Sample analysis, not a client: Darkroom Chocolate (North Walsham, UK)

Darkroom Chocolate is a UK bean-to-bar chocolate maker with an online shop. We have no relationship with them. This is a public example of what we check, made from their public pages and two live ChatGPT answers.

What ChatGPT tells a shopper (measured 4 October 2026)

Measured in a logged-out ChatGPT session, browser language en-GB, from an IP outside the UK (product cards in the answer showed prices in a foreign currency, so a UK shopper may see different cards). Full answers are saved with timestamps.

"where can I buy the best craft chocolate bars online in the UK?" (06:46 UTC) Named: Pump Street Chocolate (best overall), Cocoa Runners (best online selection), Dormouse Chocolates (best specialist maker). Also listed: Bare Bones, Duffy's, Chocolate Tree, Chocolarder. The shopping cards showed Pump Street bars with "5.0 · 82 reviews" and "5.0 · 30 reviews". Darkroom Chocolate: not mentioned.

"best bean to bar chocolate makers in the UK" (06:50 UTC) Named: Pump Street, Duffy's, Dormouse, Land Chocolate, Bare Bones, York Cocoa Works, NearyNógs, Luisa's Vegan Chocolates. Sources it showed: Chocolatier.co.uk (3 times), Speciality Food Magazine, TasteAtlas, Best in Britain, a chocolate judge's blog. Darkroom Chocolate: not mentioned.

What we found on their own product page

Page checked: darkroomchocolate.com/products/dark-flight-gift-box-4-bars (fetched 4 October 2026).

What is already right: the page carries Product data with name, brand, image, price (20.00 GBP) and stock status.

What holds it back:

1. No reviews anywhere on the product page. The page HTML shows no review widget or review text (the only review service on the site is a Trustpilot shop badge), and the Product data has no rating. In the first answer, two of the four product cards ChatGPT showed carried star ratings and review counts. A page with no reviews gives it less to work with.
2. The page description is double-encoded. The meta description reads sweet & fruity in the HTML, so search snippets and AI tools that quote it can show "&" instead of "&". The same "&" sits inside the Product description data.
3. The answer leans on third-party guides. For "bean to bar makers" ChatGPT cited Chocolatier.co.uk three times, plus Speciality Food Magazine and TasteAtlas, alongside makers' own sites. Being listed or reviewed in those guides is the off-site work most likely to change who gets named.

Three fixes, in order

1. Collect product reviews on the product pages (a Shopify review app) and include the rating in the Product data.
2. Fix the double encoding in the product description field so "&" shows as "&".
3. Ask the UK craft chocolate guides ChatGPT cites (Chocolatier.co.uk, Speciality Food Magazine) and specialist retailers such as Cocoa Runners to review or stock the bars.

Evidence: full ChatGPT answers (06:46:57Z and 06:50:15Z UTC) and a copy of the page, both saved on 4 October 2026, available on request.