Classic scene: a well formulated kpi … which is going wrong
You manage a marketing team, with the aim of increasing the share of turnover from existing customers. You assign a simple kpi: “Make 25 % of the monthly turnover via email.”
The intention is good. The canal is profitable. You want to push the retention. But a few weeks later, the first on -board effects appear:
- Increase in sending volume → saturation of hearings
- Falling opening rates
- Pic d’Unsubscribes
- Tensions with the product team who does not have the stock to follow the push
And above all: the share of the Email CA increases … only because the acquisition is late.
The KPI is reached, but it signals a false success and feeds a counterproductive spiral.
The real problem: the very structure of the indicator
A KPI of ratio (ex: % of turnover) can be handled without creating growth.
A volume KPI (ex: Number of shipments) rewards activity, not efficiency.
And an isolated kpi (email, stock, conversion) forgets the transverse arbitrations.
You don’t want a reachable kpi but a KPI that pushes to make the right decisions.
Another case: the isolated objective that parasites the strategy
Objective assigned to the Product team: Set out dormant stocks.
Indicator: “Reduce unsold 30 % in the quarter.”
The team therefore pushes these products in Homepage.
But these references have no more traction.
The campaigns cannibalize them at the expense of new products.
Conversion drops. The average basket retreats.
The KPI is respected, but it cost more than it reported.
A simple grid to avoid errors: input / output / outcome
| Kind | CRM example | Control | Relevance |
|---|---|---|---|
| Input | Number of emails sent | Pupil | Weak |
| Output | Income generated via UTM Email | AVERAGE | Strong |
| Outcome | Reseinder rate at 60 days | Weak | Very strong |
Input = what the team produces (useful as prerequisites)
Output = what the team triggers (controlled, measurable)
Outcome = what the company really wants to achieve (sustainable, strategic)
We pilot with outputs.
We follow outcomes.
And we demand inputs as a working basis, not as a goal.
Who to assign a collective kpi?
Some metrics are transverse by nature:
-
- Repeat returned depends on CRM, product, delivery, support.
- The conversion rate depends on the site, traffic, offer.
Should we avoid assigning them? No. But you have to:
-
- Cross the overall kpi with a kpi channel or operational
- Share the visibility of KPI, even if only one team carries it
- Organize coordination mechanisms, no silos
The approach to → from to intelligently structure a kpi
It is the simplest, but the most powerful approach:
“Make from 8 % to 12 % reachant rate to 90 days.”
“Increase the average value per e-mail from € 0.70 to € 1.20.”
This allows:
-
- To frame the ambition (Baseline Claire)
- To anchor the objective in a dynamic (progression, not isolated figure)
- Avoid sterile discussions of the type: “€ 10 million in email income, is that good or not?”
Toxic anti-kpi checklist
| ❌ Bad kpi | ✅ Robust kpi |
|---|---|
| “25 % of turnover via email” | “Go from 8 to 10 M € of CA utm email” |
| “5 %conversion” | “Improve 7 %income/user” |
| “Reduce stocks by 30 %” | “Set 10 references without degrading the average basket” |
| “Launch 40 creative concepts” | “40 concepts, with a 20 % min hit failure” |
What a kpi must always make it possible to do
-
- Pilot in the right direction (alignment with business strategy)
- Be understood in context (absolute growth, not flattering ratio)
- Encourage cooperation (avoid friction between teams)
- Be achieved without perverse effects (no local optimization games)
Conclusion
A good KPI is not content to be measurable. It must be useful, readable, contextualized. And above all, it must promote good behavior without producing distortions or tensions.
Always ask this question:
“If this KPI is reached, am I sure that the company is in a better situation?”
