How to organize your recipes: a system that holds

Gather everything in one place, settle a tagging system before you import anything, name recipes the way you will search for them, and delete what you will never cook. The order matters: tag first, import second, or you rebuild the same mess in a new app. Budget two hours for the first pass, then ten minutes a month. The tool comes last, replacing none of them.

Why collections scatter, and what it costs

Nobody chooses this. A collection spreads across six places on its own: screenshots in the phone gallery, browser bookmarks, Pinterest boards, saved posts and liked videos, PDFs and Word files in downloads, inherited notebooks and handwritten cards. Each has its own search and sort order, and none talk to each other.

The real symptom is not “I have too many recipes”. It is being unable to answer one question: what can I cook tonight with what I have? The second is saving the same recipe three times, having forgotten you had it. The counter-intuitive part, held throughout: the tagging decision comes before the import, not after.

Step 1: pull everything into one place

Give yourself one session with a time limit, not a completeness target. Open each source in turn and move things across without sorting, apart from obvious rejects. Three routes: paper and screenshots get photographed, files go in as PDF, Word or text, links as web pages and videos.

Cookfolio, as one example, accepts JPEG, PNG and WEBP photos, plus PDF, DOCX and TXT, and links from any recipe site, YouTube, TikTok, Instagram, Facebook and Pinterest. One limit worth knowing: for Pinterest the app follows the pin’s outbound link back to the source recipe, so a pin with no outbound link cannot be imported. Handle those by hand. Extraction is always shown for review and correction before saving, making a fast pass safe. We covered what it reads in how AI import works.

The free plan covers 5 recipes and 5 lifetime AI imports across all sources, blog and Pinterest included. Pro is €4.99/month or €39.99/year for unlimited. A Paprika (.paprikarecipes) or Mela (.melarecipes) collection imports in bulk, up to 600 entries, photos included, and 150 MB per file, on Pro. Treat the free ceiling as an exercise: five recipes forces you to name the ones you actually cook.

Step 2: settle your tags before you import anything

One main classification axis, three in total at most: dish type (starter, main, dessert, appetizer, breakfast, snack, drink), diet (vegetarian, vegan, gluten-free), constraint (quick). Cookfolio seeds those eleven tags at account creation; custom tags live under Settings, then Personalisation.

Past roughly fifteen tags, tagging costs more than it returns: you hesitate when filing, and stop using the filter. Watch for semantic duplicates: “quick” and “express”, or “dessert” and “desserts”, are different tags to the machine.

Two behaviours pull in opposite directions:

• AI import of a single recipe only matches extracted tags against tags you already own. Anything unmatched is dropped, so it can never inflate your tag list.
• Bulk import of a Paprika or Mela collection does the reverse: every category in the file becomes a tag, and any unknown name creates a new one, matched on name while ignoring case.

If you are migrating a collection, clean up the categories in the old tool before exporting, or plan a tag pruning pass right after.

For multilingual households: the eleven default tags follow the app’s language, while a custom tag stays exactly as you typed it. If you do not all use one language, lean on the defaults.

Step 3: name recipes so you find them again

Most articles skip this. It follows from a technical reality: in Cookfolio, search runs on the recipe title only, not ingredients, not notes, not the description. Searching “courgette” will not surface a recipe called “Sunday gratin” containing courgettes.

So the title must carry the ingredient or dish you would type six months from now, not the affectionate name. “Grandma’s recipe” becomes “Grandma’s apple pie”. “IMG_4821” becomes “Red lentil curry”. When in doubt, put the main ingredient first.

Two further facts shape navigation. The list is sorted newest first, with no alphabetical or frequency ordering, so title search and tag filter are your only levers. Selecting several tags widens the result rather than narrowing it: a recipe appears if it carries at least one selected tag. To narrow down, filter on one tag and refine with the title search.

Step 4: prune what you will never cook

A useful collection is a pruned one, sorted on one honest criterion: have you cooked it at least once? Cooked and good: keep it, and fill in the notes while you remember. Cooked and disappointing: delete it. Never tried in over a year: delete it too, since it is still findable online.

Said without hedging: Cookfolio stores no favourites, no five-star rating and no last-cooked date. There is no automatic signal for never-tested recipes, so build one. Either create a custom “to try” tag and drop it once you have cooked the dish, making the pile visible in one filter, or record the date and result in the notes field.

Duplicates deserve their own warning. No import path detects them: not file, not link, not bulk. Importing the same recipe twice creates two entries, silently. The only deduplication happens inside Family plan data transfers, not on imports. So search the title before importing anything you are unsure about, and run a duplicates pass while pruning.

Step 5: ten minutes of upkeep a month

A short dated ritual beats a principle. Four moves: fix rushed titles on this month’s additions; check nothing was left untagged; delete tags that fell out of use; delete untested recipes that aged out.

Reassurance on the third: deleting a tag never deletes the recipes carrying it. It is simply removed from them, and the app says how many are affected before you confirm.

Real hygiene comes from use, not tidying. A planned recipe is a validated recipe. The planner works by ISO week, Monday to Sunday, one plan per week, with no lunch or dinner split, and accepts free-text entries for a restaurant or leftovers. Note the scope of drag and drop: you move an already planned meal to another day, you cannot drag a recipe from the catalogue onto a day. More in planning your meals for the week.

The shopping list is built only from planned meals. It merges duplicates automatically, including across compatible units (ml, cl, l and g, kg), but takes no manual items, does not group by aisle and does not reorder. Changing the servings recalculates quantities proportionally, rounded to 0.1, leaving any ingredient without a numeric quantity untouched.

Four pitfalls that undo the whole system

Tag inflation. Symptom: you hesitate more than three seconds when filing. Fix: cap at fifteen tags, merge synonyms.

Silent duplicates. Symptom: two near-identical entries. Fix: search the title before importing, since no import path deduplicates.

The museum collection. Symptom: most of what you own has never been cooked. Fix: a “to try” tag, deletion after a year.

The two-headed system. Symptom: you keep saving recipes elsewhere. Fix: kill the old reflex, deleting the bookmarks folder once migration is done.

Worth being honest about the limits. There is no offline mode, so a connection is required, which matters if you cook where coverage is poor. Sharing happens through a web link, with no PDF or Excel export. There is no voice control in cooking mode, though the screen stays awake and timers are detected in the step text. If you are still digitising, how to digitize your recipes covers that pass.

For reference: Pro is €4.99/month or €39.99/year, and Family €6.99/month or €59.99/year for up to 4 accounts sharing recipes, planner and shopping list in real time, each keeping their own login.