Right-only keys are missing
Build the key population from both source lists.
Step-by-step guide · read-only ·
Loading your interactive spreadsheet…
Preparing the worksheet and formulas…
Enter one formula in G5. Return each nonblank ID occurring in A2:A9 or D2:D11 once, uppercased, in first-seen order across A then D. Report its signed numeric total from B, total from E and A minus B. Match printable ASCII IDs literally ignoring case; spaces and leading zeroes remain significant. Repeated IDs are aggregated, missing-side totals are numeric zero, and zero-net keys remain. If both lists are empty, return No keys.
Follow the highlighted cells. Switch to Your turn to edit. Answer cells: G5:J22. Enter one formula in G5; keep the rest of the output area clear. Browser calculations may differ from your Excel version.
Practice stays on this device. No worksheet data is uploaded.
Keep the union of both ID lists, combine signed repeated rows on each side, then report both totals and their difference.
Keep the union of both ID lists, combine signed repeated rows on each side, then report both totals and their difference.
=LET(joined,VSTACK(A2:A9,D2:D11),keys,UNIQUE(FILTER(UPPER(joined),joined<>"","")),booked,MAP(keys,LAMBDA(key,SUMPRODUCT((UPPER(A2:A9)=key)*B2:B9))),received,MAP(keys,LAMBDA(key,SUMPRODUCT((UPPER(D2:D11)=key)*E2:E11))),IF(SUMPRODUCT(--(joined<>""))=0,"No keys",HSTACK(keys,booked,received,booked-received)))| Normalized ID | Total A | Total B | A minus B |
|---|---|---|---|
| EAST | 125 | 125 | 0 |
| WEST | 50 | 50 | 0 |
| NORTH | -10 | -10 | 0 |
| ONLY-A | 7 | 0 | 7 |
| ONLY-B | 0 | 20 | -20 |
The remaining reserved output cells are blank. Keep that area clear when trying the formula.
The formula uses the interactive worksheet above. Adjust ranges and criteria for your own data.
The original report includes EAST, WEST, NORTH, ONLY-A and ONLY-B. EAST totals 125 on each side. ONLY-A keeps 7 and zero; ONLY-B keeps zero and 20, producing -20 in the residual column.
Rows of 10 and -10 total zero, but their ID still exists in the source. Keep that key, both totals and the residual. Filtering on a nonzero amount would discard evidence.
The report follows normalized IDs encountered in A first, then new IDs from D. It does not use a locale-dependent alphabetic sort. Reordering the sources may change report order while leaving each keyed total unchanged.
Each side permits blank rows. A blank ID may have a blank or numeric-zero amount; other amounts without an ID are left ungraded. Nonblank IDs require signed numeric cents within ±1,000,000. The 18-by-four area fits every distinct key from both declared ranges.
Two exports can have the same per-ID total and different individual transactions. Use row-level source checks when transaction identity matters. This lesson reports grouped differences without deleting or repairing the original sources.
VSTACK puts the eight A IDs above the ten D IDs. Keeping both sides prevents right-only records from disappearing.
Filter empty slots, uppercase the text and retain the first occurrence of every key. The resulting order follows A, then new keys from D.
For each key, SUMPRODUCT combines every matching signed amount in B. A missing key receives numeric zero.
Use the same key population and identity policy for D:E. Repeated receipts and credits remain in the total.
HSTACK returns key, total A, total B and A minus B. A zero residual does not hide the source totals or prove that individual rows match.
When both ID lists contain only empty slots, return No keys in one cell. All other positions in the maximum report area stay blank.
Build the key population from both source lists.
Choose keys from ID presence, not amount tests.
Apply the same literal, case-normalized key rule on both sides.
These original single-anchor tasks use a declared ASCII ID policy, signed numeric cents and bounded source rows. Browser calculations and independent rules are checked separately from native Excel/Sheets recalculation, editing and save/reopen.
Content version 1. Practice checks compare calculated results across different datasets; they do not certify every possible formula.