Why manual reconciliation fails above 1,000 orders per month
The tipping point where spreadsheet reconciliation stops working — and why it's cognitive bandwidth, not spreadsheet limits, that actually breaks.
Most Indian commerce teams start with manual Excel-based reconciliation and eventually hit a wall where it stops working. The exact point varies, but across hundreds of conversations with founders and finance operators, the breaking point is remarkably consistent: somewhere between 1,000 and 3,000 orders per month per marketplace.
The breaking point isn't technical. Excel handles millions of rows. The breaking point is cognitive. Reconciliation requires a specific kind of attention — spotting small differences across tens of thousands of data points — and that attention degrades non-linearly with fatigue. By row 500, an analyst catches everything. By row 2,000, they catch the big stuff. By row 5,000, they're mostly just approving things to be done.
What gets missed in that degradation is precisely what costs money. Commission variance of a few rupees per order looks like noise to a tired human and becomes a pattern only in aggregate. Small rounding differences that signal category misclassification. Tiny fee drifts that accumulate into meaningful margin leakage. These are exactly the things spreadsheet reconciliation is bad at and automated reconciliation is good at.
The second reason manual reconciliation breaks at scale is the compound effect of multiple marketplaces. Reconciling Amazon alone is tractable. Reconciling Amazon + Flipkart + Meesho + a payment gateway means four separate spreadsheet workflows, four separate rate cards mentally applied, and four separate sets of file formats to wrangle. Each marketplace adds more than proportional cognitive load.
Teams who've made the switch usually describe it the same way: not that automated reconciliation caught dramatically more issues in any one category, but that it caught issues in every category consistently. The silent leakage across 20 small categories, each small on its own, adds up to the first month of recovered margin paying for the tool — and that's before accounting for the analyst hours freed up.
The rough rule of thumb: if you cross 1,000 orders per month on a single marketplace, a reconciliation tool starts paying for itself. If you're multi-marketplace, the tipping point is lower — closer to 500 orders per marketplace. If you're still in spreadsheets at 5,000+ orders per month, you're almost certainly leaving more money on the table than the tool costs, and you've been doing it long enough that the leakage is now an operating assumption rather than an exception.