@johnrj1226 wrote:In 2013 eBay switched from Google to Cassini as their primary search engine - IMHO this was the largest "faux pas" or just an old fashion screw-up that eBay has made. Out of the box they had issues - recall having to redo the wording of my listing titles at least 3 times some my titles would better fit Cassini programming as stated by eBay (vs being a more seller friendly approach of adjusting their programming). Then I had over 100 items listing & sales were approaching $10K/year but sales slide down hill ever since to almost non existent - what the hay!
When I do/did a search on eBay I will/would a ton of not even closely related items to what I am looking for. 6 years later not much has changed. At least Cassini is consistent. LOL
eBay switched from Voyager (literal search) to Cassini (relational search) - Google is its own search engine - eBay buys space on google (well, they did, they're reducing ad rates). The migration from Voyager to Cassini was very messy, I agree - for a long time they had a kind of frankensearch with Cassini at the front and Voyager as back end. I'm not sure they've ever figured it out, though, but I seem to find most of what I'm looking for as long as I don't use worst best match.
@eburtonlab wrote:
EBay's AI uses an extension of the system that their API's call "Taxonomy". This is the system that the AI uses to try to build a profile of what it "thinks" the buyers will be searching for.
This means other sellers search results do NOT match yours. Nobody here will have the same results as you. Because that's how their system is designed. None of us here have results that match the next person.
I agree with your first paragraph, but I am not convinced that your second paragraph is necessarily the case (yet). While eBay is trying to model what users are searching for when using given search terms, I am not sure that eBay builds a unique model for each potential buyer -- I think it is much more of a statistical model applied across the board rather than customized to each individual.
Obviously I can only see my own search results, and the search results I obtain when I am signed out of my account. But I believe (perhaps I am mistaken) that if I specify a search by providing exact parameters including starting location for postal calculations and sort, then others should see the same results (minus any changes over time as items are sold and new items are listed). This is one of the reasons I include search URLs to illustrate the points I am trying to make rather than just listing keywords.
If every user sees unique search results, then almost any discussion about search results is pointless and I do not believe that.
For instance, when I go to ebay.com and type the following words into the search bar: pee wee herman and then I hit search, this is what I see (using "highest first" as my sort order and worldwide search as defaults, and an 02215 zip for postal calculations):
I suspect that if you follow that link you will see some 1.363 million results, the top currently being $1.1 million Joyride Halo figures that have no connection to Pee Wee Herman. If you scroll through all of those results, the most expensive Pee Wee Herman item I can find (using that spelling, but a different search) is a $2400 bicycle.
Well for that example, it's sorted by price so of course the highest priced one is going to come up first. If we switch to best match is where things would differ.
Part of the reason it "has" to differ, is because handling time and shipping location do make a difference in search. EBay tries to include at least a few "Guaranteed Delivery" listings, and prioritizes some that could get to you faster.
Also on the topic of sorted by price, eBay does have a "bug" where items don't always appear even when sorted by price. (I've shared screenshots of this on the forum, if you'd like I could look in to my history and find a link). It's intermittent, which again, is another example that one users experience does not matter much when compared to others experience.
And BTW, I also provided that information about the above mentioned "bug" to an eBay Sr Manager in person, who said not only that they would look in to it, but that they heard the same from other sellers they worked with.
Now if I take that exact search you linked, and then only change it to "best match" and touch nothing else, we can compare results.
I zoomed out so you can see more results, and here's a screenshot of what comes up for me:
Maybe share your results and we can compare?
(Edit: And for the record, to see how many of those results actually had the search phrase, in addition to where the first one on the list appears... combined with how many "sponsored listings"... It's quite comical how people can say search is "working well".)
