Matched, partially matched, and not matched are the three buckets that turn month-end reconciliation from variance theater into owned work. Month-end doesn’t fail because you lack rows. It fails because exceptions aren’t classified fast enough for the audit room.
Why do three buckets beat a sea of yellow cells?
- Matched — settle and move on.
- Partially matched — investigate with a reason, not a vibe.
- Not matched — true gaps, both sides visible.
When every row lands in one of those three, the conversation changes from “where’s the variance?” to “which exceptions matter before we close?” That is the language controllers already use. Software should speak it — not invent a fourth dashboard that nobody opens on close week.
Who owns exceptions before Day −1?
Taxonomy without ownership is just a prettier backlog. Name the owner for Partials and Not Matched before the calendar hits Day −2. Publish the rule: Matched is closed; Partial needs a reason and a next action; Not Matched is a true gap until evidence says otherwise.
Composite pattern you’ll recognize: P&L still prints black while operating cash and AR tell a different story. The recon that surfaces classified exceptions early changes the owner call from surprise to agenda.
How should software enforce the language?
Export classified Excel with variance summaries. Reserve exact matches globally before fuzzy scoring. Leave unmatched rows honest — don’t force a story. Standardize the output, then the close gets shorter.
ReconCore is built around that taxonomy: upload, map, download Matched / Partially Matched / Not Matched, purge files within 24 hours. If your team still argues about yellow cells on Day −1, standardize the buckets first — then buy the tool that already speaks them.