DEV Community

MapToSite
MapToSite

Posted on

A bakery reprints its menu 40 times a year. The QR code should be printed once.

A bakery near me writes the day's bake list on a chalkboard at 7am, and by 11am three items have a line through them. That is the whole product: what came out of the oven today, and what is already gone.

The paper under the counter cannot do that. Neither can a PDF, which is the thing most "QR menu" setups actually point at. And this is where the QR itself gets it wrong in a way nobody notices for six months: the code is printed on the table tent, the poster, the counter sticker, and the receipt. If the address behind it changes — a new domain, a rebuilt PDF, a tracking parameter added by whoever configured it — every one of those printed pieces is garbage, and the owner finds out from a customer who gave up.

So the rule the whole design follows, before any code:

The QR code holds a permanent page address. The page holds the data. Nothing about the QR depends on the menu.

Everything below is what that rule costs you in practice, and what it costs over three years.

The four things a food shop changes, and how often

I have watched what actually gets edited on the shop pages I build (Korean cafés, bakeries, barbecue places, one tattoo studio). It is not the design. It is:

What changes How often Who does it
Menu lines and prices weekly to monthly the owner, at night
"Sold out today" daily, sometimes hourly whoever is at the counter
Opening hours, holiday closure monthly, plus every holiday the owner
One notice line ("bread out at 7am, 30 loaves") daily the owner

Four things, four different rhythms, one page. If the edit path requires a designer, an account, a build step, or a WhatsApp message to me, the page is stale within a week and the QR code becomes a link to a lie.

So the edit path is a single link the owner keeps, with short keys, and a 12-line menu. Twelve is a deliberate cap: the fixed point in the layout is the printed A4 sheet, which holds twelve lines at a readable size, and the page uses the same twelve so the paper and the screen never disagree.

"Sold out" is a state, not a deletion

The instinct is to delete the sold-out item. Don't. Deleting changes the printed sheet, the layout, and makes the 11am guest think the shop never had it.

Model each menu line as a record with a status:

// one line, one record — the page, the printed A4 and the table tent all read this
{
  name: "크루아상",
  note: "프랑스산 버터",
  price: 4500,
  status: "sold_out",      // available | sold_out | tomorrow
  sort: 10
}
Enter fullscreen mode Exit fullscreen mode

Three renderings of the same record:

  • the page shows the line greyed with "오늘 품절" (sold out today) and keeps the price visible,
  • the printed A4 keeps the line, so the shop does not reprint a menu because one tray sold out,
  • the counter board can drop it if the owner wants a clean board.

That is the entire daily loop: toggle one status, the page updates, the paper stays valid.

The three-year arithmetic, with the assumptions on the table

This is the part owners actually want, and it is arithmetic, not marketing. Reprint costs vary by shop and by country, so put your own numbers in:

annual_reprint_cost = (table_menus + counter_boards) × reprints_per_year × cost_per_sheet
three_year_total    = annual_reprint_cost × 3
Enter fullscreen mode Exit fullscreen mode

With a modest Korean shop — 12 laminated table menus plus one counter board, a local print shop at roughly 2,000원 per laminated A4 sheet:

Reprint cadence Sheets per year Cost per year Three years
Quarterly 52 ~104,000원 ~312,000원
Monthly 156 ~312,000원 ~936,000원
Weekly 624 ~1,248,000원 ~3,744,000원

Now the other column, at the prices on my own order page (no subscription, no auto-charge):

One-time Ongoing Three years
Set-up, 3 months of hosting included 39,000원 — 39,000원
After month 3 — 4,900원 / month 161,700원 (33 months)
Total 200,700원
Printing — — separate, your print shop

Roughly 200,700원 sits between "monthly reprint" and "weekly reprint" in the table above, and it does not grow with how often the menu changes. That is the argument, and it is also its limit: a shop that reprints twice a year does not need this. Say so out loud, because the owner will think it anyway.

Five implementation details that decide whether it survives a year

1. No tracking parameters in the printed code. Every character in the encoded URL adds capacity, and capacity adds modules or lowers error correction. Forty characters of ?utm_source=... inside a 2cm table sticker is the difference between a code that scans first try and one that does not. Track the printed code by where it points (/t/<shop>/<table>), not by decorating it.

2. Give each table its own path, not its own domain. /t/milbat-bakery/1 through /10 stays short, keeps one certificate, and lets the kitchen see which table asked for water. Ten tables is the practical ceiling for this format — past that, people want a real POS.

3. Cache the page, revalidate the status. The page is static except for the sold-out flags. Serve it with a short s-maxage and stale-while-revalidate so a 3G phone at the door paints instantly, and the status catches up within a minute:

Cache-Control: public, s-maxage=60, stale-while-revalidate=300
Enter fullscreen mode Exit fullscreen mode

4. Print at 300dpi, keep the quiet zone, and do not laminate the code. Gloss plus overhead lighting is a scanner's worst case. If the piece must be laminated, matte, and leave the four-module quiet zone empty — no logo inside it, no border touching it.

5. The edit link is the owner's, not yours. A page the owner cannot change is a page that dies when you stop answering. Short keys, no account, no password reset loop: the link itself is the credential, so treat it like one and tell them where to keep it.

What the demo actually is

The two screens below are a real build, not a mockup — a bakery demo with sample data, the mobile shop page and the table QR view:

And the bilingual ordering page for the Korean market, which is where the three-year number above comes from: 테이블 QR 메뉴판 세팅 접수.

If you want to run the arithmetic for a shop you know, the calculator takes the numbers and shows the three-year total beside the subscription alternative: three-year cost calculator.

The kit itself — the page plus the printed files, one payment — is here: order the kit. A free version of the page exists too, no account required; the paid part is the printed kit and the edits.

The one-line version

If the QR code can outlive the menu, the shop stops paying for the same decision twice. Print the code once, on matte paper, pointing at an address that will not change, and put every daily change in the page behind it. The twelve lines, the sold-out flags and the hours are all the same record, and the paper is a rendering of it — which is the only arrangement where a 7am bake list and a 7pm dinner menu can both be true on the same sheet of paper.

Top comments (0)