Every team that adopts measurable goals discovers the same uncomfortable law within two quarters: people optimize what you measure, not what you meant.
Write "send 500 outbound emails" and you'll get 500 emails — rushed, untargeted, and quietly damaging your domain reputation. Write "ship the new onboarding flow" and something called the new onboarding flow will ship on the last day of the quarter, bugs and all. Write "improve customer satisfaction" and you'll get a spirited end-of-quarter debate about what "improve" meant.
None of this is dishonesty. It's physics. A key result is a contract between an owner and the business, and like any contract, it gets performed as written — not as intended. The skill of writing key results is the skill of writing contracts with no loopholes.
Here's how to do it, with 20 worked examples. (Everything below applies whether you call them key results, success measures, or targets — the mechanics don't care about the framework.)
The anatomy of an ungameable key result
A key result that can't be gamed has five properties. Miss any one and you've left a loophole.
1.It measures an outcome, not an activity. Activities are inputs you control; outcomes are results the business feels. "Make 50 sales calls" is an activity — you can hit it while the business goes nowhere. "Close $60K in new MRR" is an outcome. The test: could someone hit this number while the thing you actually care about gets worse? If yes, rewrite.
2.It has a number and a denominator. Raw counts are the most gameable format in existence, because the easiest way to grow a numerator is to inflate what feeds it. "100 demo requests" is trivially gamed by loosening the definition of a demo request. "Demo requests that convert to opportunities at ≥30%" polices itself.
3.It has a quality bar or a counterweight. Any single metric pushed hard enough deforms something next to it. Speed deforms quality, volume deforms conversion, growth deforms retention. The fix is pairing: the primary number plus the thing it must not break. "Resolve tickets in under 4 hours while CSAT holds at 92%+" can't be hit by firing off garbage replies.
4.It's callable every week. A key result you can only evaluate at quarter-end isn't a key result; it's a verdict. If the owner can't stand up in a weekly review and say "here's the number, we're on track or we're not," the KR is either unmeasurable or measuring something too lagging to steer by. Add a leading proxy you can read weekly.
5.It has exactly one owner. Shared KRs are pre-gamed: when the number misses, it's nobody's miss. One name per number — the person who says it out loud each week.
20 examples: gameable → ungameable
Each pair shows the loophole, then the rewrite that closes it. Numbers are illustrative — from our running fictional company, Acme Co. (~50 people) — so calibrate to your scale.
Sales
1.Activity dressed as outcome
GameableMake 200 outbound calls per month
UngameableBook 24 qualified first meetings, with ≥30% advancing to a second call
The loophole: calls are free to fake. The rewrite measures what calls are for, and the advance rate stops "qualified" from quietly meaning "answered the phone."
2.The sandbagged target
GameableClose $40K new MRR (when last quarter closed $45K with no changes)
UngameableClose $60K new MRR, with no single deal counting for more than 40%
The loophole: a target below trend is a vacation with extra steps. The concentration cap closes the second loophole — one whale masking a broken pipeline.
3.Pipeline theater
GameableAdd $500K to pipeline
UngameableAdd $300K of pipeline that reaches stage 2 within 30 days
The loophole: pipeline entries are free. Stage progression within a time window makes only real pipeline count.
4.The end-of-quarter miracle
GameableHit 120% of quota
UngameableHit 100% of quota with ≤25% of it closing in the final two weeks
The loophole: quarter-end discount-fueled cramming that steals from next quarter. The pacing constraint rewards a healthy motion, not a heroic finish.
Marketing
5.Vanity volume
GameableGenerate 1,000 leads
UngameableGenerate 150 leads that sales accepts, at ≤$200 cost per accepted lead
The loophole: "lead" is infinitely inflatable. Sales acceptance makes another team the referee; the cost cap stops buying junk volume.
6.Traffic without destination
GameableDouble website traffic
UngameableGrow organic signups from site traffic from 80 to 160 per month
The loophole: traffic can be bought, borrowed, or botted. Signups from organic can't.
7.The unmeasurable brand goal
GameableIncrease brand awareness
UngameableGrow branded search volume from 900 to 1,500 monthly, and direct traffic 25%
The loophole: "awareness" has no scoreboard, so it's declared achieved in the retro. Branded search and direct traffic are the closest honest proxies.
8.Content as tonnage
GameablePublish 24 blog posts
UngameablePublish content generating 400 organic template downloads with ≥35% email opt-in
The loophole: posts are output; nobody buys output. Downloads and opt-ins are what the posts were for.
Product & Engineering
9.The binary ship-it
GameableLaunch the new onboarding flow
UngameableNew onboarding live to 100% of signups, completion rate ≥50%, time-to-first-value ≤3 days
The loophole: "launch" is achievable at 11:59pm on the last day, broken. Adoption and outcome metrics define what launched means.
10.Velocity worship
GameableIncrease sprint velocity 20%
UngameableCut median cycle time from idea-approved to in-production from 21 days to 14, with change-failure rate ≤10%
The loophole: velocity is self-reported in points the team itself sizes — inflation takes one planning meeting. Cycle time is a clock; the failure-rate pair stops shipping fast garbage.
11.Bug-count games
GameableClose 100 bugs
UngameableCut open P1/P2 bugs from 40 to ≤10 and hold new-P1 weekly inflow under 3
The loophole: closing 100 trivial bugs while the crashers age. Severity weighting plus an inflow cap measures the actual state of quality.
12.The refactor with no receipts
GameableImprove platform reliability
Ungameable99.9% uptime on the core path and p95 page load under 800ms, measured weekly
The loophole: "improve" invites a quarter of busy work with a subjective grade. Two numbers with a weekly read make reliability a fact, not a feeling.
Customer Success & Support
13.Speed that torches quality
GameableCut first-response time to under 1 hour
UngameableFirst response under 2 hours with CSAT ≥92% and reopen rate ≤8%
The loophole: auto-replies and rushed answers hit any speed target. The counterweights make speed count only when the answer actually helped.
14.Churn defined charitably
GameableReduce churn
UngameableGross revenue churn ≤2% monthly, measured on all accounts active at quarter start
The loophole: undefined churn gets defined at quarter-end by whoever missed it (logo vs revenue, excluding "bad-fit" accounts). Fix the formula and cohort now.
15.Adoption theater
GameableRun 30 customer training sessions
UngameableRaise weekly active usage in the bottom-quartile accounts from 20% to 45%
The loophole: sessions can be held to empty rooms. Usage in the at-risk segment is the thing training exists to move.
16.The NPS survey lottery
GameableRaise NPS to 50
UngameableNPS ≥45 with ≥30% response rate across all active accounts
The loophole: survey only your happy users and NPS soars. The response-rate floor and full-base requirement kill selective sampling.
Operations & Finance
17.Hiring as a race
GameableHire 6 engineers
UngameableHire 5 engineers who pass their 90-day review, with offer-accept ≥75%
The loophole: warm bodies by the deadline. The 90-day gate measures hiring well; accept rate measures whether the offer and pitch are actually competitive.
18.Cost-cutting that eats the future
GameableCut operating costs 15%
UngameableCut opex 10% with zero cuts to the two customer-facing SLAs and no deferred security patches
The loophole: the fastest cuts are the ones whose damage lands next year. Naming what's protected forces the cut into genuine waste.
19.The report nobody uses
GameableBuild a company KPI dashboard
UngameableShip the dashboard and have it opened by 8 of 10 team leads weekly for the final 6 weeks
The loophole: internal tools "ship" and die. Usage by the intended audience is the only definition of done that matters.
20.Runway roulette
GameableExtend runway
UngameableHold monthly net burn ≤$180K while maintaining the committed hiring plan
The loophole: anyone can extend runway by freezing the company. The pairing forces efficiency, not paralysis.
The three-question audit
1. Could we hit this while the business gets worse? → add the outcome or the counterweight.
2. Can the owner call it in a weekly review — on track, at risk, off track — with a real number? → if not, add a leading indicator or admit it's unmeasurable and rewrite.
3. Whose single name is on it, and did they set or negotiate the target? → imposed targets get sandbagged next quarter in self-defense; negotiated targets get owned.
Notice what these have in common: every one is enforced by cadence, not by cleverness. A perfectly written KR that gets looked at twice a quarter will still drift and still get gamed at the margins — because the gap between the number and reality only gets inspected when someone reads the number out loud. That's why the format above insists on weekly callability: the weekly review is the immune system. The wording closes the loopholes; the cadence catches whatever wording missed.
And if the problem is upstream — you don't yet know which numbers deserve targets — the scorecard starter pack has department-by-department metric libraries with the pairing traps pre-closed.
One honest caveat: you can over-engineer this. A KR with four counterweights and three caveats is a legal document, not a goal. Use the minimum constraints that close the actual loopholes your team would exploit — usually one counterweight is enough — and let the weekly conversation handle the rest. Goodhart's law is undefeated; the aim isn't a perfect metric, it's a metric whose failure modes you'll see within seven days.