KPIs are the specific metrics organizations use to evaluate whether they are achieving their strategic objectives. This guide explains what makes a metric a KPI versus a general measurement, how to define KPIs that are actually actionable, and how analytics infrastructure should support them.
A KPI — Key Performance Indicator — is a specific, measurable value that an organization tracks to evaluate whether it is making progress toward a defined strategic objective. The definition seems simple; the implementation is where most organizations struggle.
The word "key" is load-bearing. Not every metric an organization measures is a KPI. Revenue, headcount, customer count, server uptime, support ticket volume, average session duration — these are all measurements. A KPI is a measurement that is directly connected to a specific objective the organization is trying to achieve, measured at the right frequency for the decisions it informs, and actionable — meaning someone is accountable for it and can do something meaningful in response to it moving in the wrong direction.
What Makes a Good KPI
**It is connected to a specific objective.** The objective should be explicit: "increase revenue from enterprise customers by 20% this year" is an objective; "grow the business" is not. A KPI for the first objective might be enterprise new ARR or enterprise pipeline coverage. A KPI for the second is impossible to define because the objective is not specific.
**It is measurable with available data.** A KPI you cannot reliably calculate is not a KPI — it is an aspiration. Before committing to a KPI, verify that the underlying data exists, is reliable, and can be measured at the required frequency. Organizations that define KPIs before assessing data availability create dashboards full of "data unavailable" placeholders.
**It is actionable.** Someone must be accountable for it and capable of taking action that affects it. A KPI that no one owns and that no action can meaningfully influence is a vanity metric — it may show you the state of the world but it cannot improve.
**It is measured at the right frequency.** A KPI for a daily operational process (same-day order fulfillment rate) should be monitored daily. A KPI for a quarterly strategic objective (net revenue retention) is reviewed quarterly. Measuring a quarterly metric daily adds noise without adding insight — daily fluctuations in quarterly metrics are meaningless.
**It is specific enough to diagnose.** "Customer satisfaction" is too vague. NPS score, CSAT score from post-ticket surveys, and executive relationship health score are specific and separately actionable.
KPIs vs Metrics vs OKRs
These terms are often conflated. The distinctions are practically useful.
**Metric:** Any quantified measurement — clicks, revenue, churn rate, response time. Metrics are the raw data layer.
**KPI:** A metric that is specifically connected to a strategic objective, owned by someone accountable, and reviewed at a defined cadence. All KPIs are metrics; not all metrics are KPIs.
**OKR (Objective and Key Result):** A goal-setting framework where an Objective is a qualitative direction ("become the most-trusted analytics partner in mid-market financial services") and Key Results are specific, time-bound, measurable outcomes that define success for that objective. Key Results are closer to KPIs than to general metrics — they are specific, measurable, and time-bound.
The practical difference: OKRs are aspirational goals set at the planning horizon. KPIs are ongoing operational measurements that tell you whether the business is performing as expected on dimensions that matter to strategy. OKRs and KPIs inform each other but serve different functions.
Common KPI Failures
**Too many KPIs.** An organization with 47 company-wide KPIs has none — attention is dispersed, accountability is unclear, and the signal-to-noise ratio makes it impossible to identify what actually matters. Companies that operate well on KPIs typically have five to eight at the company level, with supporting metrics that teams use operationally.
**KPIs that lag too far to be actionable.** Annual revenue reported quarterly cannot inform decisions that need to be made weekly. Lagging indicators (outcomes) must be paired with leading indicators (predictors of outcomes) that give early warning of whether you are on track to hit the lagging indicator.
**KPIs defined without data infrastructure to support them.** Committing to tracking customer lifetime value monthly is meaningless if the data required to calculate it — purchase history, support costs, churn dates — lives in four different systems with no integration. KPI definition must be accompanied by an honest assessment of data availability and quality.
**KPIs that measure activity, not outcome.** Number of sales calls made measures activity. Pipeline generated by those calls measures outcome. Activity metrics are useful as operational metrics for team management; they are not KPIs unless activity directly drives the strategic objective.
KPIs and Analytics Infrastructure
KPIs should be implemented at the semantic layer — the layer between raw data and the BI tools that consume it. Implementing KPI calculations in the semantic layer (dbt Semantic Layer, Looker LookML, Cube.dev) rather than in individual dashboards means:
- **Single definition:** Revenue is calculated the same way everywhere it appears — finance dashboard, executive report, sales performance board
- **Auditable:** The calculation logic is version-controlled and reviewable
- **Trustworthy:** When a stakeholder sees the same number in two different reports, it is the same number
- **Maintainable:** When the definition changes — fiscal calendar adjustment, revenue recognition policy update — it changes in one place
Organizations where KPI definitions live in individual Tableau workbooks or Power BI reports discover their KPIs diverging over time. Different analysts make different modeling decisions; definitions drift; stakeholders notice that two dashboards show different numbers for "the same metric" and trust collapses.
The investment required to implement KPIs well — clear objective definition, reliable data infrastructure, semantic layer encoding of the calculation, governance process for managing definition changes — is the investment required to make analytics actually drive decisions rather than inform post-hoc justification.
Our BI strategy services help organizations define KPI frameworks, implement them in the analytics infrastructure, and establish the governance processes that keep definitions trustworthy over time. Contact us to discuss your analytics requirements.
A former Microsoft data architect audits your data foundation, identifies your top priorities, and sends you a written plan. Free. No pitch.
Book a Call →