How to write KPIs leadership actually uses

Every BI engagement we’ve ever started begins with the same question: what are your KPIs? And every BI engagement’s first painful discovery is that the answer is a deck nobody owns. The framework we use to fix it is intentionally simple — three rules, then a workshop.

The most common KPI failure mode is not having too few metrics. It is having too many — a 40-row deck where every team contributed two metrics, no one is willing to remove any, and the leadership team’s actual attention is drawn to the same five every meeting. The point of the framework is not to be clever. It is to be ruthless enough that the resulting list is shorter than the one it replaces.

Rule 1 — Every KPI has exactly one owner

Not a team, not a function — a named person. If two people share ownership, neither will defend the number when it dips, and the metric quietly becomes a vanity stat. The owner is the person who has to explain a miss in the next leadership meeting. If you cannot name that person without checking a doc, the ownership is fiction.

Rule 2 — Every KPI maps to a decision

If moving the number wouldn’t change anyone’s behavior, it’s not a KPI — it’s a metric. Distinguish the two ruthlessly. Most leadership decks confuse them. Reporting metrics belong in operational dashboards. KPIs belong in leadership meetings because they drive decisions about resource allocation, prioritization, hiring, and product direction.

Rule 3 — Every KPI has a leading indicator

Lagging KPIs (revenue, churn) tell you the past. They’re necessary and insufficient. Pair every lagging KPI with at least one leading indicator the team can actually influence this week. A KPI deck that is entirely lagging metrics is a thermometer — useful for diagnosing temperature, useless for changing it.

A worked example: B2B SaaS

Here is a KPI hierarchy we sketched on a recent B2B SaaS engagement — anonymized, but structurally typical. The point is the shape: one north-star, a small number of strategic KPIs with named owners, and a leading indicator per KPI that operators can influence weekly.

North-star

• Net Revenue Retention (NRR) — owned by the CEO, reviewed monthly with the executive team

Strategic KPIs (lagging) and their leading indicators

• New ARR — owned by CRO; leading indicator: qualified pipeline coverage ratio
• Gross Margin — owned by CFO; leading indicator: cost-to-serve per cohort
• Logo Churn — owned by VP CS; leading indicator: at-risk account count from health-score model
• Activation Rate — owned by VP Product; leading indicator: week-1 product engagement score
• Sales Cycle Length — owned by VP Sales; leading indicator: stage-conversion time in stages 2 and 3

Anti-patterns we cut on every KPI workshop

• Metrics that move when we change the chart (definitional ambiguity)
• KPIs without a single named owner
• Lagging metrics with no leading counterpart
• ‘Engagement’ metrics that conflate sessions, time on page, and clicks
• Activity metrics dressed up as KPIs (calls made, emails sent, pages built)
• Multiple KPIs measuring the same outcome with different denominators

The KPI workshop

We run a single half-day session with leadership and produce a one-page KPI hierarchy: north-star, three to five leading metrics per function, defined owner, defined cadence, defined data source. Anything that doesn’t earn its row gets cut.

Output format

• North-star metric (one)
• Strategic KPIs (three to five, with owners)
• Leading indicators (three to five per function)
• Source-of-truth dataset for each
• Review cadence (weekly, monthly, quarterly)

Quarterly KPI review cadence

A KPI list that is never revisited becomes stale within two quarters. We run a quarterly review with leadership that asks four questions per KPI: is the owner still the right person, is the definition still correct, is the leading indicator still predictive, and should the KPI be retired? Most quarters, one or two KPIs change. Some quarters, half the deck does. The point is the deliberate reconsideration — KPI hygiene is a recurring practice, not a one-time deliverable.

The one-line takeaway

Then — and only then — do we touch a dashboard. A BI program built on the wrong KPI hierarchy is just a faster way to argue about the wrong numbers. Get the hierarchy right, give every metric an owner and a leading indicator, and the dashboards almost build themselves.