Reconcile corporate card statements to expense claims

A match table of claimed lines, card lines without claims and claims without card lines, with cardholder summaries.

Choose one of three approaches. The steps for that option appear below.

Choose an option above. On a phone, swipe the tabs left or right to see all options →

ChatGPT: one chat matches and drafts the follow-up

ChatGPT matches the statement to the claims, summarises unclaimed spending per cardholder and drafts the follow-up list.

  1. You

    Get ready

  2. ChatGPT

    Match the lines

  3. You

    Check the coverage

  4. ChatGPT

    Summarise and follow up

  5. You

    Tie and approve

  1. Have ready: statement and claims exports

    Use: You

    The monthly card statement CSV and the approved-claims export.

    Check

    Check both files are complete and you may upload them; keep the originals.

  2. Match lines by cardholder, date and amount

    Use: ChatGPT

    Upload both files and ask for a match by cardholder, date and amount, using Python code that is shown.

    Details

    What to paste into the tool

    Task: Match table with unmatched lists. Match each card line to a claim by cardholder code, date and amount and show the code and the tolerance used. Return three lists: matched, card lines without claims, claims without card lines, each with the match basis. List unclaimed spending as facts only and make no judgment about personal use. Per-cardholder-code totals must tie to the statement total.

  3. Check the match coverage

    Use: You

    Check the claims export covers every cardholder on the statement and the three lists miss no cardholders.

    Check

    Verify the claims export covers all cardholders on the statement; missing people break the reconciliation.

  4. Summarise unclaimed spending and follow-ups

    Use: ChatGPT

    In the same chat, paste the checked match table and ask for a per-cardholder summary of unclaimed spending and a follow-up list.

    Details

    What to paste into the tool

    Task: Draft match table: claimed-and-matched lines, card lines without claims, claims without card lines, with cardholder-level summaries. Summarise each cardholder code's matched total, unmatched card total and unmatched claim total and add a follow-up list of unclaimed card lines. Check that each cardholder code's totals add to the statement total. List unclaimed spending as facts only and make no judgment about personal use. Per-cardholder-code totals must tie to the statement total.

  5. Tie the totals, then approve

    Use: You

    Tie each cardholder total to the statement total, check every card line and claim is matched or listed, confirm the follow-up wording, then approve.

    Check

    Check the follow-up list; contacting cardholders happens outside this workflow. Totals are tied to the statement. The finance reviewer gives the final approval; nothing leaves this workflow without it.

Before you paste into any AI tool, remove any confidential information. Put it back at the end.

Tools in this workflow

Deciding between ChatGPT and Claude for this? See the comparison.

Final check

Never paste credentials or full payment details.

Verify the claims export covers all cardholders on the statement; missing people break the reconciliation.

Review the follow-up list; contacting cardholders happens outside this workflow.

Every card line and every claim in scope is matched or listed; per-cardholder totals reconcile to the statement; the finance reviewer confirms the claim-export completeness.

Check the result against your original inputs. Approve it before sharing or acting.

What still needs checking
  • Whether the organisation allows card statements and claims exports to be uploaded to each tool is not confirmed.
  • Matching tolerances for date and amount differences are for the reviewer to set; no page read here states them for any tool.

Share this page