I don't love traditional pay-for-performance.
It's the dominant playbook in most companies. Managers grade individuals against a rubric, ratings get calibrated across the org, and merit increases land on top of base salary. It looks fair, it feels objective, and it lets HR point at a process whenever someone's unhappy with their number.
In a startup, it often works against you.
Here are the three reasons I've stopped reaching for it.
1. Individual awards miss the invisible work, and they inadvertently reward competition.
The best teams I've worked with are deeply collaborative. They unblock the person next to them, they share context, they pick each other up when something falls over.
None of that shows up cleanly on an individual scorecard.
The moment you tie pay to individually-attributed outcomes, you start incentivising the wrong behaviours. You over-reward the loudest person in the standup and the one who knows how to optimise for visibility over impact. You often underpay the glue people. And the glue people are the ones holding your culture together.
2. Once it's compounded into base salary, you can't unwind it.
This is the part the playbook can fall apart on.
Someone has a stellar year, you give them a 12% merit increase. The next year they're solid but not exceptional (because there's natural mean-reversion in performance, like everything else), so you give them 5%. A year on, the role has changed, the bar has moved, or maybe they've checked out. Now you're paying top-of-band compensation for middle-of-band output.
There's no graceful way back. Base salary compounds, performance doesn't, and the system you built to reward performance has now locked in a cost you can't trim.
3. The overhead is mismatched to startup life.
Building criteria, training managers on calibration, running review cycles, defending ratings, documenting outcomes. It's a serious operational investment.
It only pays off if the goals you're measuring against stay roughly stable. Startups aren't stable. We're clamouring for product-market fit, the strategy shifts on the data, teams reshape every quarter, and the goal posts are forever on the move. By the time you've calibrated H1, the strategy has already moved on. You're auditing performance against a target that no longer exists.
That's a lot of overhead for outcomes that age out before the cycle closes.
So what do I do instead.
Three alternatives I love, and use.
1. Pay performance as bonuses, not merit increases.
If you want to reward great performance, do it with a bonus.
Cash in hand. Tied to the period it's recognising, and gone after that.
It still rewards the work, and it still motivates the right behaviour. What it doesn't do is compound. You don't have to live with last year's decision forever, and when the performance bar moves, your cost base doesn't have to move with it. That alone is worth a rethink.
2. Decide your entire workforce is high-performing, and pay them like it.
This one takes nerve.
Set your base salary bands at a higher percentile across the board (say 75th rather than 50th), and only hire people you believe deserve to be paid at that level.
You replace the entire "who deserves more this year" conversation with a much sharper one: "do we still want this person on the team?"
It simplifies comp. It raises the bar at the door. It removes most of the political theatre around comp reviews, and it sends a strong signal about the kind of place you're building. The trade-off is real. If someone stops performing, you have to have the hard conversation rather than letting them drift quietly into a lower band on the org chart. I'd argue that's exactly the kind of trade-off a startup needs.
3. Share the upside through profit or revenue share.
If you want everyone to feel ownership of the outcome, build a mechanism where they actually share in the outcome.
A pool that grows with the business and gets distributed when the business wins. Everyone rowing in the same direction, no one carving out individual recognition from a finite pie.
One thing I'd push hard on here. Don't distribute it as a percentage of salary. That re-creates the compounding problem from earlier, and it means your highest earners get the biggest absolute share, which often isn't reflective of where the value was actually created.
Normalise the payout against role level. The scope of the seat is what's earning the share, and the reward should match the scope of the seat. Paycheck size is a separate story.
A senior IC and a senior manager at the same level have both moved the business. Pay them accordingly.
Pay-for-performance isn't a bad idea. It's just usually implemented in a way that's expensive, divisive, and impossible to unwind. In a startup, you can do better with fewer moving parts.
Pay people well at the door, reward great periods with money you can take back, share the upside when the business wins, and stop pretending a calibration spreadsheet can hold the line on performance.
None of it compounds, and none of it locks you in. That's the whole system.