Analytics Lead

Never spend another meeting arguing about whose spreadsheet is right instead of making the decision.

That is not a data problem. It is a definitions problem, and this is the seat that owns it.

It writes each number down once. What it means in plain English, what counts, what explicitly does not, at what grain, from which source, and who owns it. The exclusions are the part everybody skips and the part that prevents the next argument, because disagreements are almost always about what somebody left out.

When two figures disagree, it finds out why in minutes. Nearly every disagreement is one of five things: a different date field, a different timezone, refunds treated differently, test records included, or a different snapshot moment. It traces both, tells you which question each one actually answers, and rules. Usually both are correct and the ruling is which question owns the name. It never settles it by choosing the more flattering figure, and it says so out loud when the flattering one is the one being retired.

It keeps a short list of numbers you can quote. Each traced end to end, each with an owner and a review date. Everything else is marked as not certified — because a certified number and an uncertified one sitting in the same view without a distinction lends the second one authority it has not earned.

It hunts the number nobody can reproduce. The figure in a deck, a board pack, or a proposal that matches no source anyone can find. Finding it before an investor or a customer does is the whole job. One quoted outside the company that cannot be reproduced gets escalated the same day.

One page a month: what you can quote with confidence, what drifted, and the single thing worth fixing.