Fix the cold start problem for a brand-new product's recommendations

I just launched a product with zero purchases, and the 'similar items' box has nothing to show for it.

Beginner to intermediateRead 5 minDoing: an afternoon7 steps

▶ Start

What you'll have at the end

A ranked list of similar items for the new product, generated only from its title and description, no click data yet

Before you start

You need

A catalog whose existing items already carry a stored content vector built from their own title and description, plus a 'similar items' surface that currently reads only from a table of item-to-item scores built from past customer interactions.

Not covered

The mirror problem of a brand-new shopper with no purchase history; this only covers a brand-new item that has no interactions yet, shown to shoppers the interaction-based table already knows well.

Leans on

7 steps

Where to go next

Verify your first embedding works

go there first if you don't yet have a working embedding call to produce the content vector this recipe starts from

Speed up nearest neighbor search with a FAISS index

go there once the catalog is too large for a plain sweep against every stored vector to stay fast

Checked 18 Aug 2026

Part of the Embeddings cookbook