Data vs intuition. The KPIs that actually matter for a transformation.
Transformations get instrumented with dashboards no one reads. The KPIs that matter are fewer and harder to define than the dashboards suggest, and the discipline of picking them is something most leadership teams underestimate at the start of a programme. This piece walks through the framework we use in our work to land on the small number of measures that actually drive decisions, and where intuition still has a place even when the data is good.
Why dashboards proliferate while signal weakens.
When a transformation programme of any size kicks off, every function in the organisation wants its question answered. Finance wants to see the cost trajectory. Operations wants to see the throughput. Sales wants to see the pipeline impact. The change office wants to see adoption. The board wants to see the strategic indicator. The natural compromise is to build a dashboard that tries to satisfy all of them at once, and the natural result is a dashboard with thirty-five tiles that no one in the executive room actually uses to make a decision.
In the steering committees we attend with clients, the dashboards on the screen are often more elaborate than the conversation needs them to be, and the team has implicitly accepted that the document on the wall is for reassurance rather than for decision. The pattern is consistent enough that we treat it as a warning signal. The moment the dashboard requires a guided tour to be understood, it has stopped being a decision tool.
The discipline we recommend is the opposite of the natural tendency. Pick fewer measures. Define each one at the altitude of the decision the executive will actually make from it. Resist the request to add a thirty-sixth tile, and instead ask which existing tile the new measure would replace. The friction of that question is what protects the signal of the few measures that actually matter.
The four KPI families that actually drive transformation decisions.
Across the transformation work we have run, four KPI families recur in the dashboards that worked, and one or more of them tends to be missing from the dashboards that did not. The discipline of holding to one measure per family is what makes the framework useful. Adding a second measure to any family dilutes the conversation, and we resist it more often than clients expect.
The leading business KPI.
One measure that tells the executive whether the transformation is producing the business outcome it was funded for. Revenue from the new channel, gross margin from the new product line, net retention in the segment the transformation targets. The leading KPI sits closest to the business case, and the team should be able to explain in a single sentence why a movement in this measure should make finance more confident that the programme is working.
The adoption metric.
One measure that tells the executive whether the people the transformation depends on are actually using the new way of working. Active operator count, percentage of transactions processed through the new flow, weekly active teams on the new platform. Adoption is the leading indicator of the leading indicator. If adoption is not moving, the business measure will not move in any sustainable way, and the conversation in the steering room should turn to change management rather than to performance.
The quality metric.
One measure that tells the executive whether the new way of working is producing the quality the business depends on. First-pass yield, customer satisfaction on the new flow, defect rate, audit findings on the new process. Quality is the family most often overlooked early in a transformation, because the team is preoccupied with throughput, and it is the family that almost always surfaces an embarrassing finding three quarters later when the operating model is asked to scale.
The throughput metric.
One measure that tells the executive whether the operating capacity is shifting at the rate the programme requires. Volume processed per period, cycle time, time-to-decision, transactions completed. Throughput is the family that finance pays the most attention to, and it tends to be reasonably easy to instrument, which makes the discipline of holding to a single measure more important rather than less.
Four families. One measure per family. The executive looking at the dashboard sees four numbers, can explain what each one means in a sentence, and can tell what action a movement in either direction should trigger. That is what a transformation dashboard is for, and most of the dashboards we encounter have lost sight of it.
How to pick the right level of detail.
The hardest part of the framework is not picking the four families. It is picking the right altitude for each measure. The same transformation can be measured at a dozen different altitudes, and the right one depends on who will be reading the dashboard and what decision they will make from it.
The test we use is straightforward. The executive should be able to look at the measure, form a view about whether the transformation is on track, and identify the next action that should follow from the reading. If the measure requires the executive to drill down into three layers of detail before forming a view, it is pitched too low. If the measure can move significantly without changing what the executive should do next, it is pitched too high.
In practice the leading business KPI sits at the altitude of the line in the finance report. The adoption metric sits at the altitude of the team or business unit running the new flow. The quality metric sits at the altitude of the customer experience or the regulator's perspective. The throughput metric sits at the altitude of the operating capacity the executive is responsible for. Lower than those altitudes belongs in operational dashboards owned by the teams. Higher than those altitudes belongs in the board pack.
When clients ask us to add a second measure to one of the families, the conversation tends to surface that two altitudes are needed for different audiences. The right move is usually to build two dashboards rather than to bloat one, with the executive view holding to four measures and the operating view carrying the detail the teams need.
When intuition still belongs in the decision.
Even with a clean framework and a small number of measures, the data will not carry every call. Early in a transformation the leading business KPI is often noisy enough that reading it as a clean signal would be a mistake. The adoption metric may be moving on a base small enough that statistical inference is fragile. The quality metric may be lagging by a quarter because the customer feedback has not arrived yet. In those moments the executive will reach for intuition, and the question is not whether they should, but whether they will name what they are doing.
The discipline we recommend is to mark intuition explicitly. When the steering committee makes a call that is not supported by the data on the dashboard, the minutes should record that the call was made on intuition, with the reasoning the executive offered as their basis. The point of the record is not to second-guess the call. It is to allow the team to look back six months later and ask whether the intuition was right, which is the only way the leadership team learns when the data has earned the right to lead and when it has not.
In our experience the executives who are most comfortable with this discipline are also the ones who tend to make the better calls. The ones who insist that every decision is data-driven, when the data is too thin to support that claim, tend to make worse decisions because the data ends up rationalising the intuition rather than the intuition being held up to honest scrutiny.
The executive who says "I am calling this on intuition" is being more rigorous than the one who insists the dashboard told them what to do.
The discipline of fewer numbers.
Most transformation dashboards we encounter could lose half their measures and gain signal. The discipline of fewer numbers is not a preference for simplicity. It is a recognition that the leadership team has finite attention, that every additional measure consumes some of that attention, and that the measures which survive a careful prune are the ones which actually drive decisions.
When prospective clients ask where to start with their transformation measurement, the first exercise we suggest is a retirement review. Walk through the existing dashboard, ask of each measure what decision it has actually driven in the last quarter, and retire the ones that cannot point to a single decision they changed. The exercise is uncomfortable for the teams that produced the dashboards, and it is usually the most useful piece of work the leadership team can do before defining a new set of measures from scratch.
Four measures. Each one connected to a decision. Intuition marked when it is doing the work. That is the framework we use, and it has survived several years of transformation engagements without needing to grow.
Frequently asked questions.
What dashboard tooling do we recommend for transformation KPIs?
The tooling matters less than the discipline. A simple weekly view in whatever business intelligence layer the client already uses is preferable to a custom platform. The KPI framework should fit in a single page, and the tooling should not be a reason to add complexity. We tend to start with whatever the finance team already runs.
How often should transformation KPIs be reviewed?
Weekly at the operating level, monthly at the steering committee, quarterly at the board. The cadence is part of how the discipline holds. A KPI reviewed too rarely loses signal value, and a KPI reviewed too often produces over-correction from short-term noise.
How do you handle stakeholder demands for more KPIs?
By naming the cost of additional KPIs explicitly. Each new KPI consumes attention from the leadership team and dilutes the signal of the existing ones. We propose retiring a KPI for every new one added, which forces the conversation about which signals actually drive decisions.
How does the framework integrate with finance reporting?
The leading business KPI in our framework should connect explicitly to the finance hierarchy, so that improvements in the leading indicator translate visibly to the line items the finance team owns. The other three KPI families do not need to be in the finance report, but the connection must be defensible.
What kinds of KPIs should be retired first?
Activity metrics that count actions without connecting to outcomes. Vanity metrics that track exposure rather than impact. Granular operational metrics that belong at the team level and not the executive level. The retirement test is whether removing the KPI would change a leadership decision.
How do you handle the ambiguous early period of a transformation?
Early in a transformation the leading business KPI is often noisy and the adoption metric carries more signal. We weight the dashboard accordingly for the first ninety days, then rebalance once the leading indicator has stabilised. The shift is communicated to the steering committee as a deliberate choice.
Where this lands
How we'd take this further with you.
Consulting pillar
Performance & Value
Measurement systems, KPI design and value tracking that actually move the business.
Consulting pillar
Transformation & Organisation
Operating model design, change management, capability build for what comes next.
Consulting pillar
AI-Augmented Enterprise
From maturity diagnosis to use case prioritisation to durable adoption across the organisation.
Writing is one thing. Shipping is the other. Selected work from the partners writing here.
See the work

