An interviewer says - we close an acquisition in six weeks. Both companies already report active merchants and monthly revenue to the board, the acquired company has its own glossary and a fiscal calendar starting in February, and the first combined board pack is due 30 days after close. What do you do, and what do you deliberately leave diverged?

What the interviewer is testing

Whether you can scope semantic alignment to the decisions that actually depend on it. The failing answer is a harmonisation programme. There are two warehouses, two glossaries and six weeks, and the only numbers with a hard deadline are the eight to twelve on the board pack. Everything else can stay different for a year without anyone being harmed.

The clarifying questions that change the answer

• Is the acquired business a separate reporting segment after close? If it is, the board sees two columns and a total, and you need a mapping, not a merge.
• Which metrics are in the board pack, and who signs them? That list is the scope.
• Does anyone get paid on a combined number? Commission and earn-out metrics are the ones where a definition difference becomes a legal dispute rather than a footnote.
• What is the audited fiscal calendar after close? A calendar change restates history.

The thirty-day plan

1. Freeze the list. Name the board-pack metrics and nothing else. Write each side's current definition next to the other in one table.
2. Classify every difference as a definition delta or a data delta. "Active" counted over 28 days against a calendar month is a definition delta and is settled by a decision. A merchant present in one system and missing from the other is a data delta and is settled by investigation. Teams conflate these and spend weeks reconciling a number that was never going to match.
3. Build a bridge, not a merge. One table per metric: their number, our number, and a named line for each difference with its size. A board pack with a bridge is defensible. A single blended number with no bridge is not, and the first auditor question kills it.
4. Treat the fiscal calendar as reference data with effective dates, so a report run next year for a period before close still uses February.
5. Leave the acquired warehouse in production. Point the bridge at it.

What to leave diverged

Everything not on the list. Churn definitions, funnel stages, product taxonomies, support categories. Align a definition only when one decision is made on the combined number; until then two definitions cost nothing but a mapping entry. Expect to carry dual reporting for two to four quarters, and say so to the board on day one so the second number is not read as an error.

Common weak answers

• "Run a joint glossary workshop." Six weeks, two companies, no shared authority. The workshop produces a document and the board pack still needs numbers.
• "Migrate them onto our platform first." A platform migration is quarters of work and it does not change a single definition. The semantics are the deliverable.
• "Pick our definitions because we are the acquirer." Defensible for internal metrics and dangerous for anything in their existing external reporting, which was audited as stated.

What a strong answer adds

Who decides. Name one person per metric with authority to rule, with a 48-hour escalation to the CFO, because the cost here is decision latency and not engineering effort. Add the reverse test: for each metric, what evidence would make you adopt their definition instead of yours? A team that cannot answer that is doing politics, not reconciliation.