Etsy SEO

What Etsy search actually weighs

Etsy matches a shopper's query against your title, tags, categories and attributes. That is the relevancy half, and it is entirely in your hands. The other half is performance: how often people who see your listing click it, favourite it, and buy it. A listing that converts well moves up; one that gets impressions and no clicks drifts down.

This matters more than it first appears. It means listing copy is not only a ranking input — it is also the thing that determines whether you keep the ranking. A title that wins the match but loses the click is worse than a plainer one that does both.

Titles: 140 characters, two audiences

A title has to satisfy a search index and a human being scanning a grid of thumbnails, and those two want different things. The index rewards the specific phrases a buyer types. The human rewards clarity in the first few words, because that is all that survives truncation on a phone.

The practical rule is to front-load the phrase someone would actually search, then add the qualifiers — material, size, occasion, recipient — after it. Not “Handmade Gift Beautiful Unique Candle Soy Wax” but “Soy Wax Candle, Lavender, Housewarming Gift for New Home”.

Tags: thirteen slots, and most sellers waste them

You get thirteen tags of up to twenty characters each. Two mistakes are near universal: leaving some empty, and filling them with single words.

Single-word tags put you in competition with every seller in the category. Long-tail phrases — “gift for new homeowner” rather than “gift” — match fewer searches but match them far more precisely, and the shopper typing four words is much closer to buying than the one typing one.

Repeating the same phrase across several tags does not compound. Thirteen distinct phrases beat five phrases said three ways.

Why AI-written listings come out bland

The model is not the problem; the brief is. “Write an Etsy title for a candle” contains no product, no buyer, no occasion, and no mention of the 140-character limit. With nothing to work from, a language model returns the safest phrasing it knows — which is the phrasing every other seller also got.

A brief that states the product, the buyer, the occasion, the constraint and the output format produces something you can publish. The difference in results is almost entirely the difference in input length.

Where to start

Open your worst-performing listing and check three things. Does the first half of the title contain the phrase a buyer would type? Are all thirteen tags filled with distinct multi-word phrases? Does the first photo make sense as a thumbnail at the size it is actually viewed?

Change one listing, leave it for a fortnight, and compare. Changing everything at once tells you nothing about what worked.