How Etsy ranks your listings: query, quality, context
You swap a tag, rerun the search, and your listing has not moved a single place. Three weeks later it shows up, on a phrase you were not even targeting first. Nothing irrational about it: Etsy search does two separate things, and most SEO advice mashes them into one.
Stage 1: getting picked. Stage 2: getting ordered
When a buyer types a query, the engine first builds the set of listings that match it, then puts that set in order. Two separate operations, with criteria that have nothing to do with each other.
The consequence is blunt and useful. If your listing is not in the set, no amount of work on your shop's quality will pull it up: it is not at the bottom of the list, it is not on the list. And the other way round, getting into the set on a heavily contested query guarantees you nothing if your listing has not proven anything yet.
It also explains the delays. You can rewrite your copy in a minute. The track record your listing has built up, you cannot.
Query: your copy decides the set
Getting into the set comes down to what you write: title, tags, attributes, category. That is the part entirely under your control, and the only one.
The classic trap is aiming too wide. A "necklace" tag drops you into a set of hundreds of thousands of listings where your shop has no reason to surface. A phrase like "silver name necklace" builds a much smaller set, one where your 40 sales actually count for something. That is the whole point of thinking in two to four word phrases rather than single words.
Same logic for attributes and category: they describe your listing in a closed vocabulary and make it eligible for filtered searches where free text is not enough.
Quality: what your listing has done so far
Once the set exists, it has to be ordered. Etsy publishes no formula, and nobody knows the weightings. The principles discussed publicly come back to one simple idea: the engine watches what buyers do with your listing.
Do they click when it shows up? Do they buy after clicking? Does the shop hit its delivery times and keep customers coming back? Those signals build over time, which mechanically works against a brand new listing and rewards an established one.
In practice, that means your title, your main photo and your price are not only there to convince the buyer. They also feed the signal that decides your position next time around.
Context: two buyers, two result pages
The last layer, and the one most often ignored: two people typing the same query do not see the same page. Location, language and browsing history all play a part.
Which means "I am in position 3" does not really exist. Your position as seen from your own browser, on your own shop, is the least reliable measurement you have. Your shop stats tell you which queries actually bring views, and that is the only figure that counts.
What this changes in how you work
- Pick your queries before you write. Aim at sets where your shop can exist, not at the biggest volumes.
- Separate the two jobs. Copy to get into the set, the full listing to climb inside it.
- Never judge from a search of your own. Look at the stats, across several weeks.
- Change one thing at a time. Rewrite title, tags and photo on the same day and you will never know what caused the shift.
None of these steps pays off in 48 hours, which is exactly what makes the subject tiresome. The good news is that half the work, the copy half, gets written once and stays put.
If that copy block is the one that has been sitting on your list for months, Fyncha proposes a title, 13 tags and a description in English and French from your existing listing, which you review as a diff before applying it field by field.