Three KPI mistakes we see in every project review

The review opens with a dashboard that impresses at first glance: four pages, more than thirty tiles, each with a traffic light and a trend arrow. The CFO scrolls, the management team nods. Then the first question: which of these numbers triggered a decision last quarter? After a pause, someone points to order intake, a second person to liquidity. The rest is looked at but not used – and that is already the first of three KPI mistakes we find in almost every project review.

KPI mistakes share one pattern

The three mistakes look different but have the same root: the KPI was designed from the data outwards, not from the decision inwards. Whatever was technically available went on the dashboard; whatever someone asked for was added; whether any of it carries a decision, nobody checked. Ossola-Haring, Schlageter and Schöning describe three circles – supply, demand and need – and only the intersection of all three is relevant for steering (Ossola-Haring et al. 2019, pp. 94 f.). Werner Gladen calls the same situation "scarcity amid abundance": the data on offer, the demand from management and the objective information need do not coincide (Gladen 2014).

The cost comes in three positions. First, production effort: every tile needs a data source, a calculation logic and a reconciliation with the close. With thirty tiles of which five are used, a large part of the controlling team's work goes straight to the bin. Second, attention: a ninety-minute management meeting that spends sixty minutes on what a number means has thirty minutes left for action. Third, and most expensive: false confidence. The authors of the "11 Fallacies" put it plainly: complexity that can no longer be managed feigns transparency and control, and gives management the feeling of having everything in hand (Ossola-Haring et al. 2019, p. 96).

That is why every project review we run starts with the KPI, not with the report. The frame for this is our framework "From decision to KPI": decision → question → KPI → data → report → effect → review. All three mistakes sit at the third step, and that is where they can be fixed.

Mistake 1: More KPIs mean more control

How to recognise it. The dashboard grows, but nothing is ever removed. Tiles built three years ago for a specific problem keep running although the problem is solved. There is drill-down in every direction, but nobody can say which number carried a decision last quarter. Ossola-Haring et al. describe this sprawl as the product of habit, entropy and technical feasibility; KPIs that have been produced for a long time draw their right to exist from their history alone (2019, p. 93).

What it costs. Gunhild Posselt, who ran a business on KPIs herself, sums it up in one sentence: the more detailed the KPIs, the less meaningful the management. Interpreting the numbers attracts more attention than implementing measures (Posselt 2017, p. 29). This is mechanics, not discipline: every additional number has to be placed in context, that takes time, and the time is missing at the end of the meeting when the question is what to do now.

The rule. Depth of detail follows depth of use. We distinguish three kinds of report: the glasses (dashboard, overview for the management team), the page (report, analysis for the head of a unit) and the hammer (working report, operational tool for the people doing the work). The management dashboard fits on one page; drill-down exists only where a person actually does something with the detail. Every tile faces the purpose question: what are these numbers meant to say, and what is the point of them? Whoever has no answer deletes the tile or moves it to the appendix.

Mistake 2: A value without a reference is information

How to recognise it. A tile shows "Order intake TCHF 8,060" and a trend arrow. Whether that is good, you only know with prior year, budget and seasonality in mind. In the meeting everyone brings their own reference: the head of sales compares with last quarter, the CFO with budget, the board with prior year. Three people, three statements, one number. Or the comparisons exist, but arranged differently from tile to tile. For the "11 Fallacies" this is the very first fallacy: revenue on its own says almost nothing; meaning only arises in relation to other figures, to time and to the industry (Ossola-Haring et al. 2019, fallacy 1).

What it costs. The meeting debates what the value means instead of what to do. And because the reference is missing, so is the threshold: nobody can say from which deviation onwards a number deserves attention. So every number gets the same – or equally little.

The rule. Deviation instead of value. The target is the baseline; the actual only matters as a deviation from it. The comparison logic is fixed and the same everywhere: prior year – actual – budget, with time running from left to right. Where a forecast exists, it replaces budget as the reference in the following months. A threshold – for us usually ±5 % – separates noise from signal; whatever falls within it gets no comment. And wherever possible, a degree of achievement replaces the absolute figure: 97 % of budget says more at a glance than 8,060 against 8,330.

A value only becomes information when it deviates from a reference – and only becomes control when someone is responsible for it. Which brings us to the third mistake.

Mistake 3: What is measurable is controllable

How to recognise it. The dashboard shows commodity index, exchange rate and market growth next to delivery reliability and order backlog, as if they were the same kind of number. They are not: nobody in the company can move the first three. The second sign is the KPI without an owner. It exists because the data existed – Ossola-Haring et al. describe exactly this as a KPI that sits only in the "supply" circle: technically available, but neither asked for nor needed for steering (2019, pp. 94 f.). Gladen adds a line that reviews rarely draw: a KPI either serves your own decisions or steers other people's behaviour – and the analytical KPI for a unit need not be the same as the steering KPI for its head (Gladen 2014, p. 23).

What it costs. Two things. First, no effect: Kobi notes that measures which cannot, or can only with difficulty, be influenced do not have the potential to lead to concrete action (Kobi in Eschenbach et al. 2021, p. 374). Second, the behavioural effect that shows up too late: attach pressure or a bonus to a KPI and it gets moved – not always by the intended route. Posselt: the higher the pressure on performance, the stronger the urge to influence the system (Posselt 2017, p. 38). Delivery reliability, measured against the confirmed date, improves when dates are confirmed more generously. Inventory drops when goods receipts wait a day at month end. The number turns green, the problem stays.

The rule. Kobi names four referees for prioritising measures: strategic relevance, meaningfulness, controllability and availability (Kobi in Eschenbach et al. 2021, p. 373). We apply this filter in every review; controllability is where most tiles fail. In practice: every steering KPI has an owner and a cadence – how often does this person look at it, and what happens when the threshold is breached? External figures such as indices and exchange rates stay in the appendix as context, without a traffic light. And before a KPI is tied to a bonus, we establish which behaviour it will trigger under pressure. If the honest answer is "generously confirmed dates", the problem is not the KPI but its definition.

Three KPI mistakes at a glance

Mistake	How to recognise it	Rule
More KPIs = more control	Dashboard grows, nothing is removed; drill-down everywhere, decision nowhere	Depth of detail follows depth of use: glasses, page, hammer; management dashboard on one page
Value without reference = information	Bare number with trend arrow; everyone compares against something else; no threshold	Deviation instead of value: PY – actual – budget, ±5 % threshold, degree of achievement instead of absolute figure
Measurable = controllable	External figures next to internal ones; KPI without owner; number turns green, problem stays	Kobi filter: strategic relevance, meaningfulness, controllability, availability; owner and cadence per KPI

The test in four questions

In a review, the three rules become a test we apply to every KPI. Four questions, one answer each:

1. Decision: Which decision would be taken differently if this number came out differently? If none: delete or move to the appendix.
2. Owner: Who is responsible for the number, and how often does that person look at it? Without a name and a cadence it is not a steering KPI.
3. Reference: What is it compared against – prior year, budget, forecast – and from which deviation onwards is action taken?
4. Controllable: Can the owner move the value, and which behaviour does the number trigger under pressure?

Four times yes: the KPI stays. A no on decision or controllability: it goes, or moves to the appendix. A no on reference or owner is curable – work on the definition, not on the tool. Ossola-Haring et al. see their test as a regularly recurring process (2019, pp. 94 f.). So do we: once a year, and at the latest when strategy or organisation change.

Gladen traces the relief that KPIs provide back to a single reason: the scarcity of management time and attention, with Simons' "return on management" as the return on that scarcest of resources (Gladen 2014, p. 12). The motto Gladen puts at the front fits every KPI system: fewer numbers, but smarter ones (Gladen 2014, p. 9).

What you can do tomorrow

Take the current management report and go through it tile by tile with the four questions. Count how many KPIs pass all four; in our reviews it is rarely more than a quarter. For the rest, note where they fail – usually reference and owner, both fixable without a new tool.

If you would like an outside view: our project review takes a few days at a fixed price and ends with a written second opinion – which KPIs carry, which are missing, which should go.

Sources

• Ossola-Haring, C., Schlageter, A., Schöning, S. (2019): 11 Irrtümer über Kennzahlen. Mit den richtigen Erkenntnissen führen. Springer Gabler (in German).
• Posselt, G. (2017): Renditeorientierte Führungsstrategien. Kennzahlen motivierend einsetzen. Springer Gabler (in German).
• Gladen, W. (2014): Performance Measurement. Controlling mit Kennzahlen. 6th edition, Springer Gabler (in German).
• Eschenbach, R., Baumüller, J., Siller, H. (eds.) (2021): Funktions-Controlling. Praxishandbuch für Unternehmen, Non-Profit-Organisationen und die öffentliche Verwaltung. 2nd edition, Springer Gabler – including Kobi: Personal-Controlling, pp. 365–388 (in German).