The key result that took a meeting to grade
Quarter end. The key result says “improve the onboarding experience.” One manager points to the new welcome emails and calls it a 0.8. Another points to cancellations, which did not move, and calls it a 0.3. Forty minutes later the team settles on 0.6, which means nothing to anyone. An OKR is done when two people would grade it the same way, and when hitting every key result would make the objective true.
Andy Grove, in a 1970s teaching clip that John Doerr plays in his TED talk, set the standard that meeting failed: key results “have to be measured, but at the end you can look and without any argument say, ‘Did I do that, or did I not do that?’” 2. Google's published OKR guidance turns the same idea into a rule: “It must be possible to objectively assign a grade at the end of the quarter” 1.
The writer below takes one objective at a time. You enter the direction, the key results, what will change and what must not get worse, and it flags the classic faults as you type: activities posing as results, key results with no number, goals that ask for no change, key results that could all hit while the objective misses. You leave with a one-page OKR you can paste into whatever template your company uses.
If grading it takes a meeting, it was never a key result.
OKR writer
Fill in one objective at a time. The checks run as you type and only warn; you may have a reason to keep a line they flag.
In Google's published guidance, a committed OKR is expected to score 1.0 and an aspirational one about 0.7 on average. Decide before you write, because it changes how a miss is read.
The direction, in plain words. One line, no metric needed.
Name the customer, user or economic benefit. If no one outside the team would notice, the objective is low value.
Each one a result someone could grade without a meeting: a number, a baseline if you have it, and a date.
The change test. If the honest answer is nothing, this is business as usual.
The measure that exposes the cheapest way to hit the key results.
Each team named here should carry its part in its own OKRs.
What the checks look for
Every check maps to a fault that shows up in real OKR sets. Where Google's guidance names the fault, the check follows it; the thresholds (15 words, five key results, 30 words) are our rules of thumb, not a published standard.
| Check | What it catches | Basis |
|---|---|---|
| A number in every key result | Key results nobody can grade | Google: key results must be measurable, gradable at quarter end 1 |
| No activity words | “Consult,” “help,” “analyze,” “participate”: work, not outcomes | Google: key results “must describe outcomes, not activities” 1 |
| No bare task verbs | “Launch X” with no criteria for success | Google's “who cares?” test 1 |
| Objective on one line | Objectives that hide three goals | Google: one line, or it probably isn't crisp; 15 words is our cut |
| No “maintain” or “keep” | Business-as-usual goals dressed as OKRs | Google's business-as-usual trap 1 |
| A named change | OKRs the team would hit anyway | Google: no change to activities means business as usual 1 |
| Sufficiency | Key results that are necessary but not enough | Google's insufficient-key-results trap 1 |
| Two to five key results | A single number, or a work plan | Ours |
The guardrail field has no check. It is there because any number pushed hard invites the cheapest way to hit it, the failure our page on KPI examples covers in depth.
One draft, three passes
Notice what moved between drafts. The wording checks fixed the first draft in a minute. The third key result came from the one question a string check cannot answer: what would have to be true for the objective to be true?
Reading your warnings
- No number. Add the baseline, the target and the date. If you cannot find a baseline, the first key result may be to measure one, and the OKR is probably aspirational.
- Activity or task words. Ask what the activity is for and write that instead. Google's example: not “assess Colossus latency” but “publish average and tail latency measurements from six Colossus cells by March 7” 1.
- “Maintain” or no named change. Move the line to your KPI set. Our OKR vs. KPI page explains when to promote a number back.
- Not sufficient. Add the key result that closes the gap, usually adoption, quality or the customer's own behavior.
- More than five. Cut the key results that describe how you will work. Those belong in the plan.
What a wording check cannot see
The fair objection to any OKR checker: it trains people to write OKRs that pass checks. A team can learn to avoid “help” and “launch” and still set targets it knows it will hit. Google's guidance calls that sandbagging, and its test is about resources, not words: committed OKRs should “credibly consume most — but not all” of a team's available resources 1. No text check can see your headcount.
So use the writer for what it does well, which is the mechanical half. Then spend the meeting on the half it cannot do: whether this is the right objective, whether the target is a stretch or a sandbag, and whether a miss would teach you anything. Doerr notes that at Google OKRs “are not used for bonuses or for promotions” 2; if yours feed pay, expect targets to drift toward the safe end, whatever the wording says.
Where to go next
Test your eye on the key result or task quiz, then take the OKR set to the quarterly business review, where a missed committed key result should end in a decision, not a shrug. An OKR you can grade without a meeting frees the meeting for the question that matters: what the grade tells you to change.
Questions
- What makes a good key result?
- A good key result is an outcome, not an activity, and it can be graded at the end of the period without argument: a number, ideally with a baseline, and a date. Together, the key results should be sufficient, meaning that hitting all of them would make the objective true.
- How many key results should an objective have?
- Google's published guidance sets no number. Our rule of thumb is two to five. One number rarely captures an objective and is easy to game; more than five usually means the list has turned into a work plan.
- Should an objective contain a number?
- It can, but it does not need one. The objective sets the direction in plain words and should fit on one line. The key results carry the numbers that show whether you got there.
Sources
- Google (n.d.). Google's OKR Playbook. What Matters (whatmatters.com). Key results must describe outcomes, not activities, with consult, help, analyze and participate as activity words and the Colossus latency example; key results must be measurable and gradable at quarter end; an objective that does not fit on one line; the business-as-usual, “who cares?” (“Launch X”), sandbagging and insufficient-key-results traps; committed and aspirational OKRs and their expected scores; cross-team OKRs.
- Doerr, J. (2018). Why the secret to success is setting the right goals (TED2018 talk transcript). TED. Andy Grove, in a 1970s teaching clip shown in the talk, on key results that can be graded “without any argument”; OKRs at Google “not used for bonuses or for promotions.”
More from Stratrix
“What is it worth paying to close an exposure that has no upper bound?”
One decision per pack: the framework, a working model, the sourced cases and the board material.
See the pack — $499 →Test a decision, free
Set a live decision against the companies that made it before you.
AcquisitionsThe Distance TestDivestituresThe Divestiture TestDiversificationThe Diversification TestMake or buyThe Make-or-Buy TestAll Access
Every Casebook, three Strategy Packs a seat each year and the Analyst, for teams that face these decisions all year.
$1,999 a year · $499 a seatWhat is included →Strategy Decision Cases
879 strategic decisions with dated outcomes, ranked by how often each kind holds.
24 decision typesSee which decisions hold →New analyses, by email
The defining moves in business, checked against the record, as they publish.