---
title: "Start with the KPI, not the tool"
date: 2026-11-18T13:05:00+01:00
author: Nel Claes
canonical_url: "https://www.xploregroup.be/en/insights/start-with-the-kpi-not-the-tool"
section: Insights
---






# Start with the KPI, not the tool

## Data &amp; AI

![Data AI blog](https://cdn.craft.cloud/5fa59528-9049-4864-a4f2-a0667591b3bd/assets/img/DataAI-blog.jpg)
## Start with the KPI, not the tool

Ask a data team what their platform delivers and you usually get an answer about the platform itself. How many resources are connected. How many dashboards are running. How many colleagues log in. Then ask which company KPI has changed because of it and the conversation often falls silent.

That is rarely due to the data team. It is due to the order in which data and AI projects are set up in many organizations.

## How it usually goes

The pattern is recognizable. Budget is freed up for a data platform or a first AI application. A tool is chosen. Then someone looks for use cases that fit that tool. And a year later, the CFO asks the question that was originally asked: what has this yielded?

At that point, it's difficult to give an honest answer. There is no baseline, no one has predetermined which outcome should change, and the organization sees the platform as an IT project.

With AI, that pattern becomes stronger, not weaker. Setting up a pilot is relatively quick and inexpensive. As a result, an organization can easily run five pilots simultaneously, without one being linked to a number that keeps management up on the night.

## Reverse the order

The solution is not technical. It lies in the questions you ask before you even talk about tools. These five provide guidance.

### Which KPI needs to move and by how much?

Be specific. "Working more data-driven" is not a KPI. Inventory costs, quote turnaround time, or customer churn are. Also agree on how much improvement makes the process worthwhile.

### Who owns that KPI?

It's almost never the data team. It's the sales director, the supply chain manager, or the marketing manager. Without that owner at the table, the project remains an IT exercise.

### Which decision influences that KPI?

Data only adds value if someone makes a different decision through it. Which decision is it, who makes it, and how often?

### What data do you need for that, and is it good enough?

Only here does the technology come into play. Often it turns out you need less than you thought, or that the basic data needs to be sorted out first.

### How do you measure the difference?

Set a baseline measurement before you start. Otherwise, "it works better" remains a feeling instead of a result.

## What this changes in practice

Those who start like this notice a few things. You have fewer use cases, but they are better. Stopping pilots becomes easier because you see which ones don't hit KPIs. The choice of platform or tool follows what the decision needs, not the other way around.

It also doesn't mean you have to throw away existing investments. A platform that already exists can be a good foundation. The difference is that from now on you use it for a result you have pre-specified.

## From strategy to execution, without the KPI disappearing along the way

In many projects, things go wrong at the transition between advice and execution. One team outlines the strategy, another builds, and the original demand gets lost along the way.

Within Xplore Group, the teams for data strategy, data engineering, analytics, and AI work together on the same path. Each competence center brings its own specialization, from customer data and marketing analytics to data platforms, forecasting, and AI activation. The KPI we start with is also the KPI on which we assess the results.

## Where do you start?

Include ongoing data and AI initiatives and write the KPI next to each project that needs to move it, with the owner's name. If that doesn't work out for an initiative, you immediately know where to start.


