Short answer: Platform fees are not a line item you set once. They creep — new card types, promo commissions, app subscriptions, payout timing — and most owners only notice when the bank deposit is short.
The bleed you do not budget for
Terrence runs a small product brand and sells through three rails: his Shopify store, Amazon Seller Central, and occasional wholesale invoiced in QuickBooks. Each dashboard shows green. Each takes a different slice. None of them talk to each other.
In January his bookkeeper flagged that gross margin looked healthy but operating cash was tight. Terrence blamed inventory. Then he stacked exports:
- Shopify Payments fees on the Stripe balance report
- Amazon referral + FBA fees on the settlement report
- A Shopify app billing $149/month he forgot he installed during a Black Friday panic
Individually, each fee is explainable. Together they were nearly 11% of revenue on a month with no big ad spend. The P&L still used last year's "roughly 8% all-in" mental model.
Platform tax is opex. It behaves like rent. You do not feel it until you reconcile.
We wrote a longer intel pass on this pattern: Platform tax is opex — and most owners never see the bleed. This post is the short version for operators who close books monthly, not analysts who live in spreadsheets.
Three ways fees hide
1. Aggregated processor views. Stripe's home screen shows net volume. It does not foot to your QBO deposit when refunds post late or payouts batch across month-end.
2. Stacked platforms. Shopify subscription + payment processing + apps + ads. Amazon referral + fulfillment + storage. Each bill arrives on a different cadence. Your brain averages; your cash does not.
3. Rate drift on the same stack. Domestic cards at 2.9% last year. International cards, BNPL, and chargebacks this year. If books still use a flat estimate, variance shows up as "mystery shrink" in Undeposited Funds.
Terrence's bookkeeper was not bad at her job. She was missing export-level drift — the kind you catch when gross minus fees does not equal net in the actual CSV.
What Terrence did (bounded, not theatrical)
He did not buy an AI operating system. He did not grant a stranger live QBO access.
He exported January Stripe balance transactions and ran Statement scan. Critical finding: net drift between summed charges and payouts. Effective fee rate higher than the flat 2.9% his books assumed. QBO hint: book fees per payout batch, not a journal estimate.
He re-exported charges and payouts separately. Fixed the import. Added a monthly five-minute fee check to his close routine — same slot as inventory counts, less drama.
For Amazon, he kept the settlement report separate. Different animal. Same principle: line-level exports beat dashboard mood.
The partner lane (bookkeepers reading this)
If you carry ecommerce overflow clients, fee creep is where ProAdvisor hours catch fire — not because QBO is broken, but because the founder never sends clean processor files.
Pair Statement scan with the QBO Drift checklist. Export in, findings out. No connector cosplay. Your firm still signs the close.
For month-end horror stories framed for firm owners, see Month-End Close When Stripe Fees Don't Match QBO Deposits.
Novelty without panic
Fear sells AI governance decks. Novelty sells "your processor is lying in aggregates" — which is catchy but slightly wrong. The processor is summarizing. Your books need detail. The gap is boring. Boring gaps are fixable.
We build free tools at that gap: store trust on the front (Store scan), processor trail on the back (Statement scan). Paid work starts when you want someone to implement fixes, not when you want a slide deck about guardrails.
If margin feels tight but revenue looks fine
Export last month's processor file. Run Statement scan.
Reply with your stack (Shopify + Stripe, Woo + PayPal, marketplace-only) and what does not foot — we will point you at the right check or intel note. No book-a-call pressure. Explore the tool first.