## Flatiron Books, New York
E-commerce clients are reconciliation nightmares. High transaction volume, multiple payment processors, refunds that cancel out purchases, chargebacks that look like normal charges until you look twice.
Flatiron Books had built a niche serving Shopify and Etsy sellers in the fashion and home goods verticals. Their clients loved them. Their bank feeds were a different story.
### The Volume Problem
One flatiron client processed $140,000 in sales over a 30-day period across Stripe, PayPal, and Shopify Payments. The feeds didn't reconcile automatically—manual matching was required for roughly 30% of transactions due to processor timing differences.
"I had a spreadsheet system," said founder Nina Osei. "It worked until it didn't. And by 'didn't' I mean 'I missed something and a client called me on a Saturday.'"
The duplicate that caused that Saturday call was a $340 shipping charge from the client's freight partner. It had appeared twice in the feed—once on the 3rd, once on the 17th. Both looked valid. Only one was.
The client paid it. They noticed three months later during a tax review. The relationship survived, but barely.
### After ReconAuditIQ AI
Nina runs her transaction feed through ReconAuditIQ AI before she touches a spreadsheet. The duplicate detection flags exact matches. The behavioral outlier model flags items like the $340 shipping charge—same vendor, same amount, within a suspicious window.
In her first three months, the tool surfaced:- 14 exact duplicates across her six largest clients (total savings: $12,200 in client refunds) - 23 behavioral anomalies that warranted a client check-in - 3 partial duplicates she'd never caught before
"It paid for itself in week two," Nina said. "I've raised my prices twice since and my clients haven't blinked. I can promise them no surprises."