Where it started
A curated wine shop on the Upper East Side reached out because their owner was losing her evenings to accounting. She'd spent two years doing end-of-day data entry by hand — partly because the off-the-shelf integrations she'd tried all got the specifics of her business wrong, and partly because she'd learned not to trust anything she didn't personally check.
Her question was: "should I just hire a bookkeeper?"
We asked if we could sit with her for a night first.
What we found by listening
The work she was doing by hand wasn't really bookkeeping. It was translation. Each online order had to be entered into her accounting system in a shape that handled two quirks of her business:
- New York state sales tax — which, for wine, varies by shipping ZIP and has edge cases most tools lump incorrectly.
- Wine-club member discounts — which needed to flow through as per-line adjustments, not as a generic coupon, so her margin-per-SKU stayed accurate.
Hiring a bookkeeper would have solved the labor problem but left the underlying data pipeline fragile — someone else would still be translating by hand, just not her. The real fix was to handle the translation once, correctly, and then automate it.
What we built instead
A middleware layer that sat between her online store and her accounting software, designed around the two quirks that had been tripping off-the-shelf tools.
- Every paid order triggered the translation: line items parsed, shipping-ZIP sales tax applied correctly, wine-club discounts broken out per line.
- The accounting invoice was created or updated under the matching customer record, creating a new customer cleanly when needed.
- A short end-of-day summary landed on her phone, so she could glance at it while closing up instead of opening a laptop after dinner.
- A private admin page let her re-sync any order manually if she ever wanted to, for peace of mind.
We also wrote up exactly how the tax logic and discount handling worked, in plain English, so she'd never be locked in by an opaque integration again.
The outcome
- Roughly 15 hours a month returned to the owner. She started running a weekly wine tasting with the hours — something she'd been wanting to do for two years.
- Books are current-day instead of weekly behind. She can see margin-per-SKU in real time, which has quietly changed what she stocks.
- Wine-club membership up 22% in the following six months, substantially because of the tastings the reclaimed hours made possible — adding roughly $8,000 in monthly recurring revenue from new members alone.
- Zero missed orders since launch. The one time there was a delivery hiccup, her admin page caught it instantly.
"I didn't want another subscription. I didn't want a bookkeeper. I wanted someone to stop and understand my actual business for a minute, then fix the thing. That's what they did." — Owner, the Upper East Side wine shop
Why it worked
The fastest answer would have been a bookkeeper — and the wrong one. The real work was understanding the two business-specific quirks that off-the-shelf integrations kept getting wrong, then handling them in a way that preserved her margin accuracy and freed her evenings.
Naming the real problem — and being honest that the labor-substitute answer wouldn't have actually fixed it — was the engagement. Everything we built downstream of that diagnosis lives on her infrastructure, owned by her, with the logic documented so she's not locked in by anyone, including us.