get the optimal work out of a rep, you should, in theory, tailor a comp system to that individual.

Another important study, from the late 1980s, came from the economists Bengt Holmstrom and Paul Milgrom. In their very theoretical paper, which relies on many assumptions, they found that a formula of straight-line commissions (in which salespeople earn commissions at the same rate no matter how much they sell) is generally the optimal way to pay reps. They argue that if you make a sales comp formula too complicated—with lots of bonuses or changes in commission structure triggered by hitting goals within a specified period—reps will find ways to game it. The most common method of doing that is to play with the timing of sales. If a salesperson needs to make a yearly quota, for instance, she might ask a friendly client to allow her to book a sale that would ordinarily be made in January during the final days of December instead (this is known as “pulling”); a rep who’s already hit quota, in contrast, might be tempted to “push” December sales into January to get a head start on the next year’s goal.

While a straightforward comp plan such as the one advocated by Holmstrom and Milgrom can be appealing (for one thing, it’s easier and less costly to administer), many companies opt for something more complex. They do so in recognition that each salesperson is unique, with individual motivations and needs, so a system with multiple components may be more attractive to a broad group of reps. To get the optimal work out of a particular salesperson, you should in theory, design a compensation system tailored to that individual. For instance, some people are more motivated by cash, others by recognition, and still others by a noncash reward like a ski trip or a gift card. Some respond better to quarterly bonuses, while others are more productive if they focus on an annual quota. However, such an individualized plan would be extremely difficult and costly to administer, and companies fear the “watercooler effect”: Reps might share information about their compensation with one another, which could raise concerns about fairness and lead to resentment. So for now, individualized plans remain uncommon.

Concerns about fairness create other pressures when designing comp plans. For instance, companies realize that success in any field, including sales, involves a certain amount of luck. If a rep for a soft-drink company has a territory in which a Walmart is opening, her sales (and commission) will increase. Still, she’s not responsible for the revenue jump—so, in essence, the company is paying her for being lucky. But when a salesperson’s compensation decreases owing to bad luck, he or she may get upset and leave the firm. That attrition can be a problem. Even though there are downsides to making a compensation system more complex, many companies have done this in the hope of appealing to different types of salespeople. By this approach limit the impact of luck by utilizing caps or compensating people for inputs or effort (such as a number of calls made) instead of simply for closing sales.

Using Real Company Data to Build Understanding

The big difference between earlier research on sales compensation and the research that’s come out in the past decade is that the latter is not based just on theories. Although companies tend to be very secretive about their pay plans, researchers have begun persuading them to share data. And companies have been opening up to academics, partly because of the attention being given to big data; managers hope that allowing researchers to apply high-powered math and estimation techniques to their numbers will help them develop better tools to motivate their workforce. Indeed, these new empirical studies have revealed some surprises, but they have also confirmed some of what we already believed about the best ways to pay.

Tom Steenburgh, a professor at the University of Virginia’s Darden School of Business, published one of the first of these papers, in 2008. He persuaded a B2B firm selling office equipment to give him several years of sales and compensation information. This unique data set allowed Steenburgh to look at sales and pay data for individual salespeople and use it to make assumptions about how pay influences behavior. The company had a complex compensation plan: Reps earned a salary, commissions, quarterly bonuses based on hitting quotas, an additional yearly bonus, and an “overachievement” commission that kicked in once they passed specific sales goals. He focused on the issue of timing games: Was there evidence that salespeople were pushing or pulling sales from one quarter to another to help them hit their quotas and earn incentive pay? That’s an essential question because pushing and pulling don’t increase a firm’s revenue, and so paying salespeople extra for doing that is a waste.

Even though the salespeople in the study could receive (or miss out on) substantial bonuses for hitting (or missing) quotas, Steenburgh found no evidence of timing games. He concluded that the firm’s customers required sales to close according to their own needs (at the end of a quarter or a year, say) and that the firm’s managers were able to keep close enough tabs on the reps to prevent them from influencing the timing of sales in a way that would boost their incentive payments. That finding was significant because quotas and bonuses are a large part of most sales compensation plans.

In 2011 Sanjog Misra, of UCLA, and Harikesh Nair, of Stanford, published a study that analyzed the sales comp plan of a Fortune 500 optical products company. In contrast with the firm Steenburgh studied, this company had a relatively simple plan: It paid a salary plus a standard commission on sales after achieving quota, and it capped how much a rep could earn in order to prevent windfalls from really big sales. Such caps are relatively common in large companies.

As they analyzed the data, Misra and Nair concluded that the cap was hurting overall sales and that the company would be better off removing it. They also determined that many reps’ motivation was hurt by the firm’s practice of ratcheting. Setting and adjusting quotas is a very sensitive piece of the sales compensation formula, and there’s disagreement over ratcheting: Some feel that if you don’t adjust quotas, you’re making it too easy for reps to earn big commissions and bonuses, while others argue that if you raise a person’s quota after a very strong year, you’re effectively penalizing your top performers.

Misra and Nair estimated that if this firm removed the cap on sales reps’ earnings and eliminated quotas, sales would increase by 8%. The company implemented those recommendations, and the next year companywide revenue rose by 9%.

A third empirical study of sales rep pay, on which I am the lead author, was published in Marketing Science in 2014. Like Steenburgh, we utilized data from a B2B office equipment supplier with a complex compensation plan. We examined how the components of the plan affected various kinds of reps: high performers, low performers, and middle-of-the-road performers.

We found that although the salary and straight commission affected the three groups in similar ways, the other components created different incentives that appealed to certain subsets of the sales force. For instance, overachievement commissions were important for keeping the highest performers motivated and engaged after they’d hit their quotas. Quarterly bonuses were most important for the lower performers: Whereas the high performers could be effectively incentivized by a yearly quota and bonus, more-frequent goals helped keep lower performers on track. Some people compare the way people compensate a sales force to the way teachers motivate students: Top students will do fine in a course in which the entire grade is determined by a final exam, but lower-performing students need frequent quizzes and tests during the semester to motivate them to keep up. Our study showed that the same general rule applies to sales compensation.

Researchers have begun persuading companies to share their data about their pay plans.

Our research also suggested that the firm would benefit if it shifted from quarterly bonuses to cumulative quarterly bonuses. For example, say a salesperson is supposed to sell 300 units in the first quarter and 300 units in the second quarter. Under a regular quarterly plan, a salesperson who misses that number in the first quarter but sells 300 units in the second quarter will still get the second-quarter bonus. Under a cumulative system, the rep needs to have cumulative (year-to-date) sales of 600 units to get the second-quarter bonus, regardless of his first-quarter performance. Cumulative quotas do a better job of keeping reps motivated during periods in which they’re showing poor results because reps know that even if they’re going to miss their number, any sales they can squeeze out will help them reach their cumulative number for the next period. In fact, even before we made our recommendations to the company in our study, managers there decided to move to cumulative quotas.

Out of the Lab, Into the Field

In addition to sharing sales and compensation data with academics, companies in the past several years have been allowing controlled, short-term field experiments in which researchers adjust reps’ pay and measure the effects. Before the use of field experiments, most academic experiments regarding sales force compensation took place in labs and involved volunteers (usual undergraduates) rather than real salespeople. Shifting from this artificial setting into actual companies helps make the results of these studies more practical and convincing.

Sales reps work harder for the chance to earn a reward than they do after receiving one.

As an example of one such experiment, consider recent work, my colleague Das Narayandas and I did with a South Asian company that has a retail sales force for its durable consumer products. The company uses a simple system of linear commissions—reps earn a fixed percentage of sales, with no quotas, bonuses, or overachievement commissions. Managers were interested in seeing how instituting bonuses would affect the reps’ performance, so over six months, we tested various ways to frame and time bonuses—always comparing results against a control group.

For one of our experimental groups, we created a bonus that was payable at the end of the week if a rep sold six units. For another group, we framed the bonus differently, using the well-known concept of loss aversion, which posits that the pain people feel from a loss exceeds the happiness they feel from gain. Instead of telling reps they would receive a bonus if they sold six units, we told them they would receive a bonus unless they failed to sell at least six units. To test the concept even further, the company’s managers suggested another experiment in which we paid the bonuses at the beginning of the week and then had the reps return the money if they missed the goal.

The results showed that all three types of bonuses exerted similar effects and that in every case, the group receiving the bonus generally outsold the control group. Loss aversion didn’t have much effect. We believe that’s partly because we were using cash, which is liquid and interchangeable; in the future, we might experiment with noncash rewards, such as physical objects.

We also tried to measure the impact on sales reps’ effort of cash payments that were framed as gifts (as opposed to bonuses). Whereas bonuses are viewed as transactional, research shows that framing something as a gift creates a particular form of goodwill between the giver and recipient. In our study, we used cash but told employees it was a gift because there were no strings attached—they didn’t have to meet a quota to receive it. We found that the timing of a gift directly influences how reps respond: If you give the gift at the beginning of a period, they view it as a reward for past performance and tend to slack off. If you tell them they will receive a gift at the end of a period, they work harder. We concluded that if companies want to encourage that kind of reciprocity, they need to pay careful attention to timing.

Other researchers are using field experiments to understand better how salespeople react to changes in payment schemes, but most of this work is so new that it hasn’t been published yet. One paper presented at a conference in 2014 showed that if salespeople receive cash incentives for passing tests about the product they are selling, they will sell more. (This is an example of sales compensation based on effort as opposed to results.) Another recent field experiment found that sales reps valued noncash incentives (such as points that could be used for vacations or items such as televisions) more than the actual monetary cost of the good the points could purchase. As more researchers and companies embrace the use of field experiments, sales managers will learn even more about the best ways to motivate their teams.

It Pays to Experiment

After spending a decade in academia studying sales force compensation, I sometimes wonder what would happen if I were transported back into my job as a management consultant. What would I tell sales force managers to do differently?

Some of my advice would be straightforward: I would urge managers to remove the caps on commissions or, if they have to retain some ceiling for political reasons, to set it as high as possible. The research is detailed on this point: Companies sell more when they eliminate thresholds at which salespeople’s marginal incentives are reduced. There might be problems if some reps’ earnings dramatically exceed their bosses’ or even rival a C-suite executive’s compensation. Still, the evidence shows that firms benefit when these arbitrary caps are removed.

I would tell sales managers to be extremely careful in setting and adjusting quotas. For instance, the research clearly shows that ratcheting quotas are detrimental. It’s tempting to look at a sales rep who blows through her yearly number and conclude that the quota must be too low—and quotas do need to be adjusted from time to time. But in general, it’s essential to prevent reps from feeling that unfairness or luck plays a part in compensation, and resetting quotas can contribute to that perception. And if something outside the salesperson’s control—such as an economic downturn—making it more challenging to hit a goal, I would consider reducing the quota in the middle of the year. It’s crucial to keep quotas at the right level to properly motivate people.

On the basis of my own research, I would advocate for a pay system with multiple components—one that’s not overly complicated but has enough elements (such as quarterly performance bonuses and overachievement bonuses) to keep high performers, low performers, and average performers motivated and engaged throughout the year.

Finally, I would urge my client companies to consider experimenting with their payment systems. Over the past decade, managers have become attuned to the value of experimentation (A/B testing, in particular); today, many consumer goods companies always experiment to try to optimize pricing. There are important lessons to be learned from doing controlled experiments on sales reps’ pay, because the behaviors encouraged by changes in incentives can exert a large influence on a firm’s revenue, and because sales force compensation is a high cost that should be managed as efficiently as possible. Involving academic researchers in these experiments can be beneficial: Having a trained researcher take the lead generally will result in a more controlled environment, a more scientific process, and more-robust findings. These studies also help the world at large, because research that improves how companies motivate salespeople will result in better and more-profitable businesses for employees and shareholders.