A KPI is a metric you’ve explicitly promoted to decision-making status. It tracks progress toward a concrete business outcome, not just activity. Pageviews are a metric; weekly conversion rate vs target is a KPI. The point is focus: a small set of KPIs drives prioritization, alerts, and retros, while the long tail of metrics supports diagnosis.
How to define a solid KPI fast: pick a metric tied to value (revenue, leads, retention), define the segment and time window, set a target (baseline + ambition), and decide what happens when it’s off-track. Tie each KPI to a goal, an owner, and an action playbook.
Mini formula (simple but useful):
Conversion Rate = (Conversions / Sessions) × 100%.
If your KPI is “Weekly Conversion Rate ≥ 3.5%” and you had 420 conversions from 10,000 sessions, the rate is 4.2% — green for the week.
Typical KPI examples (with targets)
| KPI | Definition | Target idea |
|---|---|---|
| Conversion Rate | % of sessions that convert | ≥ 3.5% weekly |
| CTR | % of impressions that get a click | ≥ 2.0% per channel |
| CPA | Cost per acquired customer/lead | ≤ $25 for paid search |
| AOV | Average revenue per order | ≥ $62 monthly |
| Engagement Rate | % of engaged sessions | ≥ 58% sitewide |
Pragmatic tips: keep 3–7 KPIs per team, mix leading (e.g., CTR) and lagging (e.g., revenue), and audit definitions regularly to avoid Goodhart’s Law. Compare against a living benchmark and review attribution sensitivity—different attribution models can move KPI values without any real change in user behavior.
Why KPIs matter in web analytics
KPIs translate noisy behavioral data into action. They anchor experiments, channel optimization, and roadmap trade-offs, creating a feedback loop: objective → KPI → action → result → learning.