Most strategies are not measured badly. They are measured with instruments built for something else.
Every dashboard is green and nobody can say whether the strategy is moving
The quarterly strategy review opens with the scorecard. On-time delivery holds at target, margin is up a point, inventory turns beat last year. Then someone asks whether the company is any closer to the service business the board approved eighteen months ago, and the room goes quiet. Nobody is hiding anything. The numbers on the screen simply cannot answer the question. That silence is the most common failure in measuring strategy execution: nobody defined the key results the strategy needed, so operational KPIs stand in for them.
The reflex is understandable. When a leadership team asks how it will know whether the strategy is working, the first answer is almost always “through our KPIs”. They are the only measurement system most organisations already run. They have owners, definitions, data feeds and a slot on the monthly agenda. Reaching for them feels like rigour. In practice, it is closer to tracking progress on a new building with the thermostat of the old one.
KPIs measure the health of the machine, not the change you asked it to make
A KPI tracks a process that is meant to keep running: orders shipped, cash collected, defects caught. Its job is to tell you whether the machine performs within its expected range. A strategic objective asks for something different. It asks the machine to change shape: enter a new segment, move from product sales to service revenue, halve the time to launch. The state the objective describes does not exist yet, so nobody ever built an operational KPI to observe it.
That is why a strategy can stall while every KPI stays green. The organisation keeps running the business it has, and runs it well. The business it said it would build receives no measurement at all, so its absence never shows on the dashboard.
None of this makes KPIs irrelevant to strategy. They do the essential job of a guardrail. A transformation that quietly erodes delivery performance or margin is failing, even if it hits every strategic milestone. The error is not using KPIs. It is expecting them to report on a change they were never designed to see.
There is one legitimate overlap. When the objective is a step change in an existing measure, such as lifting delivery reliability to a level the current network cannot reach, the KPI can serve as a key result. It then carries a new target and a deadline, and it is measured as evidence of a changed system rather than as routine performance.
A key result is the objective’s own evidence, and it is often unique to it
The OKR method, which Andy Grove developed at Intel and John Doerr later introduced to Google and many other companies, starts from the opposite end. It begins with the objective and asks what evidence would show, without argument, that the objective is being achieved. That evidence is the key result. Because it comes from one specific objective, it often has no counterpart anywhere else in the organisation’s reporting.
Most discussion of OKRs centres on ambition and prioritisation: setting objectives that stretch the organisation and forcing a choice about what matters most. Their value as a measurement discipline gets far less attention, and that is where strategy reviews tend to fail.
Take an objective to build a service business. Its key results might be the share of revenue from recurring contracts, the number of customers renewing after the first term, and the service margin at contract level. None of these lives in a standard operations scorecard. Each one describes the new state the strategy is meant to create.
The most common way to waste the method is to write activities in the key-result column. “Launch the service portal” and “hire three field engineers” are tasks. A team can complete all of them while the objective goes nowhere. A key result describes an outcome someone could observe in the business, not an output the team can tick off. When the column reads like a project plan, the organisation has renamed its task list and gained nothing.
One key result per objective is a single point of failure
A single key result is tempting because it is simple to report. It is also easy to satisfy without achieving the objective. Recurring revenue share can rise because the product business is shrinking. Renewals can improve because sales offered deep discounts to lock customers in. A lone measure invites people to optimise the measure itself, which is Goodhart’s law in daily practice.
The protection is triangulation. Two to four key results per objective, each looking at it from a different side, make it much harder for one favourable number to tell a false story. Grove himself argued for pairing indicators so that every effect has its counter-effect in view: volume with quality, speed with error rate, growth with margin.
A useful set usually mixes leading and lagging measures. The lagging key result confirms the outcome, but it arrives late. The leading one moves first and gives the review forum something to act on while there is still time. Objectivity, in this sense, is not a property of any single number. It comes from several independent pieces of evidence pointing the same way.
If it is not a KPI, nobody is collecting it yet
This is the step most OKR rollouts skip. A KPI arrives with its data pipeline already built. A key result usually does not. Contract-level service margin may depend on a cost allocation nobody has agreed. Renewal rates may sit in a CRM field that sales fills in inconsistently. The measure exists on the slide long before any system can produce it.
A key result without an organised data source is an opinion with a number attached. Someone has to decide where the data comes from, who supplies it, how often it is refreshed and what the baseline was before the work started. A baseline reconstructed after a quarter of effort tells you very little about the effect of that effort.
This is real work, and it needs time and people. Organisations that budget weeks for writing the strategy and nothing for making it happen tend to treat measurement the same way. The objectives get a workshop; the data collection gets a volunteer.
A measure is only as objective as its operational definition
Quality management settled this question long before OKRs became fashionable. W. Edwards Deming’s point was that a measure means nothing until it has an operational definition: a description precise enough that anyone following it would count the same things in the same way. “Customer renewal rate” is a label. The operational definition states which contracts count, what qualifies as a renewal, over which window, and which system supplies the numbers.
Six Sigma adds the second half of the discipline: measurement system analysis. Before a team trusts a process metric, it checks whether the measurement itself is sound. Is it repeatable, so the same person measuring the same thing twice gets the same result? Is it reproducible, so two different people get the same result? Is it stable over time and free of systematic bias? A gauge that fails these tests produces numbers, not evidence.
Strategic measures rarely get this scrutiny, and they need it more. They are new, they are often assembled by hand, and they carry more political weight than any operational metric. The test is simple enough to run in a review. If two people measured this key result independently next week, would they report the same figure? If the honest answer is no, the key result is not ready to judge the objective.
Validity is the final check, and no statistical test can make it for you. A measure can be perfectly repeatable and still point at the wrong thing. Before committing to a key result, ask whether a strong reading would convince a sceptical board member that the objective is advancing. If it would not, the precision is wasted.
Measurement is where strategy execution becomes accountable
A well-designed set of key results still does nothing on its own. Numbers become accountability only when a forum reviews them on a fixed cadence and has the authority to change course: reallocate resources, stop an initiative, escalate a stalled objective. Without that forum, even excellent measurement turns into reporting nobody acts on.
This is also where strategy execution works as a discipline rather than a leftover duty. The executives accountable for the objectives own their content. The measurement system needs an owner too: someone responsible for the definitions, the data flows, the baselines and the review rhythm that keep every key result honest. In many organisations, that is the natural job of a PMO that has moved beyond tracking projects.
Run the business with KPIs. Judge the strategy with evidence built for it.