Most businesses I know don’t fail with KPIs because they don’t have them, but because they build a dashboard with twenty metrics that nobody looks at after the third week. Defining useful KPIs is an exercise in cutting back, not accumulating.
Start with the decision, not the metric
A KPI only makes sense if it changes what you’ll do on Monday. Before choosing any number, answer: what specific decision would I make if this indicator goes up or down 20%? If the answer is “none” or “I’d see what happens,” it’s not a KPI, it’s a context data point.
In a sports supplements ecommerce project, the team was measuring visits, bounce rate, time on page, and social media followers. None of those numbers triggered an action. By reducing the dashboard to two indicators—conversion rate by product category and gross margin per order—they began to detect that one product line was generating a lot of traffic and sales volume but almost zero margin. That finding didn’t show up with the previous dashboard.
Three KPI patterns depending on the type of business
1. Local business with few web visits
Here the dominant KPI isn’t traffic. According to Google Business Profile documentation, call and route request metrics are the ones that reflect real intent. An auto repair shop in a mid-sized city that was only monitoring web forms concluded that its website “wasn’t working.” By measuring calls from the Google listing over eight weeks, it discovered that it was receiving more phone contact than form submissions, but wasn’t recording it anywhere. The right KPI was calls answered and average response time, not sessions.
2. SaaS or subscription service
The classic mistake is focusing on sign-ups rather than retention. A 90-day retention KPI is usually more informative than the number of registrations. Practical limitation: if your customer volume is low (fewer than 30-50 monthly subscriptions), short-term retention fluctuates too much and can lead to wrong decisions. In that case, it’s better to measure quarterly cohorts or use an activity indicator (for example, percentage of accounts active in the last 14 days) instead of the pure churn rate.
3. Professional services and B2B
In consultative sales, the cycle is long and the monthly closing KPI can be misleading. A consulting firm that only measured proposals closed per month didn’t see that its problem was in the earlier stage: discovery meetings that never led to a proposal. The indicator they changed was “percentage of first meetings that generate a proposal within 15 days.” That number did depend on their internal process, and they could improve it.
The most common mistake: KPIs that depend on third parties
A KPI should measure something your team has the ability to act on. Measuring the organic CTR of a highly competitive keyword when your website has little authority is of little use: you can’t move it in the short term. It’s more useful to measure the evolution of positions across a range of your own keywords, or the percentage of pages that receive at least one impression in Search Console.
It’s also worth being wary of composite KPIs like “digital health score.” They sound good in a presentation and explain nothing when they get worse. If you need an index, document exactly what it weights and with what weights, or it will end up being a decorative number.
How to validate that a KPI works
- Realistic review frequency: if the KPI is updated daily but you look at it once a month, it’s unnecessary. Adjust the frequency to your decision cycle.
- Clear owner: each KPI should have a responsible person, even if shared. Without an owner, there’s no follow-up.
- Defined alert threshold: the number alone isn’t enough; define when it becomes a problem. For example, “if the conversion rate drops below X% for two consecutive weeks, we review the funnel.”
- Quarterly review of the KPI itself: indicators expire. What you were measuring a year ago may no longer be relevant if your business model has changed.
What you shouldn’t expect
No set of KPIs guarantees growth. KPIs are an alert and prioritization system, not a strategy. And there are situations where a formal dashboard is counterproductive: very small teams (two or three people) who already have a direct view of the business may function better with a weekly review of three numbers written by hand than with a poorly configured advanced analytics tool.
According to industry studies on analytics adoption, a significant portion of SMEs that implement metric dashboards stop using them in the first few months. The usual cause isn’t the tool, but having chosen indicators that weren’t connected to any operational decision.
A minimal routine that works
Choose three KPIs: one for acquisition (are people arriving?), one for conversion (are they doing what I want?), and one for efficiency (at what cost or with what margin?). Define them in writing with their exact formula, their data source, and their review frequency. Review the set once a week or once a month, depending on your cycle. And eliminate without guilt any metric that has gone through three reviews without triggering a conversation or an action. That cleanup exercise is, in practice, what separates a useful dashboard from an ornament.