No matching tool is right on every list, and I do not publish an accuracy score I graded myself: a score from my test lists says little about yours. This page covers what the add-in guarantees by design, where it gets it wrong, and when another tool is the better choice.
To see it on real rows, the XLOOKUP vs AI Fuzzy Match example on the product page matches one customer list both ways: XLOOKUP returns #N/A on every renamed or misspelled name, and AI Fuzzy Match finds them and answers No match for the one customer that is not in the list.
What it guarantees by design #
- Answers come only from your list. Every match is an entry from List B, or No match. It cannot invent a name that is not in your list.
- A confidence and a status on every row. Each AI answer comes with a confidence. Every row gets one of five statuses: Matched, Review, No match, Exact or Duplicate in List B. Duplicate in List B means the matched name is in List B more than once, so the copied columns stay empty and you pick the right row yourself.
- Review below your threshold. Every answer with a confidence below your threshold (80% unless you change it) is marked Review (with a light fill by default), and Show review rows filters the sheet down to those rows.
None of this makes every answer right. It makes every answer checkable, with the doubtful ones marked. See Review results and set the threshold.
Where it gets it wrong #
- Renames and acronyms that share no letters with the name. FACEBK for Meta Platforms is an example. When List B has up to about 200 entries, every row is compared with all of it. When List B is longer, each row is offered only the 20 entries closest in spelling, so an entry that shares no letters is not among them and the row comes back No match. Check the No match rows on a long list before you treat them as new vendors, or run Search thoroughly on them, which asks about every List B entry. Work with big lists explains the shortlist.
- Parent and subsidiary look-alikes. A parent company and its subsidiary with near-identical names can be matched as the same company, sometimes with a high confidence, so the row shows as Matched. Spot-check Matched rows wherever your list holds several legal entities of one group. Raising the threshold to 90% sends more of these rows to Review, and so does the extra rule “A subsidiary is not the same as its parent company”.
When another tool is better #
- Both lists share a clean key, such as a customer number or a VAT ID: use XLOOKUP. No fuzzy matching needed.
- The differences are mostly typos and small spelling changes: Microsoft’s free Fuzzy Lookup add-in or Power Query’s fuzzy merge handle those well, for free.
- No internet, or data that must not leave your computer: Power Query’s fuzzy merge runs offline (Fuzzy Lookup too, on Excel for Windows only).
AI Fuzzy Match is for the rest: abbreviations, renames, bank descriptions and look-alike names where a wrong pair costs money. The fairest test is your own list, and the first 100 credits are free for that.