To import a recipe from a website, paste the page’s link into a recipe keeper that reads structured recipe data. Good apps pull the ingredients and steps in a second or two and hand you a draft to confirm. When a site does not publish that data, you fall back to copying the text. Both paths should end in a clean recipe you can cook from.
Why most sites import cleanly
Most cooking sites embed their recipe in a hidden format called schema.org Recipe data. It lists the title, ingredients, steps, times, and yield in a way machines can read. When you paste a link, a good importer finds that block and maps it straight into your library. This is why a well-built importer feels instant on a major food blog.
Why some imports fail
Three things break import:
- No structured data. Some personal blogs and older sites publish none. The importer has to guess from the page’s markup, which is less reliable.
- Pages that load with JavaScript. If the recipe only appears after scripts run, a simple fetch sees an empty page.
- Anti-scraping blocks. A few sites refuse automated requests outright.
The honest truth is that import can never be perfect, because the web is not consistent. The fix is not to pretend otherwise. It is to try the reliable path first, fall back gracefully, and always let you edit the result.
How to get a clean import every time
- Paste the link first. It is the highest-quality path. Let the app do the work.
- If a field looks off, fix it in the draft. A good importer opens the recipe for review before saving, so a stray line takes seconds to correct.
- For a stubborn page, copy the text. Select the ingredients and method, paste them in, and let the app sort them into sections.
- For a cookbook or a card, take a photo. On-device text recognition reads the page, and you tidy up from there.
What to look for in an importer
Reliability is the whole game. Look for an app that imports without a punitive weekly cap, that shows you the result before saving, and that keeps your recipes on your device. Trivet was built around this exact problem: import is free, it is the first thing measured, and the editor is always there as a safety net.