This started out as an exercise to evaluate the influence of the various factors in the Best Match search and sort algorithm for an eBay listing of the book How to Do Everything With Your eBay Business by Greg Holden. It yielded some very interesting results.
I listed a copy for sale a few weeks back using the suggested listing title and the pre-filled Item Specifics information based on the book’s ISBN.
I checked on the listings placement using the book title as the search query and saw that of the five books for sale on eBay.com, I was dead last in the sort order.
I thought that this would provide an interesting opportunity to experiment with the various Best Match factors to see if I could determine which ones have more influence over the others. Specifically, I wanted to see how DSR scores, shipping charges, listing type, and keywords stacked up against each other.
I learned that (at least for this search query) DSRs are weighted higher than shipping charges (specicially free shipping) which are weighted higher than keywords. It was also clear that in this category (non-fiction books), Fixed Price listings are weighted higher than auction style listings – even if the listing has a bid.
What totally surprised me was that it appears that time is a factor influencing fixed price listings (newest appearing at the top) and a significant factor at that!
Here’s my listing in all of its glory at the bottom of the search results…
The listing in the number one position is a Power Seller with a feedback rating of 23,484 (99.6% positive) with DSRs of 4.8 (Item asDescribed), 4.8 (Communication), 4.8 (Shipping Time), & 4.8 (Shipping & Handling Charges). The listing closes in 25 days. BIN price is $1.00 with a $3.99 S&H fee.
The seller in the number two position has a feedback rating of 184 (100% positive) with DSRs of 4.9, 4.8, 4.8, & 4.7 respectively. The list price is $5.00 (fixed price) plus $2.58 shipping.
Number three is offered by a seller with a feedback rating of 658 (100% postive) and DSRs of 4.9, 4.9. 4.9, and 4.8. The list price is $3.45 (fixed price) plus a shipping price of $2.58.
Item number four’s seller has a feedback rating of 453 (100% positive) with less than 10 DSRs in the past year. The asking price is $18.99 (fixed price) and free shipping.
The user ID I used for my listing in 5th place has a feedback of 144 (100% positive) with less than 10 DSRs in the past year. My asking price was $3.99 (fixed price) with $4.00 shipping.
Optimize The Title’s Keywords
My first action was to head to the BayEstimator in order to optimize the keywords in the listing title. All of the five listings had the default title eBay suggests for that ISBN, except the listing in the top position which had shortened it to include just the title of the book. The BayEstimator suggested that the keywords “book,” “for” and “U” would increase my title’s score in the Best Match algorithm.
I revised the listing to modify the title and waited for my listing to jump up in the search results. After a few minutes, the new title was live and…
No change in position. That meant that some factor in the listing above mine carried more weight than the optimized keywords in my title.
Free Shipping
OK. Time to revise again. Like my User ID, the listing above mine did not have 10 DSRs in the past year, so I thought it might be their free shipping that was providing a competive advantage over my listing in the Best Match search results.
I dropped my shipping down to free and in a couple of minutes I had moved to the number four spot.
Therefore, free shipping is more significant than keywords in Best Match.
Note also that while I bested the other listing that didn’t have an active DSR history, I did not rank higher than the sellers that have an established DSR history and good scores. Therefore, I presume that as a factor in Best Match, DSRs are weighted higher than free shipping.
Cheap Shipping?
Now I wondered what benefits under Best Match cheap shipping would provide. I revised the listing so that shipping was only a dollar – a figure well under the average for the listings returned in the search results.
The result… Back on bottom.
Therefore, cheap shipping appears to provide no benefits under Best Match - only free shipping. (Caveat – expensive shipping relative to the average in that category begets penalties in Best Match.)
Auction Style vs Fixed Price
Having another copy of this particular book handy, I decided to post another listing. This one would be in the auction format and would have the same title as my fixed price listing and boast free shipping. I placed the opening bid at 99 cents and choose a three day listing so that it would close at approximately the same time as my fixed price listing.
I was rather surprised to find this listing at the bottom of the pack. I thought the free shipping would carry more weight than the listing format, but it did not. My 99 cent auction listing was firmly lodged below the $18.99 fixed price listing. Therefore, in the non-fiction book sub-category, listing type matters and fixed price listings carry more weight than auction style listings.
Even after the auction received a bid, it was still relegated to the bottom of the heap.
Let’s try one more thing…
Still interested in the relative weight of the title keyword factor, I wanted to post another fixed price listing for the identical item with my old (unoptimized) title keywords. My hypothesis was that the two listings would be next to each other with the optimized title ranked above the original title in the search results.
I was a little shocked to see my most recent listing ranked number one – beating out the listings offered by sellers with high DSR’s!
What to make of this? My best guess is time is a factor in fixed price listings – only reverse of what we are used to with auction style listings. Specifically, fixed priced listings posted more recently (or just ending later) are weighted heavily in the Best Match search results.
I’m interested in your thoughts and observations regarding this theory.
Why did this listing make it to the top?
(One final note – There were no multiple quantity listings in this case study, so the recent sales factor of the algorithm was not evaluated.)
In summary, although keyword optimization might be the easiest factor to influence under Best Match, it carries the least weight these days. I’m not suggesting you ignore this factor. However, if you are neglecting the other factors because they are harder – you are certainly placing yourself at a competitive disadvantage to sellers who score well for these factors.
Related Posts:
Comments
RSS feed for comments on this post. | TrackBack URI
11 Responses to “Best Match Case Study: Search Ranking Algorithm Factors Ranked By Weight”
Leave a Reply
Only thing I can take away from your experiment is how difficult it is to optimize eBay listings for best match.
Once again we seem to be at the mercy of eBay’s “wisdom”
Wow! Great information! I will start conducting my own “experiments” on my listings and see how they turn out. I’ll keep you posted.
My observation of your experiment? There is no “Best Match”. It’s dependent upon the server that picks up your listing when you post it.
Try searching at a different time of the day. Your results will be different. Have a friend do the same search. His results will be different.
You can analyze Search until the cows come home. eBay randomly adds its own relevance factors to every search performed. Note the word random. You never know when recent sales, DSRs, keywords, pricing, shipping cost, etc. are going to figure into the results.
Again, just my observations.
Unfortunately, the man who created the search algorithms for “Best Match” has left the building.
Gail,
My results were consistent no matter when I conducted the query (multiple times over a several day period).
Thank you for acknowledging my post, Doug. I believe the accuracy of your experiment results for the particular book you listed, factoring in the history of the IDs you used to list it. Based on your analysis, would I get similar placement if I listed the same book? We don’t know.
I also believe different factors are used, with different weight relevance, in different categories, and even within the same category. Collectibles categories, for instance, will return a different result almost every time. Auctions show up randomly with no apparent correlation to FP placement, nor to each other with regard to end time, regardless of eBay’s assertions.
Does placement change as DSR scores change? What about number of sales? These factors change daily. Do these factors count more heavily depending on the type of listing?
I’m just trying to point out that in any analysis of Best Match, the resulting assumption will only hold true for the set of circumstances used in the experiment.
I agree with Rich Tubbs.
you use the 30 days, so it cost more than 3d or 7d, perhaps that’s why ebay put your item in 1st postion. Your 30 days listing remain in 1st postion for 30 days ?
ssorry for my poor english
Awhile back I too,experimented with a listing to see what might increase my items placement in search results.
After making several revisions,I finally succeeded in getting my
item to the second place position .
I wasn’t happy with that because my item price was lower than the seller in the
number 1 position, my shipping was lower, we had both been selling on ebays the same # of years and my DSR ratings were higher than his.
I wrote eBays and in the reply I received , it was pointed out that my item had
been revised, implying that was why my item got a lower placement.
I even went so far as ending that item and then relisting it.
When I searched the item several days later,it still showed up in the
second place position.
[I didn't bother to followup & email eBays about it, because I probably would have then received a reply eventually that said the other sellers identical listing had been listed before mine.]
I don’t know if ebays has changed this since then.
I now make it a habit to try NOT to revise items and first search items curently listed BEFORE I list an item.
There are lots of things eBay’s has chosen not to reveal about how it decides what items show first in a search result.
Very interesting!
A scerte science behing ebaying? it appears so…
Once they implemented the Best Match science I have lost 98% of my sales. Yes, that is right. I tried auctions listings, fixed price, 3-day up to 10-day, free shipping when I could and it didn’t make any difference. I think one of the issues lie is in clicks to impressions (I think that is what they call it). If your product can get view but doesn’t sell the first couple of times around I think the computer puts it at the bottom. I didn’t have that much trouble before until they started the new Best Match. I don’t think that there is a way around it. I listened to the eBay radio one day and a top rated seller called in and he said once Top Rated Seller started his business went down – once the New Best Match started it went even lower. He was making $5,000 down to $500. Ebay has the control and there is nothing we can do about it.
Doug,
I’m wondering how long the listing you posted would remain at the top of the search.
I’m assuming it will drop down in rankings as it’s ‘newness’ wears off. Have you followed that?
I’m wondering the same thing! I’m also wondering if I can get it any higher than sixth position.