A recruiter I met recently said they had eliminated incentive-based bonuses for executives. Performance, much to his surprise, improved.
Pay-for-performance. Makes sense. I don’t want to pay for no performance. That makes no sense. Yeah, but…
Paying for performance creates incentives. You’d like those incentives to:
- Encourage “good” behavior &
- Discourage “bad” behavior.
Pay-for-performance sometimes does just the opposite. Incentives are weird. Humans are weird. Systems are weird.
Definition
To make sure we’re all talking about the same thing, I’m talking about compensation that works like this:
- At the beginning of the year (or whatever cycle makes sense) the payer and the performer agree on targets for agreed measures.
- During the year compensation is less than it otherwise would be.
- At the end of the year the measures are evaluated and, assuming the goals are met, additional compensation is paid.
- A football player makes a million bucks a season and another million if they score 10 goals.
- A marketer makes a hundred grand plus another fifty if brand recognition goes up 10 points.
- A jewelry seller gets a $20 spiff if they meet their daily sales quota.
Variations
Incentive bonuses offer several dimensions along which to vary, each creating its own tradeoffs. These tradeoffs then combine with each other and with human biases to create endless combinations.
- Duration
- Scope of measures
- Where measured
- Proportion of compensation
- Binary or sliding
Duration. Daily sales goals to annual review, duration should be long enough that performer decisions can make a difference and short enough that the proximity of the payoff influences performer decisions.
(Pay-for-performance creates an automatic incentive to stick around until the bonus is paid. It’s worth taking this into account if you want/don’t want performers to stick around.)
Scope of measures. “Your” sales to company-wide measures. The wider the measure, the less control the performer has & the greater the chance of free riders. The narrower the measure, the greater the chance for gaming the measure and the Pie Problem*.
Where measured. Measure effort? Output? Outcomes for users? Impact coming back to the company? The earlier in the loop, the more control the performer has but the greater the chance for misalignment with overall goals.
Proportion of compensation. If base compensation is sufficient for basic needs then bonuses are less vulnerable to abuse, but the range goes from all bonus to bonuses being a nice, well, bonus.
Binary. Goals can be binary, come in several steps, or just be treated as sliding scale. Bonuses similarly. The bigger the steps, the more performers will twist their behavior. (E.g. Sales closing all of their business on the last day of the quarter, sometimes to the detriment of the company.)
Eliminate Bonuses?
Now we have enough background to re-examine the opening example. Why would eliminating bonuses for executives improve performance at a high-growth company?
In a high-growth, pre-IPO company everyone already has a powerful pay-for-performance bonus in the form of equity. According to the above rubric:
- Duration: indefinite but at least a few years.
- Scope: whole company.
- Where measured: this one is a little weird because “the market” does the measuring & their measure is fickle.
- Proportion: anywhere from ridiculously high for founders to modest for recent hires. Proportion changes as shares appreciate.
- Binary: equity only has real value after a liquidity event.
Given where pre-IPO lands on the various bonus dimensions we’d expect it provide powerful incentive to cooperate & little incentive to game the system.
Value is unlocked in high-growth companies from the discovery and rapid exploitation of unexpected synergies, like Gusto’s recent launch of a checking account and debit card for employees paid through Gusto’s payroll service.
Individual bonuses incentivize short-term, small-scale, low-risk improvements, none of which add substantial value to a high-growth company. From this perspective, eliminating bonuses should indeed improve performance.
Proper Context
Pay-for-performance makes sense while extracting value from known sources where the chance of unexpected synergy is small. Say you have a hit game. You want to expand and extend its profitability. Marginally better execution (marginal in the “at the margin” sense, not the “who cares?” sense) matters. Incentivizing small improvements make sense (as long as those improvements don’t risk the big stream of profit).
People ≠ Machines
I went through a period early in my exploration of software process where I thought of people as machines with a whole bunch of control knobs. My job was to find the exact right settings for those knobs to get the behavior I wanted. This was inhumane and ineffective, which I quickly figured out. Okay, eventually figured out.
My recent thinking about incentives reminds me uncomfortably of those thoughts. Here is my caveat to me (and you): human incentives are powerful and subtle. Purpose, vision, meaning, belonging, autonomy, and respect all move people to a greater degree than games of money with complicated rules.
Rules backfire. Easily. But they matter too.
When implementing monetary incentives, think about the organizational problem you are addressing (I almost wrote “solving”, duh). Think about the incentives you are creating, positive & perverse. Then pay attention to how incentives land in practice. You and those subject to pay-for-performance live in different worlds. Prepare to adjust.
- Pie Problem — when it’s easier to claim a greater share of credit for value than it is to create more value.
