OKRs for innovation teams: what to do and what to avoid

OKRs (objectives and key results) work for innovation teams when the key results measure what the team is learning and changing, not how much activity it produces. An objective states a direction in plain words; two to four key results show, with numbers, whether you got there. For exploratory work, the best key results are about evidence: assumptions tested, behaviour observed, decisions reached. Below are five things to do and five to avoid.

A short background

OKRs grew out of Andy Grove's management practice at Intel and were brought to Google by John Doerr in 1999; Doerr's book Measure What Matters (2018) describes their history and use. Google's public re:Work guidance distinguishes two kinds of OKR. Committed OKRs are expected to be fully achieved. Aspirational OKRs are deliberately ambitious, and the guidance suggests that consistently scoring around 0.7 on them means the bar is set at about the right height.

That distinction matters more for innovation teams than for anyone else, because much of their work is uncertain by design.

For how OKRs fit into the wider innovation strategy, see the pillar article How to build an innovation strategy in a few workshops.

Do

1. Write key results about learning, not about launches. "Launch the pilot app by March" is a task. It tells you nothing about whether the idea works. A learning key result names the assumption and the evidence: "By March, 40 small business owners have used the prototype, and we know whether at least half would switch from their current method." Start from an assumption map and turn the riskiest assumptions into key results. When the result is uncertain, set the threshold in advance, as in a hypothesis test.

2. Mark each OKR as committed or aspirational, out loud. Innovation teams usually carry both: committed OKRs for delivery they control (running a programme of experiments, handing a pilot to operations) and aspirational ones for outcomes they cannot guarantee. If you do not label them, leadership will read every miss as failure. State the type next to each objective, and agree in advance what score is good for each.

3. Match the key result to the horizon. Work that improves the core can use operational measures: time saved, error rates, satisfaction. Work in new areas cannot, at least not yet. Use the three horizons to sort your work, and give horizon-three projects key results about evidence and decisions, horizon-two about adoption and growth, and horizon one about performance. More on this in Three horizons in practice.

4. Include a decision as a key result. One of the most useful innovation key results is simply a decision reached on evidence: "By the end of the quarter, the portfolio board has made a fund, scale, pause or stop decision on all five pilots." It rewards closure rather than drift. A stop counts as success here. See How to stop a project well.

5. Link team OKRs upwards, lightly. Each innovation objective should connect to an organisational goal, so leaders can see why the work matters. A KPI tree or a one-page OGSM shows the link. Keep it to one clear line per objective; heavy cascading from the top tends to squeeze out the exploratory work that innovation teams exist to do.

Don't

6. Don't count ideas, workshops or participants. "Generate 200 ideas" and "Run 12 workshops" are easy to hit and say nothing about value. They also pull the team towards volume. If you must track activity, keep it in a work plan, not in your OKRs. The case against idea counts is made fully in Counting ideas is the worst innovation KPI.

7. Don't tie OKR scores to individual bonuses or appraisals. The moment scores affect pay or ratings, people set goals they know they can hit, and aspirational OKRs disappear. Innovation teams are hurt most, because their honest scores on uncertain work will often be low. Use OKRs to steer the team's work; assess individuals through separate conversations about how they worked.

8. Don't set OKRs for a year. Exploratory work changes direction as evidence arrives. Annual key results written in January are usually irrelevant by May. Set them quarterly, review them every two weeks in a short team check-in, and allow a key result to be retired openly if the team learns it was the wrong question.

9. Don't write more than three objectives. Teams with six objectives and twenty key results have a to-do list, not a focus. For a small innovation team, one to three objectives per quarter, with two to four key results each, is plenty.

10. Don't hide misses. An aspirational key result scored at 0.3 with a clear explanation of what was learned is useful. One quietly reworded mid-quarter so it looks complete is not. At the end of each quarter, show every score alongside what the team now knows. That habit, more than any template, is what makes OKRs work.

Illustrative example

Illustrative. The innovation team of a mid-sized insurer in Ireland had used OKRs for a year, mostly counting outputs: number of ideas submitted, workshops held, employees trained. All were hit; none convinced the executive team that anything had changed.

For the next quarter, the team rewrote its OKRs:

• Objective (aspirational): Find out whether small business customers will buy cover online without speaking to an adviser.
  • Key result: 30 customers complete a quote through a prototype, and we know what share would have bought it.
  • Key result: The three reasons customers abandon the quote are identified with evidence.
• Objective (committed): Make clear decisions on the current pilots.
  • Key result: Fund, scale, pause or stop decisions made on all four pilots at the quarterly review.
  • Key result: One pilot handed to an operational owner with a budget line.

At quarter's end, the team scored 0.6 on the first objective and 1.0 on the second. The executive discussion lasted half the usual time and was about what to do next, not about whether the team was busy.

Questions people ask

What is a good OKR example for an innovation team? An objective naming a question the organisation needs answered, with key results that describe the evidence: for example, "Learn whether residents will report street faults through a messaging app", with key results on the number of residents in a trial, the share who report a second time, and a decision by a set date.

Should innovation teams use OKRs or KPIs? Both, for different things. KPIs track ongoing performance of what exists; OKRs set a direction for change over a quarter. Innovation teams use KPIs for scaled services and OKRs for the exploratory work that might become new services.

How do you score OKRs for experimental work? Agree the scoring before the quarter starts. For aspirational key results, Google's re:Work guidance treats about 0.7 as a healthy result. What matters more than the number is that every score comes with what the team learned.

Where to go next

The guides used here are free: objectives and key results, assumption mapping, hypothesis test and KPI tree.