Jun & Co.Shopify · Klaviyo · ADA accessibility
If you run a jewelry store

You can’t see what a blind shopper can’t do.

A fine-jewelry store with a well-built theme, thousands of products, and one component that made an entire collection page unusable for anyone shopping with a screen reader.

Real store: a New York fine-jewelry retailer, kept anonymous · Accessibility
13checks run on three pages
11passed
490swatches with no name
½ dayto fix, including re-testing
They came with

A store nobody had complained about.

No letter, no complaint. A retail client we already worked with, and a store that looked finished. We ran the audit on three pages the way we always do: the home page, one collection page, one product page, by keyboard and then with a screen reader.

What we found

Eleven checks passed. Two failed. Both came from the same place.

The theme was better built than most: every image described, every button and form field named, text readable against its background across 704 samples, and an accessibility statement already published.

The two failures both traced to one component: the metal swatches on every product card. Each swatch was a link with no name, and each was 20 pixels wide, under the 24-pixel minimum a finger or a pointer needs. On one collection page there were 490 of them. To a screen reader, the page was the word "link" 490 times, and a shopper could not tell gold from rose gold from silver.

CheckMeasuredResult
Every photo has a description533 imagesPass
Buttons say what they do62 buttonsPass
Form fields say what to type42 fieldsPass
Text is readable against its background704 samplesPass
The page says what language it is in3 pagesPass
Keyboard users can skip the menu3 pagesPass
Phones can pinch to zoom3 pagesPass
Screen readers can jump between sections3 pagesPass
Tab order follows the page3 pagesPass
Each page has a real title3 pagesPass
An accessibility statement existssite-widePass
Links say where they go1,125 links526 fail
Buttons are big enough to tapall controlsFail

What we fixed

One line, in one template.

The product-card template builds every swatch. One line there gives each one its name: the product plus the metal, so a screen reader says "Huggie hoop earrings, rose gold vermeil, link." Half a day, including re-testing the page with a screen reader afterward.

What we recommended

Software can count names and measure pixels. They cannot tell you whether a keyboard shopper can see where they are on the page, or whether the text laid over the hero images is readable in every case. Both were left open for a person to test before sign-off, instead of marked passed on a guess.

The third is the swatch size itself: 24 pixels is the minimum, and the same template controls it.

We also noticed

The good news is worth saying out loud.

Most of what a demand letter points to was already right on this store: image descriptions, labeled forms, readable text, a published statement. That is rarer than it should be, and it is the kind of thing we put in the report as clearly as the failures, because a founder deserves to know what is working.

What they have now

A page a screen reader can shop, and the record.

A collection page where every swatch says what it is, a dated report of exactly what was tested and what was found, and a short list of what to check with keyboard and screen reader testing next. The 490 were a subset of the 526 unnamed links the audit counted. A measured result on three pages, not a claim about the whole store.

Sound like your store?

Start with the $297 audit