How can I check that an imported roster has one row per person?

Use a stable person identifier, such as an employee or member ID, rather than names alone. Names can be repeated or formatted differently, so they are poor evidence that two rows represent the same person.

Make three comparisons: total imported rows, number of distinct person IDs, and the expected set of IDs from the roster you intended to import. If every expected person should appear once, the total and distinct counts should match the expected headcount, and the IDs should match the expected set. Matching totals alone is not enough: one missing person and one extra row can cancel each other out.

Keep an untouched copy of the import while checking it, so investigation does not change the evidence you are comparing.

Group or summarize the imported rows by the person ID and count the rows in each group. Any ID with a count greater than one needs review; it may be a duplicated record, or there may be a legitimate reason that person has multiple entries. Power Query supports grouping by columns and applying a Count rows operation [S1].

For example, suppose an illustrative import has 102 rows and grouping produces 100 distinct IDs, with ID-17 appearing three times and ID-42 appearing once. The result flags ID-17 for investigation; it does not prove which row, if any, should be removed. Compare the flagged rows’ other fields and the source records before deciding what to do. Do not delete repeats automatically if the roster’s rules allow multiple records per person.

Compare the imported ID list against an authoritative expected roster for the same reporting period. Write down IDs expected but absent from the import, plus any imported IDs not on the expected list. Those are separate exceptions to explain; a matching row total cannot show that the identities match.

If you use Power Query’s column distribution profile to inspect repeated or distinct identifier values, check its coverage setting: the profile is limited to the first 1,000 rows by default, and can be changed to the entire dataset [S2]. A profile count is useful for inspection, but it does not identify absent expected people. For that, you still need the expected ID list and a direct comparison. Record the comparison date and roster version so the result is interpretable later.

Use this sequence before treating the import as reconciled:

1. Preserve a copy of the original import and note the expected roster’s version or date.
2. Confirm the person ID field is populated and uses a consistent format; investigate blanks or apparent formatting differences rather than silently changing IDs.
3. Count imported rows and group by person ID to identify IDs with more than one row.
4. Compare imported IDs with the expected ID list; record missing and unexpected IDs separately.
5. Review exceptions against the source records and document the decision for each; keep unresolved items visible.
6. Re-run the counts and comparison after any corrections, and report the scope and date checked.

Copyable report (replace placeholders): Roster checked: [name/version/date]. Imported rows: [number]. Distinct IDs: [number]. Repeated IDs: [number/list]. Expected IDs missing: [number/list]. Unexpected IDs: [number/list]. Unresolved exceptions: [number/list]. Status: [reconciled / exceptions remain].

This check can expose count and identifier discrepancies, but it cannot establish that every other field is accurate or that an exception is an error. Resolve those questions with the roster owner or authoritative source.