Nobody has seen this before, so the site has to teach.
Most marketing sites are selling a familiar thing better than the next company. This one had a different problem. Photograph the inside of your fridge, get recipes back from what is actually in there, is not a category anyone is already shopping in. If a visitor does not understand the mechanic within a few seconds, they leave, and no amount of persuasion further down the page ever gets read.
The second challenge was justifying a subscription. On the surface a recipe app sounds like a novelty, the kind of thing you try once. The value only becomes obvious when you put it next to the number: a typical family bins around a thousand pounds of food a year, and most of that is food they forgot they had.
There was also a real product decision hiding in the copy. Scanning a fridge is the headline, but scanning a shopping receipt is often the better first experience, because it stocks the whole pantry in one go. Both had to be visible without making the product sound complicated.
Three steps, then the number.
We built the page around comprehension first and persuasion second. The hero states the mechanic in one sentence, then an immediate three step explainer walks through scan, get recipes, cook and save, using real screens from the app rather than illustrations.
- Three step explainer directly under the hero, using real app screens so the mechanic is unmistakable
- Both scan modes given equal weight, fridge and shopping receipt, without complicating the pitch
- Food waste figures used as the commercial argument, with the source named for each number
- Feature grid covering automatic expiry dates, cook now versus almost there recipes, the shopping list and the savings tracker
- An old way against new way comparison that frames the product as a habit rather than a novelty
- Free tier stated plainly with limits shown, and a paid tier positioned as paying for itself in saved food
- Marketing site and app deployed separately, so the site stays static and instant while the app scales on its own
- Prompt handling hardened against injection, since the app feeds user photos and text straight into a language model
We built both halves: the app in React with an OpenAI vision pipeline, Stripe billing and Redis backed limits, and the marketing site as a single static page with real app screens.
Fridge to dinner in about thirty seconds.
The app is live at scan.tryfridge.com with the marketing site in front of it. A scan takes seconds, expiry dates are estimated automatically so nothing needs typing, and the site itself loads in around six tenths of a second.
Two ways to stock a pantry
Point the camera at a fridge, or snap a shopping receipt and stock everything in one go.
No dates to type
Use by dates are estimated for every item, then the app nudges before anything spoils.
Site loads in well under a second
A single static page with real app screens, deployed separately from the app itself.
These three are our own products, not client work. We build them with the same process we sell: one clear message, a page that loads instantly and a route to purchase that does not make anyone think. Building and running our own apps is also how we stay honest about what actually converts.Wellington Web Co. Studio, in house build
Launching something people have not seen before?
New ideas are won or lost in the first five seconds of a page. If yours needs explaining properly, that is exactly the kind of work we like.