← Insights

Why two teams report different numbers

It is almost never a broken tool. It is that nobody ever wrote down what the metric means, so each team defined it plausibly and differently.

Sarthak Jain, Co-founder · · 5 min read

Sales says the number is 412. Marketing says it is 380. Finance says both are wrong. Everyone assumes a system is broken, and a week disappears into reconciling dashboards.

The systems are usually fine. The definition never existed.

Every metric hides four decisions

Take something apparently simple, like "leads this month". It silently encodes:

  • What counts as one? A form submission, a person, or a company? Two people from the same account are one lead or two, and both answers are defensible.
  • When is it counted? At submission, at qualification, or at the point it entered the CRM? Month-end sits differently under each.
  • Whose timezone? A month boundary in IST and in UTC are five and a half hours apart, which is enough to move a number that gets reported to a board.
  • What is excluded? Test entries, internal domains, spam, duplicates, and the ones the team manually disqualified.

Four decisions, each reasonable, and two teams making them independently will differ by ten per cent without either being wrong.

Write the definition next to the number

The fix is not another dashboard. It is a definition that lives where the number is displayed - what counts, when it is counted, what is excluded, and who owns the answer when it is disputed.

Once it is written, the argument changes from "your number is wrong" to "I think the definition should exclude X", which is a conversation that can actually be resolved.

One source, many views

The second half of the fix is structural. Each metric gets one computation, in one place, and every dashboard reads from it.

The moment the same metric is computed in the BI tool, in a spreadsheet, and in a product query, they will drift. Not because anyone was careless — because someone will correct one of them and not the others.

Instrument the definition, not the screen

A practical consequence: put the logic in the data layer, not in the chart. If "qualified lead" is a filter someone drags into a dashboard widget, that definition exists only inside that widget. If it is a modelled field, everything downstream inherits it, and correcting it corrects everything at once.

The test

Ask three people in different teams to produce the same number independently. If they cannot, you do not have a reporting problem. You have an undefined metric, and no amount of tooling fixes that.

If this describes your situation, start with a baseline.

We document where you stand before recommending anything - so whatever we claim at 90 days has a before to be measured against.