OKR Tracking for Product Teams: Build a System Beyond Spreadsheets
Spreadsheet OKRs go stale by week three. Here is how to connect each key result to the tool that already holds the number, run a 20-minute weekly check-in, and score the quarter honestly.
Share
OKR tracking is the weekly habit of checking each key result against live data and deciding what to change, not a quarterly ritual of typing numbers into a spreadsheet. An automated OKR system pulls each key result straight from the tool that already holds it, such as product analytics, the issue tracker or billing, so the check-in starts from current numbers and spends its time on decisions.
Key takeaways
Google grades key results from 0.0 to 1.0 and calls 0.6 to 0.7 the sweet spot for ambitious goals (Google re:Work). A tracking system that only shows "done / not done" can't support that.
Label every OKR as committed (expected score 1.0) or aspirational (expected average about 0.7). Mixing them up is the first trap in Google's internal playbook (What Matters).
Only 20% of product teams in Teresa Torres's benchmark describe themselves as outcome-focused; about 30% still work mainly on outputs (Product Talk, 2024). Automating the tracking of output OKRs just makes a to-do list faster.
Keep OKRs out of pay reviews. Both What Matters and Christina Wodtke warn that linking them to bonuses makes people set goals they know they can hit.
What is OKR tracking, and why do spreadsheets fail at it?
OKRs (objectives and key results) pair a qualitative objective with two to five measurable key results. Tracking is everything that happens between setting them and grading them: updating values, judging whether you are on course and changing plans when you are not.
Spreadsheets fail at the middle part for three reasons:
Someone has to copy the number in. If the key result lives in Amplitude, Mixpanel or PostHog and the OKR lives in a sheet, a person pastes it in every week. When that person is busy, the sheet goes stale.
There is no definition attached. "Activation rate" in the sheet may not match "activation rate" in the dashboard. Without a linked query, two people argue about the number instead of what to do about it.
History disappears. A cell holds the latest value, so you lose the trend line: is the key result moving, flat or going backwards?
The result: OKRs written in week 1, ignored until week 11, scored from memory.
What should a product OKR look like before you automate it?
Automation multiplies whatever you feed it. Fix the OKRs first.
Outcome key results, not output key results
An output is something you ship; an outcome is a change in what customers do. Torres separates business outcomes (revenue, cost) from product outcomes, which measure a customer behaviour or sentiment inside the product (Product Talk, 2024). Product teams should own product outcomes, because they can move them within a quarter.
Output key result
Outcome key result
Example
Ship the new onboarding checklist by 15 March
Week-1 activation rises from 22% to 30%
Data source
Jira or Linear status
Product analytics
Can be automated?
Yes, but it only proves the work happened
Yes, and it shows whether the work mattered
What a miss teaches you
Delivery was late
The bet was wrong, or too small
If a key result reads like a ticket, it belongs in your delivery tool. Our guide to product OKRs with examples has rewrites for the most common output-shaped goals.
How many objectives and key results?
Google's guidance is three to five objectives with around three key results each (Google re:Work). Christina Wodtke, author of Radical Focus, is stricter: one company OKR unless you run several business lines, with key results "most often three" (O'Reilly, 2016). For a product team of 3 to 10 people, one objective with three key results per quarter is usually enough.
Committed or aspirational?
Google's internal OKR playbook splits OKRs into two types (What Matters):
Committed: the team agrees to deliver it and rearranges schedules and resources to do so. Expected score: 1.0. Anything less needs an explanation.
Aspirational: a stretch towards where you want to be, with an unclear path. Expected average score: 0.7, with high variance.
Your tracker needs a field for this: 0.7 on an aspirational key result is a good quarter; on a committed one it is a miss.
How do you build an automated OKR tracking system?
Budget an afternoon for the first set-up and 20 minutes a week after that.
Write each key result as metric, baseline, target and date. "Week-1 activation: 22% → 30% by 30 June." No baseline, no automation: you can't chart progress from an unknown starting point.
Name the system of record for each key result. One source per metric, written next to the key result. If two tools disagree, pick one and note why.
Save the query, not just the number, so anyone can rerun it and get the same answer.
Connect the source to wherever the team looks: an OKR tool, a product system or a sheet fed by an API. The value must refresh without anyone copying it.
Refresh daily, not hourly. Hourly numbers invite reactions to noise.
Add the human fields the data can't fill: confidence (0-10), owner, the current bet and what changed since last week.
Review weekly and grade quarterly. The weekly check-in below uses these fields; the quarterly grade uses the final values.
Map your stack: which tool holds which key result?
Key result type
Example
Typical system of record
Refresh
Product outcome
Week-1 activation 22% → 30%
Amplitude, Mixpanel, PostHog
Daily
Engagement or retention
Weekly active teams 140 → 200
Product analytics or data warehouse
Weekly
Revenue
Expansion MRR +€8k
Stripe, Chargebee, Salesforce
Weekly
Customer sentiment
Onboarding CSAT 3.9 → 4.3
Survey tool, support desk
Weekly
Delivery (committed only)
Payments migration live by 30 June
Jira, Linear
Daily
Delivery data belongs here only for committed OKRs with a real deadline.
How do you run a weekly OKR check-in?
If you aren't reviewing OKRs weekly, you are auditing them, not managing them. Wodtke's version is a Monday meeting to check progress, confidence and priorities, and a Friday session where teams show what they shipped (O'Reilly, 2016). The weekly pulse below fits in 20 minutes because the numbers are already on screen.
The weekly pulse check (copy this)
Metric review: What is the current value of each key result? Pulled automatically, never typed into the meeting.
Trend analysis: Are we on track for 0.7 (aspirational) or 1.0 (committed)? Compare this week with the straight line from baseline to target.
The "why" factor: If a metric isn't moving, what hypothesis can we test this week to change the trajectory?
Resource blocker: What is stopping us from focusing on this objective?
Then update one human number: confidence, from 0 to 10, that you will hit each key result. Wodtke recommends setting targets at a confidence of about 5 out of 10, a coin flip, so a stretch goal starts at 5 and moves from there (O'Reilly, 2016). A confidence that falls two weeks in a row is the earliest warning you will get, usually before the metric itself drops.
Paste the result where the team already reads updates:
OKR pulse, week 6 of 13
O: New teams get value in their first week
KR1 Week-1 activation: 22% → 30% now 25.5% (+1.0 wk) confidence 6/10
KR2 Time to first report: 3 days → 1 day now 1.8 days confidence 7/10
KR3 Second user invited by day 14: 15% → 25% now 15.5% confidence 3/10 ↓
Bet this week: invite prompt after first report (KR3)
Blocker: analytics event for "invite sent" fires twice; fix before Thursday
When a key result stalls, ask which customer problem you misread, not which ticket was late. For the wider 30-minute ritual this check-in sits inside, see the PM weekly review.
How do you score OKRs at the end of the quarter?
Google grades each key result from 0.0 to 1.0, averages them into the objective's score and expects an average of 0.6 to 0.7 across all OKRs (Google re:Work). Grade about two weeks before the quarter ends, as Wodtke suggests, so the scores inform next quarter's planning rather than arriving after it (O'Reilly, 2016).
With automated tracking, scoring a metric key result is arithmetic: (current − baseline) ÷ (target − baseline). The judgement is in reading the score:
Score
Aspirational OKR
Committed OKR
1.0
Target was probably too easy; raise it
Delivered as promised
0.6-0.7
The sweet spot
A miss; write a short post-mortem
0.3-0.5
Learned something; was the bet wrong or the target?
Serious miss; check how it was planned
Below 0.3
Drop, or rewrite with a new bet
Escalate; the plan or resourcing failed
Google also makes OKRs public so everyone can see what other teams are working on (Google re:Work). A live, read-only view does that without a status meeting.
Worked example: a 5-person team tracks one objective (illustrative)
Say a 5-person B2B SaaS team sets one aspirational objective for a 13-week quarter: "New teams get value in their first week."
KR1: Week-1 activation from 22% to 30% (system of record: PostHog saved insight).
KR2: Median time to first report from 3 days to 1 day (warehouse query, refreshed daily).
KR3: Teams inviting a second user by day 14 from 15% to 25% (PostHog funnel).
At week 6, the tracker shows KR1 at 25.5%, KR2 at 1.8 days and KR3 at 15.5%, against a straight-line pace of 19.6% for KR3. Confidence drops from 5 to 3, so the check-in focuses on KR3. Two customer calls show people don't notice the invite option until they export a report, so the team moves the invite prompt to the moment the first report finishes.
At grading in week 11, KR1 is at 28.4% (score 0.8), KR2 at 1.4 days (0.8) and KR3 at 19% (0.4). The objective averages 0.67, inside Google's 0.6-0.7 band. Without weekly tracking, the KR3 problem would have surfaced at week 11, too late to fix.
Common mistakes with OKR tracking
Tracking business as usual. Google's playbook names "business-as-usual OKRs" as a trap: goals written around what the team already expects to do (What Matters). Keep uptime, bug counts and support response times on a health dashboard, and reserve OKRs for change.
Tying OKRs to pay. What Matters, the team behind John Doerr's book, says not to connect OKRs and compensation, because people aim lower to protect their bonus (What Matters, 2026). Wodtke warns the same.
Dictating every OKR from the top. What Matters says healthy organisations aim for half of their goals to come from the bottom up (What Matters). Leadership sets the objective; the team proposes the key results it believes will show progress.
Individual OKRs inside a product team. Marty Cagan argues that when engineers, designers and PMs each carry their own objectives, they pull in different directions; a product team should share one set of team objectives (SVPG, 2020).
Automating a vanity metric. If a customer could be better off while the number stays flat, pick a different key result. Our guide to picking a North Star metric uses the same test.
Which OKR tracking tool should a product team use?
The right option depends on team size more than features.
Option
Good for
Watch out for
Spreadsheet fed by an API or connector
Teams of 1-5 with one objective
Breaks when the person who built it leaves; no confidence history unless you add it
Dedicated OKR software (for example Quantive, Perdoo, Lattice Goals)
Companies with many teams cascading OKRs
Another tool to keep in sync; product evidence still lives elsewhere
A product system that holds OKRs next to roadmap, specs and customer evidence
Product teams that want each key result tied to the bets meant to move it
Less suited to HR-led, company-wide goal cascades
Keep your OKR data portable, because tools disappear. Microsoft stopped building Viva Goals on 5 December 2024 and retired it on 31 December 2025, leaving customers to export via API, Excel or PowerPoint (Microsoft Learn).
When OKRs and roadmap disagree, the roadmap is usually wrong. An outcome-based roadmap should carry the same numbers as your key results, so a single update serves both.
Where Specky fits
Specky holds OKRs in the same Product Graph as customer calls, feedback, specs and tickets. Each key result can link to the features and specs meant to move it, and after launch Specky checks whether the bet moved the number, so the "why" behind a stalled key result is one click from the key result itself. It pairs with analytics rather than replacing it; we explain the difference in product intelligence vs product analytics.
Frequently asked questions
How often should you track OKRs?
Check key results weekly and grade them once a quarter. A weekly check-in is frequent enough to change a bet mid-quarter and slow enough to ignore daily noise. Christina Wodtke recommends a Monday check-in on progress and confidence, plus a Friday session to show what shipped. Grade about two weeks before the quarter ends so the scores shape the next quarter's plan.
What is a good OKR score?
For aspirational OKRs, 0.6 to 0.7 on Google's 0.0-1.0 scale is the target. Scoring 1.0 every quarter suggests goals that are too easy. Committed OKRs are different: Google's playbook expects them to score 1.0, and anything lower needs an explanation. Always label which type an OKR is before the quarter starts, or the score can't be read.
Can you automate OKR tracking completely?
You can automate the numbers, not the judgement. Values for outcome key results can flow from product analytics, billing or a warehouse without anyone copying them. Confidence, the reason a metric stalled and the next bet still need a person, or you get green dashboards and no decisions.
Should OKRs be tied to bonuses?
No. What Matters, the organisation behind John Doerr's Measure What Matters, advises against connecting OKRs and compensation, because people sandbag goals to protect their pay. Google's own guide says OKRs are not the same as employee evaluations. Reward team outcomes and how people work, and keep OKRs as a planning and learning tool.
Are OKRs a good fit for small product teams?
Yes, if you keep them small: one objective and three key results per quarter for a team of 3 to 10. Small teams gain most from automated tracking because nobody has time to maintain a spreadsheet. Skip cascading OKRs down to individuals; at this size it adds overhead, not focus.
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