The comp plan trap (and how to escape it without burning trust)
Every comp plan we’ve ever shipped was wrong within twelve months. The plan stays the same, but the business changes around it — new product lines, a different ICP, a shift in deal mix — and the incentives quietly start pointing the wrong way. Here’s how we caught it last year and what we changed without losing the team.
The setup
We sell a B2B platform with two product lines: the original product (mature, steady margins, $40k average ACV) and a newer add-on (smaller, faster sales cycle, lower ACV but higher attach rate to retention). Sales reps are paid commission on total ACV, with a flat 8% rate and accelerators above quota.
This plan was correct when we wrote it because the add-on barely existed. By the time the add-on was 22% of revenue, the plan was actively miscalibrated. Reps were rationally choosing not to sell the add-on, because the $200 commission on a $2,500 add-on wasn’t worth the time taken away from the next $40k deal.
How we noticed
We didn’t notice through the comp data. We noticed through the customer success team. CS started flagging that customers without the add-on had churn rates 1.7× higher than customers with it — and that this gap was widening. Our retention math depended on the add-on getting attached, but our incentives were quietly pushing reps the other way.
We pulled the data: in Q3, only 31% of new customers had the add-on, down from 54% the previous year. Our forecast for the year assumed 60%.
The hard part
Comp plans are emotional. Reps build their financial lives around the assumption that this quarter’s rules will be next quarter’s rules too. Mid-year changes feel like the rug being pulled out, even when the change is fair, because the fairness depends on a whole context the rep doesn’t fully see.
Our first instinct was the worst possible one: announce the new plan in the next all-hands and start it the following Monday. We didn’t do this, partly because someone in ops correctly pointed out that we’d lose two reps within a week. The cost of replacing them at full ramp is roughly $180k each. The expected gain from the new comp plan was about $300k a year. The math was bad even in the optimistic case.
What we did instead
Step 1: Tell the team there’s a problem, before the solution.
We held a 30-minute meeting with sales leadership and the full sales team. We showed the data: the gap between attach-rate forecast and actuality, the churn delta, and the resulting revenue-at-risk number. We did not propose a new plan. We said: “The current plan isn’t working, the data is clear, and we want your help figuring out what to do.”
This is the part that feels indulgent and isn’t. Reps know the plan isn’t working before you do — they’re the ones working it. Naming the problem out loud and asking for their thinking turns the conversation from “leadership is changing my pay” into “we’re fixing something together.”
Step 2: Run a working group with three reps.
We picked three reps with different tenure and territory profiles and worked with them for two weeks on plan design. They saw all the modeling, including what each version would have paid each individual rep last year. By the time we had a draft, they had already pre-sold it to the rest of the team in side conversations.
This is the cheapest insurance you can buy on a comp change. The cost is two weeks of three reps’ partial attention. The benefit is that the eventual rollout has internal champions before the announcement.
Step 3: Ship it forward-looking, with grandfathering.
The new plan was: 8% on core product, 15% on add-on (the higher rate is necessary because deal sizes are smaller; we modeled this carefully against gross margin). Critical detail: it applied to new deals starting next quarter, not retroactively, and quota for the affected quarter was held flat — reps weren’t expected to make up the difference for the design lag.
Grandfathering matters. The reason mid-year comp changes burn trust isn’t the change itself — it’s the implication that work already done now pays differently than it was supposed to. Drawing a clean line between “old work, old rules” and “new work, new rules” removes the betrayal feeling.
Step 4: Pre-commit to a sunset date for revisions.
We told the team: this plan is stable through the end of next fiscal year. If the math is wrong again, we will eat the cost rather than change it mid-year. Putting that commitment in writing was uncomfortable. It also turned out to be the most reassuring single thing we said.
What happened
Add-on attach rate moved from 31% to 47% within one quarter, and stabilized at 51% the quarter after. Average deal size went down slightly because the add-on is smaller, but total commission paid out per rep went up because units per deal went up. Nobody quit.
The retention math improved more slowly — that’s a 12–18-month lagged metric — but the early indicators (NPS at 90 days, expansion conversion at month four) all moved in the expected direction.
What I would do differently
Two things.
First: I would have caught this six months earlier by tracking attach rate weekly as a leading indicator instead of reading it quarterly out of the comp report. The data was always available; we just weren’t looking at it.
Second: I would have included a CS rep in the working group. Their context on what makes customers stick was the missing piece in our original plan, and adding it earlier would have saved a year of slow drift.
The general principle
Comp plans don’t fail because of bad design. They fail because the world they were designed for stops existing, and nobody updates the design until it’s breaking visibly. The fix isn’t to design a perfect plan — it’s to build a habit of looking at the plan’s assumptions every six months and asking, “is this still true?”
For us that habit now lives in a recurring quarterly meeting between the VP of Sales, the CFO, and the head of CS. They review three numbers (attach rate, blended ACV, comp-as-percent-of-revenue), and if any of them have moved more than 10% from the plan’s assumptions, the plan goes back into review. So far that hasn’t triggered another rewrite, but knowing the trigger exists keeps everyone honest.