Defining KPIs: How a Business Goal Becomes a KPI That Management Trusts
Monday morning, management meeting. Sales reports that two key accounts are complaining about delivery performance. Production reports 92 % for the same month. Both sides calculate correctly – and still the next half hour goes on the question of which number is right. (The figure is an example; the situation is one we know from many engagements.)
Such differences usually have mundane causes. Internally, a delivery counts as on time within a tolerance band, say two days early or late; the customer works with the day they need the goods. What gets measured is the shipment ex works, what gets experienced is the arrival – in between lies transit time. And customers often compare against their requested date, which was never confirmed to them.
Added to this is a detail few people have on their radar: most ERP systems do not store the first confirmed date in the first place, it is overwritten with the first rescheduling. Yet that date would be the most honest benchmark for a company’s own performance. Without a binding definition, on-time delivery remains an opinion with a percent sign.
The two deviations also tell different stories. Missing the first confirmed date points to planning and processes. Holding it and still receiving complaints says something about the market: that customers expect more flexibility and availability than can currently be committed to. Both matter – they just do not belong in the same number.
Why good data alone is not enough
In the articles “Peak Performance through Transparency” we described why companies need transparency over a few, well-chosen KPIs and how reporting is built across several altitudes. This article goes one step deeper: how does a business goal become a single KPI that management trusts?
In our experience, the ERP system itself is rarely the bottleneck – the data in it often is. Where the use of the system is not standardized, too many degrees of freedom remain: dates are maintained differently, order lines closed differently, fields filled at will. In the evaluation, those degrees of freedom turn into large deviations. What is missing is an agreement on what exactly is measured, where the data comes from, how the system has to be used for it and who answers for the number. We record this agreement in a KPI profile. Getting there takes four steps.
Step 1: From business goal to lever
The starting point is not “What can we measure?” but “What do we want to achieve?”. A business goal such as “increase gross profit” or “respond faster to market and customers” is too abstract to measure directly. So we ask: through which levers does the company actually influence this goal? For “respond faster”, these are for example the reliability of order fulfilment, lead time and material availability.
The rule is simple: every lever gets one key KPI, and every KPI is assigned to a lever. A KPI that serves no lever triggers no decision. It fills reports, but not the agenda.
Levers come with trade-offs, and those have to be managed deliberately. Raising material availability usually drives inventory up; cutting inventory risks missing parts. We therefore run such pairs together: the key KPI and its counter-KPI sit side by side in the same cockpit, with a target range instead of a one-sided goal.
Step 2: From lever to performance parameter
For each lever we determine the performance parameter that represents it best. We distinguish two kinds. Lagging indicators show what has happened – for example last month’s on-time delivery. Leading indicators show what is going to happen – for example material coverage for the orders of the next four weeks, the order backlog relative to capacity, or the book-to-bill ratio, the relation between incoming orders and invoiced output. Above 1, the backlog grows faster than it is worked off; below 1, the company is living off its backlog.
A useful cockpit needs both. Lagging indicators alone turn management into historians, leading indicators alone into prophets. For every important lever it should therefore be clear which KPI steers looking back and which looking ahead.
Step 3: The KPI profile
Now the KPI is described so precisely that two people working independently arrive at the same result. The profile has ten fields:
- Name – one term used the same way throughout the company.
- Purpose and lever – which goal the KPI supports.
- Formula – numerator, denominator and period.
- Data source in the ERP – table, field and posting point.
- Measurement standard – the rules for edge cases.
- Master data prerequisite – what must be maintained in the system for the number to be right.
- Frequency – how often it is measured and reported.
- Target and tolerance – derived from the business goal, with an intervention threshold.
- Responsible person – who answers for the KPI, not who calculates it.
- Committee – where the KPI is discussed and measures are decided.
On-time delivery shows why the inconspicuous fields matter most. The measurement standard settles three things: the reference date, the measuring point and the tolerance. In our example, performance is measured against the first confirmed delivery date, at order line level, at shipment ex works; an order line counts as on time if it leaves the building no more than two working days before the date and not a single day after. Each of these is a decision. Measuring against the most recently confirmed date rewards rescheduling. Counting orders instead of order lines makes an order with one missing line look just as punctual as a complete one. And measuring ex works leaves out transit time – defensible, as long as everyone knows that the customer experiences the arrival.
Equally important is the distinction: on-time delivery measures reliability against the company’s own commitment. Whether the company meets the customer’s requested date is a different question and a separate KPI, delivery to customer request. Both have their place, but they do not belong in the same number – and they lead to different measures: the first to planning and processes, the second to the delivery concept, inventory and capacity.
The master data prerequisite is where many KPI projects fail. If the ERP overwrites the first confirmed date every time it is rescheduled, on-time delivery cannot be calculated cleanly – however good the formula. What is needed first is a field that keeps this date, and a rule on who maintains it. The profile exposes such gaps before the first number is reported.

Step 4: From profile to cockpit
Only once the profiles are in place is the reporting built. We assign the KPIs to three views: the company as a whole for management, sales, and order fulfilment. Each view contains only KPIs that its committee actually makes decisions on. On-time delivery thus appears in detail in the order fulfilment view and as a condensed value in the overall view.
In the management control loop, the cockpit then becomes a working tool: set the goal, measure, identify the deviation, decide on a measure, check the effect. Because the definitions are fixed, the discussion is no longer about the number but about the cause. We show how such a cockpit is built on the CEO cockpit page.
Speed beats perfection
A common mistake is to start with the tooling question. Selecting and implementing a business intelligence tool takes months – and during that time people keep arguing about numbers instead of causes. We therefore build the first evaluations deliberately in Excel, straight from the ERP extracts.
This has two advantages. First, the cockpit is up in weeks rather than quarters. Second, building it in Excel reveals exactly what a BI tool will later require: missing fields, unclean master data, edge cases nobody has settled. That clean-up is the actual work. Once it is done and the profiles have proven themselves in daily use, the reporting can be implemented quickly in Power BI or a comparable tool – without experimenting on the live system.
From practice
At an internationally active industrial group in automation technology, we defined the order fulfilment KPIs this way within one month. The effort lay less in calculating than in clarifying: Which date applies? Who maintains it? Who answers for the number? The result was a shared understanding across departments – and meetings that discuss measures instead of numbers.
Three key takeaways
- The definition matters more than the formula. Most conflicts about KPIs do not arise in the calculation but in edge cases that nobody has settled.
- Every KPI needs a lever, a counter-KPI and a responsible person. Without the first, things get measured; without the second, optimized one-sidedly; without the third, not steered at all.
- Credibility is decided not by the ERP but by the discipline with which it is used. A KPI profile shows early which master data and which maintenance rules are missing – and a first evaluation in Excel shows it faster than any tool project.
Would you like to know whether your KPIs show what you need for your decisions? The Quick Health Check takes five minutes to complete and shows which state your company is in and which fields of action offer the greatest leverage – as a five-page profile, reviewed by an experienced consultant and free of charge. Start the Quick Health Check.
Further articles:
Peak Performance through Transparency (Part 1 / 2) (link)
Peak Performance through Transparency (Part 2 / 2) (link)
The Leadership Control Loop: From Goal to Execution (link)
Where does your company stand?
Would you like to know whether your KPIs show what you need for your decisions? The Quick Health Check takes five minutes to complete and shows which state your company is in and which fields of action offer the greatest leverage – as a five-page profile, reviewed by an experienced consultant and free of charge.
Start the Quick Health Check

