A Marketplace Fee Audit Guide for Online Sellers
Marketplace fees are calculated by systems processing millions of transactions, against rate cards that change, categories that get remapped, and weights that get re-measured. Errors happen in both directions, and they are your responsibility to find: no marketplace proactively refunds a fee you did not dispute. A fee audit is the systematic check that every deduction in your settlements matches what the published rate card and your contract say it should be. Here is how to run one without it consuming your month.
Know what you should be charged
The audit starts from the rate card, not the settlement report. For each marketplace you sell on, collect the current fee schedule for your categories: commission percentages, shipping rate structure by weight band and zone, collection fee basis, fixed or closing fees by price band, and the fee-reversal policy on returns. Marketplaces publish these in their seller documentation, and they change, so date-stamp what you collect. From the rate card, you can compute an expected fee for any order: category commission on the sale value, shipping for the item's registered weight and the delivery zone, plus applicable fixed fees. That expected figure is the yardstick for everything that follows. Where your contract includes negotiated rates, file those alongside the public card; the yardstick is whatever you actually agreed, not just what is published.
Compare expected to actual, order by order
With expected fees computable, the audit is a per-order comparison against the itemised deductions in your settlement reports. The classic findings:
- Category misclassification: your product mapped to a higher-commission category than it belongs to.
- Weight disputes: shipping charged on a higher weight than the item's actual or registered weight.
- Fee lines outside the rate card: charges that do not correspond to any published fee head.
- Missing reversals on returns: the refund deducted, but reversible fees never credited back, one of the most common and most claimable findings.
- Duplicate deductions: the same fee applied twice to one order across settlement cycles.
Tolerance matters: rounding and minor timing effects create noise, so set a threshold below which differences are ignored, and focus attention on systematic patterns, the same error across many orders is where the material money is.
A practical cadence
| Frequency | Work | Goal |
|---|---|---|
| Monthly | Fee totals by head as a percentage of sales, compared to prior months | Catch drift: rate changes, remappings, new fee heads |
| Monthly | Order-level expected-versus-actual on flagged orders | Confirm and quantify specific errors |
| Quarterly | Full pass on returns: was every reversible fee reversed? | Recover the most commonly leaked category |
| On rate-card updates | Refresh expected-fee logic | Keep the yardstick current |
Raising claims that get answered
Marketplace support responds to specific, verifiable claims and struggles with general complaints. A claim should contain: the order ID, the settlement ID and line, the fee charged, the fee expected with the rate-card basis, and the difference. File through the marketplace's own dispute or case process, batch similar errors together, and track claim status alongside your reconciliation so accepted claims are matched to their eventual credit and rejected ones are consciously closed. Keep a simple claims register even at small volume, claim date, order IDs, amount, status, outcome; marketplaces occasionally reject first submissions that succeed on resubmission with better evidence, and the register is what stops accepted-but-uncredited claims from quietly expiring. This is also where the evidence trail in your accounting pays off directly: if every voucher links back to settlement lines, assembling a claim is retrieval, not research, the principle behind evidence-backed accounting.
Where automation fits
The comparison engine of a fee audit, expected versus actual per order, is mechanical, and doing it by hand monthly is why most sellers audit fees once and never again. An automated reconciliation pipeline does the heavy part continuously: TallySutra imports your Amazon, Flipkart and Meesho settlement files, itemises every deduction into its own ledger in the generated TallyPrime vouchers, and pushes unexplained or anomalous lines into an exception queue where they age visibly until resolved. Fee drift then shows up in your monthly ledgers without a separate exercise, and the order-level detail for claims is already organised; our exceptions guide covers the resolution workflow. Whether you automate or not, the discipline is the same: know the expected fee, compare it to the actual, and never let unexplained deductions ride. Money you do not question is money the system assumes you agreed to pay.
Frequently asked questions
How much money do fee audits typically recover?
It depends entirely on your volume, categories and return rates, and no general figure would be honest. The audit tells you your own number: run the expected-versus-actual comparison on one recent month and let your data answer.
What is the most commonly missed fee error?
Missing fee reversals on returns: the customer refund is deducted, but fees that should be credited back are not. It hides inside returns totals and only order-level checking of returned orders surfaces it.
How far back can I claim fee errors from a marketplace?
Each marketplace sets its own dispute windows, and they are typically limited, another reason to audit monthly rather than annually. Check the current policy in your marketplace's seller documentation before assuming a claim is possible.
TallySutra turns Amazon, Flipkart and Meesho reports into reconciled, reviewed TallyPrime vouchers — duplicate-safe, with every rupee traceable to its source row.
Book a 30-minute demo