Skip to content

← All work

CS-05 · 2025 – 2026

Proving a sales metric meant something different from its documentation

A key qualification flag was documented one way and behaved another. I showed it with data, then made the contradiction something the business could decide.

~42%
of active prospects carried the qualification flag
Measured. Its documented meaning assumed full coverage.
From a doubt to a decision
  1. Notice the mismatchDocumented meaning vs how the flag behaved
  2. Check it against the dataReconcile the definition with actual rows
  3. Quantify the gap~42% of active prospects covered
  4. Put it to the ownerEscalated as options
  5. Apply it consistentlyAcross the warehouse and reports

The method is the point. Evidence first, then a decision the owner can actually make, then consistent application.

The problem

A key qualification flag was documented as meaning one thing, but it behaved like another. On top of that, early-funnel definitions (how the business decides a lead has become a real prospect) were inconsistent.

That is the kind of problem that looks small and does a lot of damage. A metric whose written meaning differs from its real behaviour can’t be trusted, and everything downstream of it inherits the doubt.

What I did

  • Reconciled the documentation with reality. Instead of trusting the written definition or the loudest opinion, I compared what the flag was documented to mean with what it actually did across the data.
  • Quantified the gap. The flag covered only about 42% of active prospects, where the documented meaning assumed it applied to all of them. Measured.
  • Facilitated alignment on definitions with the people who owned them, so the fix would be a shared decision rather than my interpretation.
  • Drove consistent application of the agreed definition across the warehouse views and the downstream reports. Self-reported: standardised across the warehouse and 8+ reports.
  • Escalated the contradiction as a decision. I laid out options for the owner, rather than quietly choosing a meaning myself.

My role

I owned a workstream inside a larger data-quality initiative led by someone else. My part was the analysis, the definition alignment and the consistent rollout across the data layer and reports.

Architecture and trade-offs

  • Evidence over argument. A definition dispute is easy to lose on opinion. A measured coverage gap is much harder to wave away.
  • Decisions belong to the owner. The data could show that the documented meaning was wrong. It could not say which meaning the business wanted. Framing that as options kept me on the right side of that line.
  • Consistency before cleverness. Fixing one report would have made the others look wrong. The value came from applying one definition everywhere, including the places nobody had complained about yet.

Impact

  • Found that the flag covered about 42% of active prospects against the full coverage it was assumed to have. Measured.
  • Standardised the definition across the warehouse and 8+ reports. Self-reported.
  • A contradiction that had been quietly producing inconsistent numbers became an explicit, owned decision.

The hardest part

Proving, with data, that the documented meaning was wrong, and then making the problem decidable. Finding the gap was the easier half. The harder half was presenting it so the business owner could choose between clear options instead of defending a document.

Skills shown

Data governance, analytical rigour, SQL, stakeholder facilitation, metric definition and consistency.