Key performance indicators (KPIs)

What are key performance indicators?

KPIs (Key Performance Indicators) are a small number of measures that show whether a team is moving towards an important goal and help it make the next decision. Not every number in a report is a KPI. Follower count, page views or emails sent may be interesting, but without a connection to a goal they easily become vanity metrics.

A good KPI works like a signpost: it says what is being observed, over what time and for what reason. For example, “number of orders” is not enough to assess a store if we do not know whether growth came from returning customers, a new campaign or a lower margin.

Start with the decision, not the dashboard

First name the goal and the person who can influence it. Only then choose the measure. If the goal is to shorten request handling, measure the time from a case arriving to its resolution and the share of cases requiring another contact. If the goal is more valuable enquiries, lead source, brief completeness and progression to a conversation are useful — not page visits alone.

For every KPI, answer four questions:

• What goal does it describe?
• Who owns the outcome?
• What action will we take if the result worsens?
• What data and definitions are needed to calculate it?

If there is no answer to the third question, the measure may be interesting, but it is not yet a management tool.

Examples of KPIs in different situations

An online store may track purchase rate, average order value, abandoned baskets or the share of returns. A service company may track response time, qualified enquiries, value of active projects and customer retention. For a digital product, activation, return after first use and completion of a key task can matter.

The same measure can mean something different in different models. A high bounce rate on an article need not be a problem if the reader got the answer and ended the visit. It becomes a signal for analysis when the user should continue but does not.

Leading and lagging indicators: why pair them?

A lagging indicator describes an outcome that has already happened: revenue, margin, retention or completed projects. A leading indicator shows behaviour that may lead to that outcome earlier, such as time to first response, completed onboarding or regular use of a key feature.

The outcome alone tells you whether the goal was met, but it often appears too late to react easily. A leading indicator alone may rise without improving the business. Pair them and test the relationship. If more completed onboardings do not improve later activity, verify the hypothesis rather than celebrate the movement on a chart.

Is bounce rate a good KPI?

Only when its definition and goal fit the page. In Google Analytics 4, bounce rate is the percentage of sessions that were not considered engaged. An engaged session lasts longer than 10 seconds, has a key event or contains at least two page or screen views. This differs from the former definition of “enter and leave without another page view”. Current Google Analytics definition

Do not transfer historic benchmarks such as “a good result for a blog is X–Y%” to a current report. First check the configuration, traffic source and the page’s task. A definition article may meet a need in one visit; a high share of unengaged sessions in a multi-step form needs diagnosis. Compare similar pages and your own trend, not a random industry table.

Set a definition and measurement cadence

The team must understand a measure in the same way. Record exactly what you count, which data source you use, the comparison period and exclusions. An “active user” may mean logging in, creating a document or closing a task — without a definition, a report will only appear precise.

Not every KPI needs daily review. Operational measures need a more frequent cadence than quarterly goals. The timing of the decision is what matters: the report should arrive while there is still time to act.

A good definition test is to walk through one real scenario. If qualified enquiries fall, the team should know whether to first check traffic sources, the site message or how contacts are qualified. If they rise but sales do not, establish whether lead quality, the offer or the sales stage changed. A KPI does not solve the problem itself; it orders the questions and ownership of the answer.

Fewer measures, a better conversation

When a dashboard contains everything, the most important information disappears. Choose a few main KPIs and keep supporting metrics for diagnosis. If sales fall, then examine traffic sources, funnel stages and qualitative data. Do not try to manage every layer every day.

KPIs are especially useful when they connect data with experimentation. First formulate a hypothesis, then check whether the change improved the relevant measure without harming other important outcomes. Our article on growth hacking describes this loop too.

Before the next report meeting, briefly record the result, possible cause, decision and person responsible for the next step. That way indicators do not end with a presentation, but become part of the working rhythm. If the definition or data source changes, note it with the report — comparing incomparable periods creates a false sense of control.

If you need reporting that does not end with a table of numbers, see how we approach data and analytics. When optimising a purchase journey, conversion rate is a natural complement.