Org, people & execution
Objectives and Key Results (OKRs)
A quarterly goal-setting cycle that pairs a handful of ambitious, qualitative objectives with three to five measurable key results each, published across the organisation so teams align, track progress weekly and score honestly at the end of the cycle.
Also known as OKRs, Objectives and key results, iMBO (Intel Management by Objectives). First set out by Andrew S. Grove; popularised by John Doerr in 1983; the primary source is cited in full below.
Where this is contested
Attribution is layered rather than clean: Drucker's management by objectives (1954) supplied the philosophy, Grove built and named the practice at Intel from the 1970s and documented it in 1983, and Doerr carried it to Google in 1999 and codified it in 2018. No single figure originated OKRs as now practised.
- Format
- Process / loop
- Level
- Corporate · Business unit · Product · Team
- Best for
- Plan execution · Prioritise
- Decision stage
- Plan · Execute · Review
- Difficulty
- Intermediate
- Time to apply
- A day or two to draft a first company set; the real cost is the recurring cadence of weekly check-ins and quarterly scoring, and the benefit only compounds after two or three cycles.
Plate · The model
The components
Set objectives and key results
The drafting stage: a few significant, qualitative objectives per level, each carried by three to five measurable key results that describe outcomes rather than activity.
Signals of strength
Three to five objectives at each level, no more · Every key result has a number, a baseline and a named owner · Key results describe outcomes rather than tasks · At least some of the targets feel uncomfortable
Align across teams
The negotiation stage: OKRs are published, teams write their own in response to the top-level set, and cross-team dependencies are agreed before the quarter starts rather than discovered inside it.
Signals of strength
Anyone in the organisation can read any team's OKRs · A meaningful share of team OKRs originates bottom-up · Dependencies were negotiated between teams directly · No shadow goal system runs alongside the official one
Execute and track
The in-quarter stage: OKRs live in the weekly rhythm of the team, with confidence ratings that move as evidence arrives and failing key results triggering action mid-cycle.
Signals of strength
OKRs appear in weekly team meetings rather than only at quarterly reviews · Confidence ratings change during the quarter · A slipping key result produces a decision, either intervention or explicit de-prioritisation
Score and reset
The closing stage: each key result is scored, the misses are read for information, and the next cycle starts with the lessons priced in.
Signals of strength
Scores are published, including the failures · Average scores on aspirational OKRs sit near 0.7 rather than 1.0 · Some objectives are killed rather than rolled over by default · Scoring takes hours, not weeks
When it earns its keep
- The organisation is scaling and informal alignment has broken down: teams work hard on things that no longer add up to a coherent whole.
- Strategy exists on paper but never survives contact with the quarter, and you need a mechanism that converts it into a small number of measurable commitments.
- You need to concentrate effort. OKRs force explicit choices about what will not be pursued this cycle, which is most of their value.
- Cross-functional dependencies keep surfacing at the deadline. Published OKRs let teams see what others are committed to before the collision.
And when it doesn't
- The organisation cannot yet measure the things that matter. Key results without trustworthy data become theatre; fix measurement first.
- You intend to wire the scores into pay and bonuses. Doerr is explicit that separating OKRs from compensation is what keeps the stretch honest; tie them to pay and people will sandbag.
- The work is genuinely steady-state operations with stable service levels. OKRs are for change and stretch; routine performance belongs in ordinary metrics.
- Leadership wants a one-off planning exercise. OKRs only pay back as a repeating cadence with in-quarter tracking; a single quarter proves nothing.
How to run it
Before starting, gather the inputs the analysis depends on:
- A clear strategy or annual direction from which quarterly objectives can be derived, or the exercise becomes goal-setting in a vacuum.
- Baseline data for anything that will serve as a key result, with a named owner for each metric.
- A cadence commitment: time for quarterly setting, weekly or fortnightly check-ins, and an end-of-cycle scoring session.
- Leadership willing to publish their own OKRs first and to be scored in public alongside everyone else.
- 1
Set a small number of objectives
Choose three to five objectives per level, each significant, qualitative and time-bound. An objective is a direction worth fighting for this quarter, expressed in language a new joiner would understand. Ten objectives is a list of everything the team already does, and a sign the choosing has not happened.
- 2
Attach measurable key results
Give each objective three to five key results that describe outcomes with numbers attached. Grove's test still applies: at the end of the period you should be able to say, without argument, whether each one was achieved. 'Improve the returns experience' is a hope; 'cut refund processing from nine days to three' is a key result.
- 3
Align rather than cascade
Publish the top-level set, then let teams write their own OKRs in response, negotiating dependencies directly with each other. Doerr's account of Google has roughly half of OKRs originating bottom-up. Mechanical cascading, where each layer inherits fragments of the layer above, produces compliance without ownership and takes weeks the quarter does not have.
- 4
Track in the open
Review OKRs in the ordinary weekly rhythm of the team, with a simple confidence rating against each key result. The point of in-quarter tracking is to trigger action while there is still time to act; a key result discovered to be failing in week thirteen was actually failing in week five.
- 5
Score and grade honestly
At cycle end, score each key result, typically on a 0.0 to 1.0 scale, and distinguish committed OKRs, where the expectation is delivery in full, from aspirational ones, where around 0.7 counts as success. The score matters less than the sentence that explains it.
- 6
Reset and learn
Run a short retrospective before the next setting session. Some objectives carry over deliberately, some are killed, and the ones that scored 1.0 with no effort get harder. The loop, run repeatedly, is the framework; a single pass is just a to-do list with numbers.
Reading the result
A published, measurable statement of what the organisation is trying to change this cycle, a live view of progress against it, and an end-of-cycle scorecard whose misses carry as much information as its hits.
- A perfect scorecard is a warning, not a triumph. Grove treated consistent full marks as evidence the targets were set too low, and the convention that 0.7 on a stretch key result is success exists to protect ambition.
- Read the misses for information. A 0.3 with a clear account of why is worth more than a 0.9 nobody can explain, because it tells you something about the business you did not know in week one.
- Look at where OKRs originated. If everything flowed top-down, the alignment stage did not happen and next quarter's ownership problem is already visible.
A worked example
A homeware e-commerce scale-up uses OKRs to fix its post-purchase experience
A UK online homeware brand has grown from £8m to £25m revenue in three years and now employs around 90 people. Growth has been carried by paid acquisition, but repeat purchase rates have stalled and delivery complaints are climbing. The leadership team adopts a quarterly OKR cycle, with a first company objective aimed squarely at the post-purchase experience.
- Set objectives and key results
- The company objective is 'Make delivery and returns a reason to recommend us'. Key results: cut average dispatch-to-delivery time from 5.1 to 3.5 days; raise post-delivery NPS from +12 to +30; reduce damage-in-transit refunds from 4.0% to 1.5% of orders. A first draft included 'launch the new returns portal' as a key result; it was rewritten as the outcome it was meant to produce, refund processing time down from nine days to three.
- Align across teams
- Warehouse, customer service and web teams wrote their own supporting OKRs against the company set. The warehouse took dispatch time; the web team took returns; customer service, asked to inherit the NPS number wholesale, pushed back that it could not own a metric driven mainly by couriers and packaging, and instead committed to first-response and resolution times. That argument, had it not happened in week zero, would have happened in week twelve as blame.
- Execute and track
- OKRs were reviewed in Monday leads' meetings with red-amber-green confidence. By week five the damage-refund key result was red and holding. Investigation traced most breakage to one flat-pack range and its supplier's packaging, which produced a mid-quarter decision to re-specify cartons rather than wait for the scorecard to record the failure.
- Score and reset
- Quarter end: dispatch time 0.9, damage refunds 0.7 after the packaging change, NPS 0.4. The NPS miss was the informative one: analysis showed the score moved with courier reliability in two regions, which the warehouse could not fix from inside. The next cycle's objective became re-tendering the courier contract, something nobody had proposed in the original planning.
The read. The cycle delivered two of three key results, but its real product was the NPS miss, which redirected the next quarter toward the courier contract rather than more warehouse optimisation. The scores forced the arguments, the arguments produced the learning, and the learning set the next objectives. That loop, rather than any single quarter's numbers, is what the company bought by adopting OKRs.
Pitfalls
- Writing tasks as key results. 'Launch the portal' is activity; the key result is whatever the portal is supposed to change, with a number on it.
- Covering everything the team does. OKRs describe the change you are trying to make this cycle; business as usual belongs in ordinary operational metrics, and mixing the two buries the stretch under the routine.
- Cascading mechanically down the hierarchy so each level inherits fragments of the level above. It feels rigorous, takes weeks, and produces compliance without ownership.
- Linking scores to compensation. The moment a 0.7 costs someone money, every target in the building quietly shrinks to what is already achievable.
- Set-and-forget: drafting OKRs in week one and rediscovering them in week thirteen. Without the weekly tracking rhythm the framework is a quarterly essay-writing exercise.
What the critics say
The goal-setting research behind frameworks like OKRs has documented systematic side effects: narrowed focus that neglects everything off-goal, increased unethical behaviour, distorted risk appetite and eroded intrinsic motivation. Ordóñez and colleagues argue goal setting should be treated as prescription-strength medication rather than an over-the-counter remedy, a warning that applies directly to aggressive key-result regimes.
Ordóñez, L. D., Schweitzer, M. E., Galinsky, A. D. and Bazerman, M. H. (2009) 'Goals Gone Wild: The Systematic Side Effects of Overprescribing Goal Setting', Academy of Management Perspectives, 23(1), pp. 6-16.
Deming argued that management by objectives, the tradition OKRs descend from, should be eliminated altogether: most performance variation is systemic, so a numerical target without a method changes nothing except behaviour around the measurement. The critique lands hardest on OKR programmes that set numbers without changing the system that produces them.
Deming, W. E. (1986) Out of the Crisis. Cambridge, MA: MIT Press.
Direct evidence that OKRs outperform other goal systems is thin. The method's reputation rests heavily on its association with Intel and Google, firms whose success has many causes, and survivorship bias does much of the selling; well-documented adoption failures rarely make the keynote circuit.
Sources and further reading
- Grove, A. S. (1983) High Output Management. New York: Random House.
- Doerr, J. (2018) Measure What Matters: How Google, Bono, and the Gates Foundation Rock the World with OKRs. New York: Portfolio/Penguin. ↗
- Drucker, P. F. (1954) The Practice of Management. New York: Harper & Row.
- What Matters: 'OKRs History: Where Do OKRs Come From?' (the Grove and Intel origin story). ↗