Esta página aún no está traducida al español. Ver las páginas en español · Ver en inglés

How to answer: “Walk me through how you would check whether a report's numbers are right.”

Whether you verify numbers independently — sources, samples, reconciliation and documented assumptions — instead of trusting a total that looks right.

The answer, step by step

  1. Start with the report definition: source tables, filters, time period.

  2. Trace a small sample of records back to the source.

  3. Reconcile totals with an independent figure (finance, the source system).

  4. Document assumptions and unresolved differences; give an example.

A strong sample answer

Benchmark answer for this question

Fictional example for learning only. It is not your experience; replace every detail with facts that are true for you.

I start with the report definition: which source tables it uses, which filters and which time period. Then I pick a small sample — usually ten to twenty records — and trace each one back to its source by hand. Next I reconcile the report's totals with an independent figure, like the finance system's monthly revenue, using a SQL query I write myself rather than the report's own logic. Any difference gets documented with the likely cause before anyone uses the report. For example, in my last role at a nonprofit in Chicago, I checked a donor report this way and found it counted refunded gifts. The donor total was $18,400 too high for the quarter. I flagged it with the three refunds as evidence, and the report was corrected before it went to the board. Careful data analysis means proving the number, not just producing it.

Why this example works

It uses independent checks, makes assumptions visible and proves the method with a concrete example.

  • Names a concrete method or tool: Report definition, a 10–20 record sample, an independent SQL reconciliation
  • Gives an example: “I checked a donor report this way and found it counted refunded gifts”
  • Specific, not general: “$18,400”, “ten to twenty records”, “three refunds”, Chicago
  • Claims are supported: The method is backed by the example
  • Connects to the role: Names SQL and data analysis

What you can learn from it

Name the data source, sample method, reconciliation and how you document unresolved differences.

Make it yours — with true details only

Use a real time you checked numbers — a budget, an invoice batch, a dashboard. Name only tools you have used (Excel, SQL, a BI tool).

From generic to evidence

Generic: “I would look at the numbers and see if they make sense.”
With evidence: “I'd trace a sample of 10–20 records to the source and reconcile totals with finance using my own SQL query; that's how I found a donor report counting refunded gifts.”

Mistakes that cost points

  • “I double-check the math” — Describe independent checks
  • Trusting the report's own logic — Reconcile with another source
  • No example — Give one time you found a problem
  • Hiding unresolved differences — Document them

Before you answer, check

  • Did I start with sources and filters?
  • Did I describe a sample check?
  • Did I reconcile with an independent figure?
  • Did I give an example?

Practise it in your own words

Write your version and get instant, explained feedback. Nothing is stored.

Practise this question

Related questions

All interview answers