N-grams: why the keywords that worked three years ago no longer match
Three years ago you could fill your 13 tags with single words: "necklace", "silver", "boho", "gift". And your shop got by. Today those same tags barely bring anything in. Your shop didn't decline, search changed its logic: it no longer reads words, it reads phrases.
What an n-gram is
An n-gram is a run of words that forms one unit of meaning: "name necklace", "moon necklace silver", "mothers day gift". Two to four words that, together, describe a precise intent. That's exactly what a buyer types into the search bar: nobody searches "necklace" on its own when they want a new baby gift for their sister. They type the whole phrase, because the phrase is what says what they want.
Marketplace search engines followed. Powered by NLP, natural language processing, they now appear to score how well the full query matches your fields rather than counting words in common. Etsy doesn't publish the details of its algorithm, but the observable outcome is consistent: listings written in buying phrases show up for searches made of buying phrases.
Why single words stopped working
Back when matching happened word by word, the tag "necklace" was enough to exist, at least in theory, on every query containing "necklace". In practice you were competing with every jewelry listing on the platform, on a word that says nothing about your product.
A single word stacks up two flaws: it describes no buying intent, and it puts you where competition is highest and your specificity invisible. A key phrase does the opposite: less search volume, but queries where your listing is an exact answer. For a small shop, that's the only ground where the match is winnable.
What it changes for your tags
You get 13 tags of 20 characters each. Twenty characters is the size of a short phrase, not a word. That's no accident: use the whole budget.
A fictional example, a silver crescent moon necklace:
Before: "necklace", "silver", "moon", "jewelry", "gift", "women". Six tags spent, zero buying intent covered.
After: "moon necklace silver", "crescent moon charm", "celestial necklace", "moon gift for her". Four tags, four plausible searches, and nine left to cover other angles: material, occasion, recipient, style.
The practical test: read each tag out loud and ask whether anyone would type it as is into the search bar. If the answer is no, replace it. The full method is in Optimize your Etsy tags.
What it changes for your title
Same logic for the title: each segment should read like a plausible query. "Moon necklace sterling silver, crescent pendant, birthday gift for her" strings together readable key phrases. Stuffing synonyms separated by vertical bars belongs to the single-word era: engines understand phrases, repetition adds nothing. The detailed structure is in Etsy listing titles.
Where to find your n-grams
Three free sources, in order:
- Etsy's autocomplete. Type the start of your phrase and note what the bar suggests: those are real queries, already worded the way buyers type them.
- Your stats. The queries already bringing visits to your listings are validated n-grams. Also look at the ones showing your listings without generating clicks: often the title fails to confirm what the query promised.
- Your customers' words. Messages and reviews carry the spontaneous wording of your market, phrases no tool would have suggested.
Updating without rewriting everything
No need to redo fifty listings in one weekend. Start with your three to five best listings, swap single-word tags for phrases, realign the start of the title, then watch for a few weeks: listings get reindexed, but not instantly.
If you want to speed things up, Fyncha rebuilds title, 13 tags and description in English and French from key phrases of 2 to 4 words, with a diff before anything is applied: you review, adjust and validate field by field.