One catalogue, many channels
The channels want the same catalogue in different shapes. Maintaining one feed per channel by hand is how catalogues drift; maintaining one enriched source and projecting it is how they do not.
What they share
The core is the same everywhere: a stable id, a title, a description, an image, a price, availability, a link. Any channel can be served from a catalogue that has those correct, which is why feed quality is not a Google-specific investment.
Where they differ
| Channel | What it wants that Google does not |
|---|---|
| Meta | Square or 4:5 imagery, richer lifestyle assets, its own category taxonomy |
| TikTok | Vertical video per product, short punchy titles |
| Tall imagery, strong scene context | |
| AI shopping surfaces | Dense structured attributes and unambiguous identifiers |
| Marketplaces | Their own category tree and mandatory per-category attributes |
The pattern in that column is consistent: every channel except Google wants more creative, and AI surfaces want more structure. Both can be produced from a catalogue you already have.
The AI shopping surfaces are the new one
An assistant answering "which of these three is best for a small kitchen" is reading attributes, not advertising. It needs dimensions, materials, compatibility and power ratings, which are the fields most catalogues leave blank. Enrichment done for Google covers the same ground.
The structure that holds
- One source of truth for product data, in the store or the PIM.
- One enrichment layer that fills attributes, normalises values and generates missing creative.
- Per-channel projections from that layer: field mapping, taxonomy mapping, image crops, video formats.
- Per-channel diagnostics fed back to the same place, so a rejection in one channel is fixed once.
This is what Cobiro’s multi-channel push does: an enriched catalogue projected into Google Shopping, Meta and AI search surfaces from one place, with stock and price changes propagating to all of them.