Product OKRs: 12 Examples and How to Write Outcome Goals
Most product OKRs are to-do lists in disguise. Here are 12 outcome-based examples, a six-step way to write your own, and the rules that turn OKRs into a tool for saying no.
Share
Product OKRs are quarterly goals that name a change in customer behaviour (the objective) and two to four numbers that prove it happened (the key results). "Launch feature X" is not a product OKR; it is a task. A good one reads like "New teams get value in week one: activation rises from 22% to 30%", and leaves the team free to choose what to build to get there.
Key takeaways
OKRs began with Andy Grove at Intel and reached Google in 1999, when John Doerr presented them to Larry Page and Sergey Brin (Penguin Random House).
Only 20% of product teams in Teresa Torres's benchmark call themselves outcome-focused; about 30% mostly work on outputs (Product Talk, 2024).
Google recommends three to five objectives with around three key results each, and expects an average score of 0.6 to 0.7 (Google re:Work). Small product teams should use fewer.
The most useful thing a product OKR does is tell you what not to build this quarter.
Marty Cagan, who has long written about OKRs for product teams, says they fail in companies that run feature teams and feature roadmaps (SVPG, 2020).
What are product OKRs?
OKR stands for objectives and key results. The objective is qualitative and says where you want to go. The key results are measurable and say how you'll know you got there.
Product OKRs are the version a product team owns. Torres splits outcomes into business outcomes, such as revenue, and product outcomes, which measure a customer behaviour or sentiment in the product (Product Talk, 2024). Product OKRs should mostly carry product outcomes, because a team can move them within a quarter. Revenue lags by months and depends on sales and pricing as well.
Josh Seiden's definition, which Torres cites, is the cleanest test: an outcome is "a change in human behavior that drives business results". If your key result could hit 1.0 while no customer behaves differently, it is an output.
Andy Grove built OKRs at Intel on top of Peter Drucker's 1950s idea of management by objectives (What Matters). John Doerr learned them from Grove in the 1970s while working at Intel. In the fall of 1999, shortly after his firm invested $12.5 million in Google, Doerr presented OKRs to Page and Brin. He later wrote them up in Measure What Matters, published on 24 April 2018 (Penguin Random House).
Why do most product OKRs turn into task lists?
Because writing "ship X" is easier than deciding what success looks like. John Cutler's 2016 list of feature-factory signs describes the result: teams that don't measure the impact of their work, celebrate shipping over results and keep roadmaps of features instead of goals (Cutler, 2016).
Output OKRs cause three problems:
They dictate the how. "Launch the onboarding wizard" removes the team's freedom to find a cheaper fix, so engineers become ticket-takers.
They grade the wrong thing. You can score 1.0 on "ship the wizard" while activation stays flat.
They can't say no. Any feature can be argued into a list of features. An outcome gives you a test: does this request move the number?
Task-list OKRs
Outcome OKRs
Objective
"Deliver the Q2 roadmap"
"New teams get value in their first week"
Key result
"Ship onboarding wizard by 30 May"
"Week-1 activation: 22% → 30%"
Who chooses the solution
Whoever wrote the OKR
The product team, through discovery
Review cadence
End of quarter
Weekly check-in
When the idea fails
Ship it anyway; it's on the list
Swap the bet, keep the key result
12 product OKR examples you can adapt
These are illustrative: the baselines and targets show the shape, not benchmarks. Replace them with your own numbers. "A" means aspirational (expected score around 0.7); "C" means committed (expected 1.0), following Google's playbook (What Matters).
#
Area
Objective
Key results
Type
1
Activation
New teams get value in their first week
Week-1 activation 22% → 30%; median time to first report 3 days → 1 day
A
2
Activation
Self-serve signups set up without help
Setup-related support tickets 120 → 60 per month; setup completion 48% → 65%
A
3
Retention
Teams make us part of their weekly routine
Weekly active teams 140 → 200; week-8 retention 31% → 38%
Data-loss incidents 3 → 0; p95 save latency 900 ms → 300 ms
C
12
Platform
Payments migration done before the provider deadline
Migration live for 100% of accounts by 30 June; 0 failed renewals during cut-over
C
Notice that only examples 11 and 12 are committed, and only 12 carries a delivery date. Cagan calls these high-integrity commitments: binary, tracked separately from key results and meant to be the exception, not the rule (SVPG, 2020).
Example 8 is a process goal. Use one like it only when the habit itself is the problem you are fixing, and drop it once the habit sticks.
How do you write product OKRs? A 6-step method
Start from strategy and one metric that matters. Read your strategy and your North Star metric. The objective should move one input to it.
Write the objective as a customer change, in one line. Google's playbook offers a test: "If your Objective doesn't fit on one line, it probably isn't crisp enough."
Pick 2-4 key results with baseline, target and date. Measure the baseline first if you don't have one. Google suggests around three key results per objective (Google re:Work).
Check the key results are sufficient. Google's playbook warns about key results that are necessary but not sufficient: hitting all of them should mean the objective is met.
Let the team propose the key results. What Matters says healthy organisations aim for half of their goals to come from the bottom up (What Matters). Leadership sets the direction; the team that does the work proposes the numbers.
Label committed or aspirational, then write a "not this quarter" list. Name three tempting things the OKR rules out. That list is what makes OKRs a tool for focus.
An OKR's power lies less in what it makes you do than in what it lets you ignore. If the quarter's objective is "trial-to-paid rises from 9% to 12%", a request for a new dashboard theme fails the test on sight: it doesn't move the number.
Use the OKR in the conversation, not as a wall:
"Does this move one of our three key results?" If yes, it competes on evidence with the current bets.
"If not, does it protect a committed OKR or a health metric?" If yes, it may still be urgent.
"If neither, it goes on the next-quarter list, with the reason written down."
This turns a "no" into a shared rule rather than a personal opinion. For scripts that keep relationships intact, see how product managers say no.
How much capacity should OKRs take?
Not all of it. Google's playbook says committed OKRs should use most, but not all, of a team's resources, and a team that meets every OKR with spare capacity may be sandbagging (What Matters). Our rule of thumb for a small team: plan OKR bets for about two-thirds of capacity and keep the rest for bugs, technical debt and support. Ignoring that maintenance doesn't make you faster; it piles up until it stalls delivery.
How do you know when to kill an OKR bet?
Check key results weekly. If a bet hasn't moved its key result after 3-4 weeks of a 13-week quarter, treat that as evidence, not bad luck. Ask whether the bet was wrong, too small or aimed at the wrong customers, then change it.
The sunk-cost trap is real: a feature that took six weeks to build feels too expensive to drop. But the key result is the commitment, not the feature. Swapping the bet while keeping the outcome is the system working as designed.
A compact loop for each quarter:
Define the outcome: "Increase week-8 retention from 31% to 38%."
Define the constraint: "We have six weeks to test three ideas."
Discover: test assumptions with customers before writing production code. Our assumption mapping guide shows how to pick the riskiest one.
Ship and measure: release the smallest version, read the key result, then persevere, pivot or stop.
Common mistakes with product OKRs
Low-value objectives. Google's playbook gives the example "Increase task CPU utilization by 3 percent": it can score 1.0 without anyone benefiting. Reword around a user or economic benefit.
Individual OKRs inside a product team. Cagan argues that separate objectives for engineers, designers and PMs pull a team apart; they should share one set of team objectives (SVPG, 2020).
Too many OKRs. Five objectives with four key results each is 20 numbers. A team of 6 can't run bets against 20 numbers.
OKRs that duplicate the roadmap. The roadmap and the OKRs should carry the same outcomes. An outcome-based roadmap makes this natural: each Now item maps to a key result.
Tying OKRs to bonuses. Teams set targets they know they can hit. Keep OKRs for planning and learning.
When OKRs don't work
Cagan, who has written about OKRs for product teams for years, says he no longer recommends them to most companies. His reason: teams that are handed feature roadmaps can't use a technique built for empowered teams that are handed problems (SVPG, 2020). If leadership will still dictate the feature list, fix that first. OKRs also add little for a solo founder before product-market fit; one weekly goal and a learning log will do.
Where Specky fits
Specky keeps OKRs in the same Product Graph as customer calls, feedback, specs and tickets, so each key result can link to the bets meant to move it and the evidence behind them. After launch, it checks whether the bet moved the number, which makes the weekly "keep, swap or stop" decision faster.
Frequently asked questions
How many OKRs should a product team have?
One objective with two to four key results is enough for most product teams of 3 to 10 people. Google's guidance allows three to five objectives with around three key results each, but that is for larger groups. More OKRs dilute focus: every key result needs someone running bets against it and checking it weekly.
What is the difference between OKRs and a product roadmap?
OKRs set the target for a period, usually a quarter. The roadmap shows the sequence of problems and bets across several periods. They should share numbers: each item in the roadmap's Now column should map to a key result. If they disagree, one of them is out of date, usually the roadmap.
Who invented OKRs?
Andy Grove developed OKRs at Intel, building on Peter Drucker's management by objectives from the 1950s. John Doerr learned them from Grove in the 1970s and introduced them to Google's founders in 1999. Doerr's 2018 book Measure What Matters made the method widely known outside Silicon Valley.
Should product OKRs ever include features?
Only as committed OKRs with a real external deadline, such as a contract, a regulation or a partner launch. Even then, track them separately from outcome key results and keep them rare. Features normally belong under a key result as bets the team can swap if they don't move the number.
What are good OKRs for a product manager?
A PM shares the team's product OKRs rather than holding separate personal ones. Good team key results for a PM to own are product outcomes such as activation, retention, trial conversion or task success rate. If a PM needs a personal development goal, keep it out of the OKR system and in a one-to-one.
GDPR-aligned. EU workspace data is stored in Frankfurt, Germany. GDPR and EU data Built in Vienna, Austria. A product of influence solutions fm GmbH (FN 675374y). Imprint 14-day trial. Card required, and nothing is charged until day 14. 30-day refund after subscribing. See pricing