Reviewing an import

Every import route — AI, quick, or CSV — lands here before anything is saved. Review & verify is the last point at which a mistake is free, and it is worth spending real time on for a large book.

The header tells you the shape of what is about to happen:

• a count of recipes marked READY
• how it was read — e.g. “extracted by AI — edit any field before saving”
• new ingredient(s) · new unit(s) · new station(s), with “created automatically as drafts on import”

Read the “new” counts before anything else

Those three numbers are the best early warning you have.

A plausible number of new ingredients means the import matched your catalog. An implausible one means it did not — 41 new ingredients on a ten-recipe import into a full catalog usually means names arrived in a different form (yellow onions vs Yellow onion) and you are about to double your catalog.

On a brand-new brand with an empty catalog, a large number is expected and correct.

New units and new stations deserve more suspicion than ingredients, because you have far fewer of them. A new unit is often a typo (Tbsp. with a period), and a new station is often a step’s text mistaken for a station name.

Each row

Every recipe is one row, and everything on it is editable:

• Title — fix it here rather than after import
• Type — Production or Prep. Worth checking on every row: a spice blend or a sauce imported as Production will not behave as a building block. See Prep.
• Station — — none — or one of yours
• A checkbox — ticked means it will be created. Untick to exclude a row without losing the rest of the import.

Expanding a row shows what was actually read:

• Description — “Menu description / notes (shown in Front-of-House views)”
• Ingredient rows — quantity, unit, name, each editable, with a ✕ to drop one
• a NEW pill on any ingredient not already in your catalog
• a step count, e.g. “2 steps · edit the full method after import”

The NEW pills are your dedupe check, per row. If cornmeal and egg are marked NEW and your header says “2 new ingredient(s)”, those agree and the import matched everything else. If a common ingredient you know you have is marked NEW, the name arrived in a different form — fix it in the row now, and you avoid a merge later.

Warnings

A row that could not be read cleanly gets a Check pill and an explanation on the exact field, e.g.

kosher salt: couldn't read quantity "to taste"

with the quantity field outlined.

A warning never blocks the import, and that is deliberate: “to taste” is a real instruction with no number, and refusing to import it would be wrong. The warning tells you the field is empty on purpose or by accident, and you are the one who knows which.

⚠ Do not skim past warnings on a large book. They are per-ingredient, and a missing quantity produces a recipe that looks complete on the line and tells a cook nothing about how much to use. The steps are the thing to check on a photographed or scanned source, since a transcription error there is invisible in the row summary — open the row.

Committing

Import N recipes as drafts creates them. Everything arrives as a draft, so:

• nothing reaches the line from an import
• each recipe still goes through your normal review and publishing
• you can fix the full method afterwards in the editor

Start over discards the whole extraction without creating anything.

What to do straight after a big import

1. Ingredients → duplicates filter, and merge what arrived twice. Undo is available for two minutes, so do it deliberately. See The ingredient catalog.
2. Check the recipe types — Production vs Prep, in the list view.
3. Set default stations, so the Heart-of-House grouping and your station analytics are not all Unassigned.
4. Add allergens to the new ingredients. Import does not invent them, and an ingredient with none set is a claim that it has none.