Fewer indicators, more decisions
IN BRIEF
A KPI earns its place when you know what decision it supports, how it is calculated and who acts if it crosses a threshold.
Most teams have plenty of KPIs. Often they have too many, use different definitions and have no clear action when a figure moves.
When we work with your teams, we always ask which decision each indicator should help them make. If that answer is unclear, the KPI will be unclear too.
A KPI is useful when someone knows what to do if it crosses a threshold. If your service level falls, who investigates and how soon? If your margin falls, what level of detail do you examine? These questions separate indicators that guide action from figures that are merely interesting.
A dashboard with twenty charts can still be unhelpful. Three well-defined indicators can guide an entire meeting. The right number depends on the process, but an SME’s summary page usually benefits from focusing on three to five priority signals.
A dashboard should bring forward the signals that call for a decision.
KPIs, metrics and objectives serve different purposes
An objective describes the result you want, such as improving margin. A metric measures something, such as the number of order lines. A KPI is a metric you choose to track an objective, with a target, a review frequency and a planned response. Revenue might be a KPI for executives and background information on the production floor.
The method in six steps
1. Start with the business objective
Describe the objective before putting a number on it. You might want to protect margin, reduce delays or improve availability. Choose the KPI afterwards to make progress visible. Starting with an available metric can lead you to manage what is easy to count rather than what matters.
Choose a clear objective, such as reducing stockouts, improving on-time delivery, raising margin on priority products or shortening the financial close. Ease of extraction alone is a poor reason to select a KPI.
2. Identify the decision to make
Ask what will change if the indicator turns red. Rescheduling production, contacting a customer, changing a price and investigating scrap are all observable decisions. If no action or discussion follows, the indicator is probably secondary.
3. Define the formula and scope
A short formula can hide many decisions. Which costs go into the margin calculation? What period determines whether a customer is active? We write these choices in business terms and validate a typical case as well as an edge case. That gives you a KPI you can defend.
Document the numerator, denominator, inclusions, exclusions, reference date, unit and filters. A service level can be calculated by order line, order, quantity or value, and each gives a different result. The review committee should approve one definition.
4. Check the source and frequency
Is the data available at the right level of detail? Does it arrive early enough to act? Even a perfect measure is of little use for daily operations when it arrives three weeks late. Name a data owner and track rejected records, delays and completeness.
5. Set a target and thresholds
Base the target on a budget, customer commitment, operational capacity or internal plan. Avoid arbitrary thresholds. Use a warning band when the precision of the data does not justify an abrupt change from green to red.
6. Name an owner and establish a review routine
The person responsible for the KPI may be different from the person who produces the data. The owner coordinates the response to a deviation. Record when the KPI is reviewed, who takes part and what decisions are recorded.
A useful KPI connects strategy, reliable measurement and action.
Pair outcome indicators with leading indicators
Revenue, margin and the share of late deliveries reflect outcomes that have already occurred. An order book, conversion rate, inventory coverage or workload can signal what is coming. A balanced dashboard uses both types. Outcome indicators confirm performance, while leading indicators give you time to act.
Examples of KPIs by function
| Function | Objective | Possible KPI | Response to a deviation |
| Sales | Build a healthy margin | Margin by customer and product group | Review product mix, discounts or pricing |
| Finance | Close the books faster | Days from month-end to approval | Address the highest-priority bottleneck |
| Logistics | Reduce stockouts | Stockout rate and inventory coverage | Adjust replenishment |
| Production | Meet the production schedule | Schedule adherence and scrap | Reschedule or investigate the cause |
| Customer service | Resolve requests faster | Median resolution time and cases outside the SLA | Reallocate capacity |
The five most common pitfalls
The common thread is a lack of follow-through. An accurate, attractive and regularly updated indicator can still have no effect on what people do. Remove or de-emphasise KPIs that prompt no questions, decisions or learning.
Avoid vanity metrics that improve without reflecting value, multiple definitions under one name, excessive decimal places or false precision, comparisons across dissimilar periods or groups, and KPIs with no target, owner or expected action.
Build a definition sheet for each KPI
Create a short record of each indicator’s business name, objective, definition, formula, unit, level of detail, source, owner, review frequency, target, thresholds and access rules. Include a calculation example and the date of the last change. This reduces arguments over the figures and makes handover easier.
Test the dashboard in a meeting
Watch how quickly the discussion moves. If your teams still spend fifteen minutes disputing the definition, revisit the model. If they spot the variance, open the relevant detail and agree on an action, the KPI is doing its job. We observe that use before considering the work complete.
Show the dashboard in a real review and listen to the questions. Do participants understand the colours? Can they drill down to useful detail? Do they make a decision? Remove elements people never use. The quality of the discussion is part of a KPI’s value.
Datavizin’s ECSTAA model connects exploring sources, structuring data, analysing information and acting on it. The Data Starter Pack includes defining the scope of three to five indicators for an initial dashboard.
NEXT STEP
Pick a priority objective and complete a KPI definition sheet. If the source, formula, target or action is still unclear, resolve it before designing the visual.