On machine learning and other Etsy search changes...

Hold on tight... Etsy is a sea of changes

Sometimes it feels like there is a sea of change at Etsy. I know that I felt that way earlier this year. Since then, there have been further transformations afoot.

For one thing, Etsy stated that they are doing even more testing than in the past. They use the term "experiment velocity."

Etsy has not announced these changes per se. But they are happening.

Etsy has added many new search features:

• Item attributes can now be searched in some cases. (Make sure yours are all filled out.)

• New filters are available (items on sale, free shipping, some attributes on the left side).

• Badges may appear for things like "best seller" and what they call a "scarcity".

• Guided search, which are orange-toned buttons that appear at the top of some broad searches.

Some changes to Etsy search

• Shop feedback (star ratings) now display in search results.

• Similar items buttons are under search results (only show up if your cursor is over that item).

Artificial intelligence & search

Etsy has begun to use machine learning technology ("artificial intelligence"). That is why Etsy purchased the company Blackbird Technologies last year.

What is Etsy's algorithm?

The goal is to determine the intent of a buyer when making a search. (Here is more about the related "Context Specific Search Ranking.")

In the past, search results were returned to shoppers that best matched their exact phrases in tags and titles. And the best "quality scores." 

Items used to be shown one per page per shop as per the "fairness" part of the old algorithm.... but that has changed...

Some things have been removed

Gone is the "shop diversity" element. There is no longer a limit of one item per shop per page of search results. "Clumping" is now the norm. This is a big change for some shops.

Also gone is the opportunity to test search to see how your items place. Why? As mentioned above, your search results are becoming contextualized and personalized. Results are different for each buyer. 

As artificial intelligence is self-learning, the personalization is still developing. It is learning to make connections between the terms that searchers use and what their real intent is. (For example, when someone types in a word, what is it that they eventually buy?) Etsy stated that factors include search history, user data and "as a first step" sales data. It is far from perfect, but you can expect it to get better with time.

Any improvement in Etsy search is good news, right?

What is next at Etsy?

We do not know. And there is no recommendation to make any major changes to titles or tags. As always, use a variety of long-tail keywords. Use accurate titles and tags. Those are the best ways to capture different queries that buyers might make. And ultimately sell your items.

(And, of course, offer cool stuff.)

Etsy is working on more interesting ("revolutionary") discovery tools... So plan on more changes!