Should your bakery open a second location? Before you tour a single storefront, read your own sales export. Here is what roughly 170,000 line items told one Washington, DC bakery — and the five questions to ask your own POS data this week.
The short answer: don’t decide from instinct or street buzz — your POS system already recorded what kind of location your revenue rewards. When this bakery read theirs, the biggest finding wasn’t an address at all: their market stall was quietly matching the shop, which changed what “a second location” should even mean.
Seylou Bakery & Mill, a whole-grain bakery that mills its own flour in Shaw, was weighing where to grow. We started where every expansion decision should start: the Square export they already had. Ten months of it — roughly 170,000 line items across about 290 trading days. (A note first: the numbers on this page have been changed to protect the owners’ confidentiality — counts are rounded and every dollar figure is scaled by an undisclosed constant. The comparisons, shares and ratios the findings rest on are unchanged.)
| Trading days | Revenue per day | What it costs to keep open | |
|---|---|---|---|
| Shop (Shaw) | 277 | $2,796 | Rent, utilities, fit-out |
| Market stalls | 90 | $2,796 | A table |
That is not a rounding trick. Across 90 market days, the stalls averaged the same revenue per day as the shop did across 277. Split by day it gets starker: the Sunday stall averaged $3,465 a day — 24% more than the shop’s average day — in about five hours, outdoors, with no lease.
And it’s dependable, not lucky: 45 Sundays, never a missed week, the take ranging $2,327–$4,116. Markets are 24% of total revenue off 90 days of trading.
The export contains no costs, so we cannot tell you which channel is more profitable — and we won’t imply it.
The stall carries staffing, transport and market fees. The shop carries rent but sells $173,000 of high-margin coffee that the stalls sell none of. Those pull in opposite directions, and the data can’t say by how much.
What it can say: the revenue case for market stalls is far stronger than most owners assume — strong enough that a stall deserves to be priced properly against a lease before anyone signs one. That’s a question a bookkeeper can answer in an afternoon, and it became the first item on the client’s list.
89% of shop revenue lands before 3pm. The peak is 9–11am; after 4pm the shop is close to dormant.
This matters more for choosing a site than almost anything else. The right question about a candidate street is not “how busy is it?” but “how many people are here between 8 and 11 in the morning?” Those are very different places — a block full of evening restaurants can be busy and useless to a bakery.
The data also showed weekends carry the shop, and that coffee — the highest-margin item — sells almost entirely at the shop, not the stalls. Every one of these findings reshaped what “a good location” means for this specific business.
The sales profile became the yardstick for a site-selection model built almost entirely on free public data:
The full interactive report — maps, rankings, methodology, and the honest dead ends where free data stops — is published in full:
You don’t need a consultant to run the first pass. Export your sales history (Square, Toast, Clover all do this) and ask:
None of this requires more than a spreadsheet. All of it should happen before anyone tours a storefront.
Not until you have read your own POS export. Your sales history already shows what kind of location your revenue rewards — which hours, which days, and which channels carry you. In this case, the data showed a market stall matching the shop’s revenue per day, which changed the question from “which storefront” to “is a storefront even the next move.”
Export your sales history (Square, Toast, and Clover all do this) and run the five questions above: revenue per trading day by channel, revenue share by hour, which weekdays carry you, what sells in which channel, and how consistent each channel is. A spreadsheet is enough.
Revenue and profit are different questions. Here the stall matched the shop’s revenue per day, but the export contained no costs — so the honest answer is that a stall deserves to be priced properly against a lease before anyone signs one.
Mostly free data: your POS export, US Census demographics, LODES daytime-worker data, OpenStreetMap and Google Places. Those layers ranked 541 real leasable addresses in this case study.
Everything on this page came out of one sales export — a file that took the owners minutes to send. If you’re weighing a second location, a stall, or a lease, the cheapest step available is finding out what your own numbers already know, before a landlord is holding your signature.
A full written read of your export — the same analysis at the top of this page, on your numbers: channels, hours, weekdays, item mix, and the questions to settle before a lease — is $199, delivered within five business days of your export arriving. On the typical unit in this report, that is about one day’s rent, spent before the lease instead of inside it.
CodeMyVibe is one person plus AI tooling, priced for small business, and honest about what the data can’t say — this page shows exactly what that looks like.