I have noticed something different in the past week. For a few shops, I am seeing the tags moved up on the listing page-- to right below the photos on the left side (above the reviews.) Section header says "Explore more options" then shows 14 tags. The fact I am most interested in with this change(?)/test(?) is that there is apparently also an increase in character limit with this change. I noticed it right away when I first stumbled across the change, but didn't count the characters until I saw it again today. The longest tag for that particular listing was 28 characters.
Anyone else seeing this? I don't know how I feel about the elevated tag location, but I would like to have more than 20 characters to use for my tags! :)
The phrases they use for this test seem to be ones that buyers have entered into the search bar, which would cause this item and similar ones to be included in searches for that type of product. That's as far as we can tell. They are called "phrasal tags" and are related to search activity, gathered from site-users click data.
Etsy won't be generating those phrases - they will be ones that are derived from actual real-time click data generated by users of this site.
The point of it seems to be to direct viewers to a more specific set of "you also might like these" products from around Etsy. It is like a word-based version of the images they show under our listings now, but much more product-refined & related to the actual type of product the viewer might want to see.
The purpose will be to increase sell-through rates across the entire site, not for any one specific shop. Similar to what those product pictures we currently see do.
The "improvement" for buyers is that they allow the viewer to have more control over what they look at next, instead of only seeing products selected for them by Etsy.
The improvement for Etsy is that it will likely teach the AI algorithm more about what people actually want to look at after they open an existing product page, whether as the result of a search or not. It is a good way to gather additional data about buyers' intentions, data that is actually entered by buyers themselves through their click activity.
In that way, AI learns that people who opened Product A actually want to see a specific type of product grouping. This tells AI more specifically what items to include in searches whose results might include the product page the viewer originally opened. Other data can also be gathered, such as which category of products the viewer selects from.
This way of displaying tags will give Etsy a lot more info than simply showing a list of tags the seller chose to put in a listing. We all know that seller inputted tags can sometimes be way off point, not the "best" tags for the product as far as buyers are concerned, or simply a fishing expedition by the seller trying to cover all possible searches.
Tighten up your category choices, titles and phrasal keywords folks! This is a good demonstration of how AI algorithms are "taught" to work to make connections that in turn populate search results pages.
Really, as soon as Atrributes were introduced, the writing was on the wall for tags. I'm not at all surprised to see that they are being superceded with an AI-based solution to help buyers locate products as they move deeper into the site.
The way Etsy uses tags is not in line with how the rest of the retail online world does - it is a very strange system that is really just a hangover from the days when keyword density was a factor in ranking - which ended a long time ago.
Allowing sellers to put their own tags in, then using those self-entered tags as a ranking factor site-wide is just silly. We all know how tags are mis-understood and misused on this site, so Etsy including them in ranking is just going to corrupt results. I'm really happy to see they are being phased out.
The next logical step in AI devlopment (once tags are out of the way) is to look for confirmation of meaning within the body of the description, so that the important places for keywords will be the title and the description - especially the top paragraph. This is how it is done across all major search engines, which is the direction Etsy is moving in.
In a talk last year, Josh said that they were using ALL parts of the listing data, including the description. From what we can tell, that is happening already in a "test" way so far. I guess as the test for using phrasal tags under listings progresses and returns more useful data, the algorithm will start including more data from the description at about the same rate tags are retired.
That's the "normal" way of rolling this sort of change out. The algorithm needs "confirmation" that it is making the right connections about keywords, so it would seem the places Etsy is going to use to gain that confirmation is Title & Description, with Attributes as a lesser input. (Attributes have other functions that just keyword confirmation.)
This is a very positive step - one that will help search results be better. Josh has been talking about this type of thing for a while now, if you read what he says in the context of how AI algorithms work in the retail environment. And of course Rachel Glaser has made many comments in this direction as well.
Once Etsy ignores sellers' self-inputted tags entirely, visibility of products in search results will depend on sellers using the correct keywords and phrases in their title & description - ones that are relevant to that one specific product. No more keyword stuffing or trying to attract every single person to a listing whether it is what they are looking for or not.
Sellers "lose" nothing as it is not difficult to build relevant phrasal keywords into descriptions, & buyers can search within each store to find more things with keywords relevant to themselves if they want to. As a result, the general site-wide search will start to function as a search system should - once all the seller-tag-influence is removed.
If you look at the source code, Etsy calls those links "seller-tags-experiment". They've replaced our public tag links with different links (you cannot see the real tags at all now on those listings, not even in the source code), & are calling it a tag experiment. The list of links is called a tag list in the source code. Yet if I find a listing with those "tags", I am seeing it everywhere, unlike many tests they run here. But, it isn't on every listing.
While they may be searches buyers have done, they are also most likely searches that the machine learning process has already associated with that listing. Looking at my jewellery shop, many are not exact phrases I entered in my titles or tags, but I am ranking well for many of them, even when searching incognito. I've been doing ok for those searches for a while now, & I didn't even use the exact phrases. The more popular my listing, the more generic that "tag list" is. The less popular my listing is, the less likely it is to have the tag experiment at all. I can guess which of my listings will have it with about 80% accuracy.
So, this experiment appears to show you which items have received a lot of positive search interactions, and what terms they do well for. It's very useful data for sellers.
just wanted to update that I am not seeing this experiment today on one browser, but still am on the rest.
If they are using terms and phrases that actually fing the listing then that sounds like a good thing. Could be better than we sellers guessing at tags to use.
"If they are using terms and phrases that actually fing the listing then that sounds like a good thing. Could be better than we sellers guessing at tags to use."
But what if the tags that are used do not apply? One of my Great Blue Heron prints has 'tags', however one of the tags is "white heron bird art".
'White' is incorrect in any sense and I'm not sure where that is coming from. I suppose it is a suggestion in case someone doesn't know what they want? At any rate, the tags pull buyers right out of the shop into the vastness of Etsy land - not sure that is a good thing. But at least they are below the fold (and the description) although above the reviews.
@ZehOriginalArt thank you! I have been looking for an example like that. Just to make sure, you don't have "white" accidentally selected as a colour attribute, do you? The item does come up in your shop in a search for "white", so that would also establish that these "tags" are searchable. words added to the listing. I see you have the word "white" in the description, but you have it in many other listings as well that are not searchable, so that is not likely the reason for its appearance in the new tags. Hard to be sure, though, since not all listings have the tag experiment.
I am sorry that your listing is now incorrectly tagged, but I really appreciate the example! I couldn't find one in my shop & got tired of looking LOL
@ZehOriginalArt, if you think about that example in the way Etsy does, having that tag there is "correct" for Etsy's purposes.
I have said many times that Etsy and Sellers are NOT aligned on the outcome they want from search functions. You are looking at it from a seller's perspective, which absolutely is not the viewpoint Etsy comes from.
As a seller of a "blue heron" item, you see "white heron" as a search link to be incorrect. Well, from Etsy's perspective it is perfectly fine and a useful link to display to people looking at a "blue heron" listing. You don't know how many people start their Etsy journey by looking at a blue heron pictures, but end up buying a white heron one - but Etsy DOES know that. Your assumption is that all viewers want to see blue heron pictures - but that is just an assumption on your part. They may be just as happy to see any heron pictures at all - and it seems in the case you refer to above, that is indeed what click data has informed the algorithm is the case.
Also, in the phrasal keywords shown the link to "white heron" is only one of many. A viewer who doesn't want to see white herons will not click that link. Those viewers who DO click that link obviously want to look at white heron pictures. If the viewer has got that far down your listing page without purchasing your item, then etsy wants them to have choices of what to look at next. The hard truth is that if the viewer clicks away, they weren't going to buy your item - so Etsy wants them to buy something else.
Etsy's goal is to make sales anywhere on this site, they do not care what shop an item sells from. The search algorithm is designed specifically to achieve Etsy's purpose.
As a result of how AI calculations work, Etsy now has a very good understanding of BUYER behavior, one that is based on cold hard data gathered from buyers' own click activity. That will apply to at least 80% of buyer actions on this site - and that is the number Etsy is interested in. The data is constantly changing in response to real-time site activity, so it is updated to capture all emerging behavior immediately - it is unbelieveably responsive & flexible.
Sellers' goals are to get sales within their own shop only. We do not want buyers being led off to a competitor's shop. That means that what we want is vastly different than what Etsy wants to achieve. Seller behavior is very easy to understand from Etsy's POV. We sellers all want the same thing - to sell our stuff. Nothing complex there. What we 'want" from Etsy isn't really important in the big scheme of things.
There is no way of reconciling these two totally different goals. That is how it works on marketplace sites. The only way a seller can channel their own customer activity is to have your own website. All marketplace sites are designed to grow the SITE, not individual shops.
I am involved in a couple of development groups for AI algorithms in retail. The way sellers/retailers look at this stuff couldn't be more different than the way site owners look at it. Buyers look at it the same way site owners do, which is one of the reasons why Etsy is growing its sales and buyer-registrations so rapdily as it makes all these changes. It is only sellers who don't like it, because we have a different agenda that is focussed on our OWN bottom line.
When Etsy adds the word "white" to your listing when it is not correct, & makes the word searchable, it is definitely a concern to everyone, buyers and sellers alike.
@LandandSeaGems I think that is one of her actual tags, according to what I see in this browser (where I don't see the "tag experiment" today, so can see all the tags). If those aren't actually your tags, @ZehOriginalArt I apologize for assuming.
@CindyLouDoesStats @ZehOriginalArt Yep, you’re right. Sorry. I think I need to quit while I’m (not) ahead.Lol. Spent the last 3 days fighting with the now non functioning Etsy search on Ipad and missing navigation bar, trying to make some purchases and I think I’m just frustrated. I hate the Etsy app but guess I have to use it now.
OK, yes I do use white as one of the colors in my attributes for the heron print. Colors are - primary color: "gray", secondary color: "white". So that is probably where it is coming from. I used those two colors since they were the closest I could get to the heron colors. Geesh. If they decide to keep this, I'm going to have to be really careful about attribute colors, I guess.
On the pink orchid painting, 'Hawaii flowers' is one of the tags so all is good there!
Quite honestly, there is absolutely no point in worrying about all this. especially while it is in test mode - what you see today may or may not be seen tomorrow.
Sellers could instead just get on with the business of making sure their titles and descriptions are utilizing the best, most specific keywords for the item within the listing. Get category choices nailed by researching in-site. Use Attributes where you can - where it makes sense (and don't use ones that don't apply because that's as bad as entering wrong tags currently is).
Beef up the resolution and pixel count (and number) of your product images. Redo listings that are dormant as far as activity on them goes. Delete products that are just never going to sell on Etsy. Increase the phrasal keyword count in your About page. Prepare drafts of listings for new stuff that is suitable for the next event in the annual retail calendar so you can swing buyers your way during that cycle. Work on your next newsletter or social media campaign.
There is so much for sellers to do that spending any time on what is likely going to be a test that evolves over several months before anything is actioned site-wide, is a complete waste of time.
@onedaysgracenz I’m actually posting from buyer mode....I hope not too many buyers have experienced what I have for the last few days while trying to shop here. I can roll with the seller stuff but I hate to see anything that might impede a buyer from purchasing. I have been trying to find and purchase 2 paintings for 3 days now.....I’m away from my shop and using my IPad.
@LandandSeaGems there's no denying that search here is a hit & miss affair right now! That's why I am so pleased that seller-inputted tags seem to be on their way out - hopefully that will be a factor in improving the quality of search results.
The other side of my optimistic viewpoint is that it is likely we won't see "quality" search results before 2020. The AI learning curve is lengthy and it's only really been 18 months so far. On top of that is Etsy's migration to the cloud, also expected to take until 2020 to complete. My guess is that will be late 2020, not early next year at all.
It's definitely a work in progress and the whole renovation of the site has been disruptive for sellers. IMO there was no choice for Etsy, they had to bring the site into the age of technology that now drives the internet. A painful process for sure, but the signs are improving now so I think we are over the worst of the impact on buyers. Etsy's published data seems to suggest buyers are happy in general - although of course there will be a % who don't fit into the box. Some buyers will give up on Etsy too of course, but overall it seems more new ones are replacing ones who leave at a faster rate.
The impact on sellers is another matter. As I have said many times, not all sellers will survive the way this site is heading. All mall renovations mean some shops are literally driven out of business, and I guess we all hope it won't be ours.
So I just noticed this evening this was going on. My initial reaction is total frustration. A customer is immediately directed to other options and clicks leading them right out of the listing/my shop. You pay for extra advertising promoting, get them to your shop and poof etsy immediately gives them other options to other shops. I do not complain much but this really erked me. I understand the need to do different things/increase customers/competition from Amazon ebay etc. I didnt even complain with the increase in fees ( that had not been done before). But this was very frustrating. I like many others do this as a business. I work a full time job for medical but etsy I have remained commited to. This one is very annoying and I do not think is fair/I hope they do not maintain this after the test. Put it another place on the page but leave my description below the listing. Dont make the consumer go hunting for my details .
I agree 100%. IMHO photos and item descriptions should have the prime real estate and everything else should be farther down or on a separate page with a clickable link, if space is an issue. Customers aren't buying reviews or keywords from us but items and they need item specifics more than other information.
I've been seeing this in my shop too the last few days. It only shows on my more popular listings with alot of views/sales. The listings that have had less activity do not show the new 'tags' under the image....