A KPI (key performance indicator) is a measure that shows how close the business is to one of the few goals it has chosen for the period, with a written calculation rule, an owner, a target and a threshold. The short answer to how to set KPIs: start from the goal, not from whatever happens to be measurable.
First answer "what are we trying to achieve?", then pick one lagging indicator that shows the result and one or two leading indicators that signal it in advance, and tie each to a definition card that states its source and formula.
The KPI problem usually starts with measuring too much
In many companies the monthly management pack runs to dozens of lines: revenue, order count, customer numbers, stock, collections, scrap, returns, absenteeism, website visits. Each one measures something, but none is tied to a decision. The meeting notes that "sales are down" and moves on to the next line without discussing why, or who will do what.
The second common problem is that the same indicator is calculated differently in each department. Sales counts revenue by order date, accounting by invoice date, and finance nets off returns. Half the management meeting goes on arguing which number is right. The third is that the KPI has no owner: when it turns red, nobody is clearly responsible for acting.
The cost of wrong or undefined KPIs
A wrong indicator does not just waste time; it lets you make the wrong decision with confidence. Data quality expert Thomas C. Redman, writing in MIT Sloan Management Review in 2017, estimates the cost of bad data at 15% to 25% of revenue for most companies. That is an expert estimate rather than a measured statistic, but it shows that vaguely defined KPIs, hand-stitched spreadsheets and calculation rules that change from team to team carry a real price.
Measurement that is not tied to a goal is also a recurring issue in large change programmes:
BCG's 2020 research on digital transformation found that 70% of digital transformations fall short of their objectives; only 30% met or exceeded their target value and produced sustainable change.
A project that does not define measurable goals at the start cannot say at the end whether it succeeded; we apply the same principle to AI investments in measuring AI project ROI. In Türkiye, the measurement infrastructure itself is often the gap: according to TurkStat's 2025 survey on ICT usage in enterprises, only 6.5% of Turkish enterprises with 10 or more employees used business intelligence (BI) software, falling to 4.9% among those with 10–49 employees. For most smaller firms, KPIs still live in manually prepared spreadsheets.
How to set KPIs in seven steps
- Write the goal in one sentence. "Grow" is not a goal; "increase repeat-order revenue from existing customers this year" is. Three to five goals per period is plenty; if everything is a priority, nothing is.
- Choose one outcome measure per goal. This is the lagging indicator; it tells you whether the goal was reached (for example, repeat-order revenue as a share of total revenue). It is confirmed at month end and describes the past.
- Add one or two leading indicators. A leading indicator is a behaviour that drives the lagging one and can be measured more often: customers with no order in 60 days, quotes issued, quote conversion rate. The chance to intervene lives in the leading indicator. We explain how to spot customers whose order frequency is falling in customer churn analysis.
- Fill in a definition card. For each KPI, write the formula, the data source (which field in which system), what is included and excluded (returns, cancellations, intercompany sales), update frequency and owner. An indicator without a definition card is not a KPI.
- Set a target and a threshold. Use past data to set a realistic target and a "red" threshold. Also write down who does what, and how quickly, when the threshold is crossed.
- Calculate automatically, with one rule. Recalculating a KPI by hand every month lets the definition drift. The indicator should be computed straight from the source system, following the definition card. For KPIs fed by several systems, that single rule usually lives in a data warehouse and ETL process.
- Review quarterly. Drop indicators that nobody looks at, that never lead to a decision or that are being gamed. If the goal changes, the KPI should change too.
Examples of leading and lagging indicators
| Goal | Lagging indicator (outcome) | Leading indicator (early signal) | Owner |
|---|---|---|---|
| Grow repeat business | Share of revenue from repeat orders | Customers with no order in 60 days | Sales manager |
| Improve cash flow | Average collection period | Receivables falling due within 7 days | Finance manager |
| Raise production efficiency | OEE | Unplanned downtime minutes per shift | Production manager |
| Customer satisfaction | Complaint rate | First response time | Customer service |
| Reduce stock cost | Inventory turnover | Value of stock with no movement in 90 days | Supply chain |
To expand the sales and customer rows of the table, see sales dashboard metrics for an indicator set built for sales teams, and how to measure NPS for tracking customer satisfaction with a single number.
The same logic applies to sector-specific indicators. In manufacturing the most common outcome measure is OEE; splitting it into availability, performance and quality shows which loss to tackle first (how to calculate OEE). In energy-intensive plants, energy consumed per unit produced is tracked instead of the total bill, because it strips out changes in output volume; we cover the metering side in factory energy monitoring.
A checklist for a good KPI
Before adding an indicator to the list, ask: Is it tied to a goal? Will someone do something differently when it moves? Are its formula and source written down? Does it have an owner? Can it be calculated automatically? Can it be gamed (if you only count "tickets closed", tickets may be closed unresolved, which is why call centre speech analytics measures quality as well as speed)? If any answer is no, fix the indicator or remove it.
Cascading company KPIs to departments
If the KPIs on the executive screen are not linked to departments' daily work, teams invent their own measures and coherence is lost. A KPI tree helps: the company goal and its lagging indicator at the top, the departmental measures that drive it below, and at the bottom the leading indicators teams can influence week by week. "Average collection period", for example, breaks down into "share of orders written on standard payment terms" for sales and "share of receivables due within 7 days that received a reminder" for finance, each with its own definition card and owner.
The most common KPI mistakes
Vanity metrics. Website visitors, follower counts or total registered customers feel good when they rise but lead to no decision on their own. They only become useful when linked to a conversion or revenue measure.
Tracking only lagging indicators. Management that watches only revenue and profit learns about problems after the month has closed. Without leading indicators, a KPI system becomes a rear-view mirror. And the work does not end when a KPI turns red: in production, the "why" is answered by root cause analysis with manufacturing data.
Turning the measure into the target. When an indicator is tied directly to bonuses or penalties, teams find shortcuts to improve the number. Every KPI linked to pay should be paired with a balancing quality measure, such as error rate alongside speed.
Not writing the definition down. Everyone assumes "revenue" means the same thing; until returns, discounts and intercompany sales are settled in writing, the same KPI will show two different numbers in two reports. Keep definition cards in one place with version history rather than in email attachments: in SmartFiles, part of the Smart360 family, every file uploaded under the same name is saved as a new version, and earlier versions can be downloaded or restored, so you can see when a definition changed.
Once your KPIs are set, our guide to building a management dashboard covers how to bring them together on one screen, and moving from Excel reports to BI covers the switch from hand-built spreadsheets to automated reporting. Data warehousing, GIS and other topics are gathered in our Data & Analytics guide.
How we build KPI systems at Digital Bridge
We do not hand over a list of KPIs and walk away; we make sure the indicators are calculated automatically, from the source systems, with one rule. A typical engagement looks like this:
- Goals and KPI workshop. With your management team we choose the period's three to five goals and the lagging and leading indicators for each, and fill in the definition cards. This often sits within our digital transformation consultancy, alongside your wider roadmap.
- Mapping data sources. We work out which field in which system feeds each indicator: ERPs such as Logo, SAP and Mikro, CRM, production and e-commerce systems. Missing or conflicting data is fixed at source through data governance and quality work.
- One place to calculate. Where data comes from several systems, we consolidate it in a data warehouse and calculate each KPI once, following its definition card, so everyone sees the same number.
- Executive dashboard. We present the indicators on a BI dashboard built for your business, with threshold alerts and drill-down to invoice or work-order level. For production KPIs, the dashboard can connect directly to MES production management data.
- Review rhythm. After the first three months we look together at which indicators led to decisions and which did not, and we simplify the list.
Scope, phases and cost are set out in a written proposal after the needs analysis.
Next step
Open your latest management report and write two notes next to each line: which goal is this indicator tied to, and what decision was made last month by looking at it? The lines where both notes stay blank show where your KPI set needs trimming. Then get in touch and we will review your goals and data sources together and draft your first KPI set, complete with definition cards.