The relative size factor test divides each payee's largest payment by its second-largest. A ratio near 1 means the biggest payment looks like the others. A ratio of 5 or 10 means one payment towers over the rest, which can signal a keying error, a padded invoice or a fraudulent payment hidden among a vendor's normal charges.
What does the relative size factor test measure?
The relative size factor test measures how far a payee's single largest payment stands above its next largest. The ratio is largest divided by second largest, computed separately for each vendor, employee or payee. It ignores the overall size of payments, so a small supplier and a large one are judged by their own history.
That is its strength. A 15,000 payment is unremarkable for a freight company billed 12,000 a month and alarming for a cleaning firm paid 600. A test on absolute amounts flags the first and misses the second. The ratio does the opposite, because it compares each payment only with others to the same payee.
The test finds one kind of problem well: a single payment out of line with a payee's normal pattern. Typical causes include an extra digit typed into an amount, an invoice inflated for a kickback, or a fraudster who borrows a real vendor's name for one large payment. It does not find many small thefts, because those keep the ratio low.
Computing the relative size factor in a spreadsheet
Computing the relative size factor needs a list of payments with a clean payee name and a positive amount on each row. Build a summary with one row per payee, find each payee's largest and second-largest payment, and divide. Any spreadsheet can do it in four columns.
Steps, assuming the payments sit in an Excel table named Pay with columns Payee and Amt:
- Normalize the names. Make sure each payee appears under one spelling, or the ratio will split one vendor into several.
- List the payees. On a new sheet, put
=UNIQUE(Pay[Payee])in A2. - Count and find the top two. In B2,
=COUNTIFS(Pay[Payee],A2). In C2,=MAXIFS(Pay[Amt],Pay[Payee],A2). In D2,=LARGE(FILTER(Pay[Amt],Pay[Payee]=A2),2). - Divide. In E2,
=IFERROR(C2/D2,"one payment"), then sort by E, largest first.
Excel's LARGE function returns an error when asked for a rank beyond the number of values, per Microsoft's LARGE function reference, so payees paid once show the label instead of a ratio.
Use only outgoing payments, and run refunds and credits as a separate list. Mixing them in can make a vendor's second-largest figure a refund, which distorts the ratio.
A worked relative size factor table for five payees
A worked table shows how the results read in practice. The example below covers one year of payments from an operating account. The payees and amounts are invented for illustration.
| Payee | Payments | Largest | Second largest | Ratio | Action |
|---|---|---|---|---|---|
| Ridge Freight | 31 | 18,900.00 | 2,310.00 | 8.18 | Pull invoice and approval |
| Northgate Consulting | 3 | 12,000.00 | 1,500.00 | 8.00 | Pull contract, check vendor setup |
| Atlas Tools | 9 | 3,950.00 | 3,120.00 | 1.27 | None |
| Harbor Paper Co | 24 | 1,140.00 | 1,085.00 | 1.05 | None |
| City Water | 12 | 486.20 | 471.90 | 1.03 | None |
Ridge Freight is the clear case: 31 payments, mostly around 2,000, and one of 18,900. That one payment is either a real special shipment, a keying error for 1,890.00, or something worse. The invoice and approval answer which.
Northgate Consulting has a similar ratio but only three payments, so the ratio says less. With so few data points, the more useful question is whether the vendor is real and was set up properly.
Atlas Tools, Harbor Paper and City Water have ratios close to 1, which is what steady billing looks like.
How should you read a high relative size factor?
Read a high relative size factor as a pointer to one document, not as a finding. There is no fixed cut-off. Sort by ratio, set a threshold that gives a list you can review in the time available, often somewhere between 2 and 5, and pull the invoice, approval and payment record for each flagged item.
For each flagged payment, check:
- The invoice. Does it exist, and does it match the amount paid?
- The approval. Who approved it, and was that within their limit?
- The bank record. Did it go to the vendor's usual account? A changed bank account on the one large payment is a serious sign. If the bank records for that period are missing, rebuild them first, as described in how to reconstruct bank records.
- The source document. If you are working from statements supplied by the client, confirm they are genuine before acting on what they show, using the checks in how to detect fake bank statements.
A confirmed bad payment then becomes part of a loss figure, built as set out in how to calculate fraud loss.
PreciseBankFlow extracts the payment lines from each statement with a confidence score on every amount, so the list feeding this test comes from every statement line and any doubtful figure is marked before you divide by it. Computing and judging the ratio stays with you. You can ask for early access.
Payees where the relative size factor test misleads
The relative size factor test misleads when a payee has very few payments, when its bills vary by design, or when the largest payment is a normal annual charge. In each case a high ratio is expected and says nothing about fraud. Note these payees and judge them by other tests.
Common cases:
- Two or three payments. With so little history, any variation gives a high ratio.
- Annual or quarterly bills. An insurer paid monthly plus one annual premium will always show a high ratio.
- Project vendors. A contractor paid a deposit and then a final balance has two very different amounts by design.
- Cards. Card statements mix many merchants under one issuer. Run the test by merchant, not by card issuer, and pair it with the review in employee credit card misuse.
- Many small thefts. A payee used for repeated small fraudulent payments keeps a low ratio. Duplicate tests, round-amount tests and a review of new vendors catch what this test cannot.