Order-Level vs Summary Reconciliation: What to Choose

By TallySutra Team · 16 July 2026 · 5 min read

There are two ways to reconcile marketplace settlements. The summary method compares totals: sales for the month versus settlements received versus bank credits, with a bridge for fees and timing. The order-level method matches every individual order from sale through settlement to payment. The first takes an hour; the second, done manually, can take days. Most sellers and many accountants choose the hour, and the choice is understandable, but it has a specific, quantifiable cost that this article lays out. By the end you will know which method your situation actually requires.

What summary reconciliation can and cannot see

Summary reconciliation answers one question well: is the overall picture roughly right? If settlements banked are wildly off from expectations, a total-level view catches it. What it structurally cannot catch is anything that offsets: an order short-paid by 400 next to an unexpected timing credit of 400 nets to zero at summary level. Fee overcharges spread thinly across thousands of orders look like a slightly high fee ratio, indistinguishable from a mix shift. Refunds issued without the fee reversals due back sit invisibly inside the returns total. The summary method is a smoke detector, not an audit: it tells you if the building is on fire, not that the tap in room six is leaking money continuously.

What order-level matching surfaces

• Short payments: orders settled below the expected net, visible only when expected and actual are compared per order.
• Fee errors: deductions above rate card for the category, identifiable and claimable per order ID.
• Missing fee reversals: returns where the refund was deducted but reversible fees were not credited back.
• Aged pendings: delivered orders that never settled, indistinguishable from timing at summary level until they are individually tracked and aged.
• Duplicates and anomalies: the same order charged twice, adjustment lines with no explanation.

Each finding comes with the order ID and settlement line attached, which is precisely what a marketplace claim needs. Summary findings, by contrast, are unarguable internally and unclaimable externally.

The honest comparison

When each method is the right call

Summary reconciliation is defensible when volumes are small, margins are comfortable, and the seller consciously accepts leakage risk in exchange for time, a few hundred orders a month with low return rates, say. It is also a reasonable weekly pulse-check layered on top of monthly order-level work. Order-level becomes the right call as any of three things grow: volume (more rows for errors to hide in), return rates (more reversals to go wrong), or the stakes of the books themselves, GST scrutiny, audits, funding due diligence, where order-level detail behind the ledgers is what makes the books defensible. The trap to avoid is drifting: sellers who chose summary at three hundred orders a month are often still on it at three thousand, not by decision but by inertia.

Automation changes the trade-off

The entire case for summary reconciliation is the manual cost of order-level work, and that is exactly the cost automation removes. When settlement files are matched programmatically, order-level is no slower than summary: import the reports, and every order is matched, pending or flagged, with the exceptions queued for human attention, the workflow described in resolving reconciliation exceptions. This is the model TallySutra runs: order-level matching from uploaded Amazon, Flipkart and Meesho files, results posted as balanced TallyPrime vouchers after review, so books and reconciliation are the same artefact rather than two jobs, see the features page for the pipeline and pricing for the free pilot tier. Once the cost side of the trade-off collapses, the choice answers itself; the only sellers who should still be on summary-only reconciliation are those who have checked the price of order-level and found it genuinely not worth it, and at current tooling prices, that is a short list. The month you do switch, expect findings: a first order-level pass over books that have only ever seen summary checks almost always turns up an aged tail nobody was watching, and clearing that tail is the switch paying for itself.

Frequently asked questions

Related reading

• Build vs Buy: Marketplace Accounting Automation
• Marketplace Accounting Mistakes That Cost Sellers Money
• The Multichannel Seller's Accounting Guide
• Reconciliation Exceptions: How to Resolve Them