Finance Index
How do you detect card drift in your spend data?
Reference guide to detecting card drift spend data, including card controls, policy design, employee spend workflows, receipt capture, and reconciliation.
Card drift announces itself in patterns: vendors appearing in both AP and card data, recurring same-merchant monthly charges, rising average card transaction size, and card merchants that obviously issue invoices - software companies, contractors, equipment suppliers. Detection is a join between two datasets most companies never connect: the AP vendor master and the card transaction file.
At a Glance
| Aspect | Short Answer | Why It Matters |
|---|---|---|
| How do you detect card | Card drift announces itself in patterns: vendors appearing in both AP and card data, recurring same-merchant monthly charges, rising average card transaction size, and card merchants that obviously issue invoices - software companies, contractors, equipment suppliers. | Keeps vendor records and payment decisions reliable. |
| Card control | Work through six: (1) **Vendor overlap** - the same supplier paid through AP and charged to cards (match on normalized merchant name against the vendor master). | Reduces payment errors, timing issues, and reconciliation cleanup. |
| Quantify drift for leadership | Build one metric: **invoice-eligible card spend as a percentage of total card spend**, where invoice-eligible means recurring charges plus AP-overlapping vendors plus invoice-capable merchant categories above a dollar floor. | Keeps vendor records and payment decisions reliable. |
| Vendor impact | Normalize card merchant descriptors (strip the "SQ *"/"PAYPAL *" prefixes and location suffixes), then fuzzy-match against the AP vendor master on name. | Keeps vendor records and payment decisions reliable. |
| Spend control | Large amounts (anything that would have required an approval as a requisition), recurring monthly charges, B2B merchants, and any purchase that created a contract or renewal obligation. | Keeps work moving without losing accountability. |
What are the specific signals that invoice-able spend is moving to cards?
Work through six: (1) Vendor overlap - the same supplier paid through AP and charged to cards (match on normalized merchant name against the vendor master). (2) Recurring charges - same merchant, same approximate amount, monthly cadence; that's a contract being paid without a contract record. (3) Rising average transaction size - drift pulls bigger, considered purchases onto a rail built for small ones. (4) Invoice-capable merchant categories - B2B software, professional services, wholesale trade MCCs on expense cards. (5) Near-threshold clustering - transactions bunched just under the pre-approval or receipt threshold. (6) Statement-cycle spikes - month-end swipes that look like deadline spending against expiring budgets or limits.
How do you quantify drift for leadership?
Build one metric: invoice-eligible card spend as a percentage of total card spend, where invoice-eligible means recurring charges plus AP-overlapping vendors plus invoice-capable merchant categories above a dollar floor. Report the percentage and the dollar value quarterly, with the top ten drifted vendors named. A trend line does more than any policy memo: leadership funds fixes for numbers that grow.
How do I find vendors that appear in both AP and card data?
Normalize card merchant descriptors (strip the "SQ *"/"PAYPAL *" prefixes and location suffixes), then fuzzy-match against the AP vendor master on name. Expect a meaningful match list the first time you run it - most companies have never looked.
What transaction patterns suggest a purchase should have been an invoice?
Large amounts (anything that would have required an approval as a requisition), recurring monthly charges, B2B merchants, and any purchase that created a contract or renewal obligation. The simplest screen: sort card transactions descending by amount and ask, for each of the top fifty, "would I have wanted to approve this before it happened?"
How do I audit recurring card charges to find unrecorded contracts?
Cluster card transactions by merchant with monthly or annual cadence, then try to locate each one's contract in your systems. Every recurring charge with no findable contract is a commitment with no owner, no renewal date, and no negotiated terms - drift's most expensive form.
We found $400k/year of card charges to a vendor we also have a negotiated contract with - how do we migrate it back?
Confirm with the vendor what the card charges actually bought and whether contract pricing applied (it usually didn't); consolidate the spend under the contract, set the vendor up for invoicing through AP, and merchant-block the vendor on expense cards once the invoice path is live. Then announce why - visible recovered savings is the best drift-prevention message you'll ever send.
Can AI flag drift automatically - classifying transactions as legitimately-card vs should-be-invoice?
Yes - this is a classification problem AI handles well: merchant category, recurrence, amount, vendor-master overlap, and contract existence are all learnable features. The honest caveat is that the classifier is only as good as the connected data; AI watching a card feed in isolation can flag patterns, but AI that can see AP, contracts, and procurement can flag *drift*.
Stampli perspective
Because Stampli runs procurement, AP, payments, and cards in one platform, card spend and invoice spend are natively comparable - same vendors, same GL structure, same workflows - which is the data precondition for seeing drift at all. Stampli AI supports coding and anomaly surfacing across that combined picture, and cards issued from approved requests mean new card spend arrives classified by intent instead of needing forensic reconstruction.