Entry · 2 weeks
When two reports give two figures, nobody trusts either.
The metrics audit takes two weeks to leave the ten metrics your leadership decides with under one definition, with a business owner, a technical owner, and reconciling across systems. It's the entry point to everything else we do.
What happens when a metric has two values.
Neither gets defended: trust in both is gone. Leadership stops looking at that KPI, the hunt for someone to blame begins, and each department uses whichever figure suits it. You can live with that for a long time. What you can't do is build anything on top of it, least of all an AI project.
What happens in the two weeks.
- Together with leadership we pick the ten metrics you actually decide with: revenue, CAC, LTV, churn, margin, whichever are yours.
- A forensic audit of each one: order of magnitude, the same formula cross-checked between dashboards and systems, a sample of data types and the integrity of every join. The errors that cost the most are the ones that raise no alarm.
- One definition per metric, in writing, with its business owner — who decides what it means — and its technical owner — who answers for how it's calculated.
- A new view: something about your business you can't see today, such as where sales leak out of the funnel.
The guarantee.
You end with your ten metrics under one definition, two owners and reconciling, or we keep working at no cost until they do.
The only condition is the one that makes the outcome possible: you name the owners and they attend the sessions. Without owners, no definition lasts.
What you keep.
- The dictionary of your ten metrics: technical formula, who uses it and for which decision, frequency, criticality and what it depends on.
- The list of errors found, with their impact and the order to fix them in.
- A clear verdict on which data or AI initiatives are viable today, which need foundations and which should be dropped.
- And if you want it to stay that way, the option of the monthly fee. If not, you take everything with you and whoever you choose delivers it.
Frequently asked questions
- Why don't metrics reconcile between departments?
- Almost never bad faith: it's the gap between the business definition and its technical implementation. The KPI gets defined, the correct query turns out hard and the easy one gets built; or each department calculates with its own date, exchange rate or idea of an active customer. Without an owner to stop it, versions multiply.
- How long does it take to define a company's KPIs?
- Two weeks are enough for ten metrics if there's a decent database, dashboards already in use and involvement from both business and the technical team. From scratch, with no data architecture, it takes longer. Ten metrics are enough to start; thirty is the sensible ceiling.
- What is a KPI dictionary?
- A living document with every leadership metric and, for each: the technical formula, who uses it and for which decision, how often it refreshes, how critical it is, which tables it depends on and its two owners. It dies when nobody maintains it, which is why it's tied to the monthly fee.
- What if our metrics are already fine?
- Then the audit confirms it in writing, which is worth having, and you still get the new view of your business. The guarantee is on the outcome: you end with your ten metrics defined, owned and reconciling.