My Photo

CSE Strategies (Sister Blog)

« Washington Post article on eBay feedback changes - what do *you* think? | Main | Interview: Exclusive interview with 'BMX' - the seller that hacked BestMatch »

April 10, 2008

Sellers are learning how to 'hack' BestMatch - is this really good for eBay Buyers?

Outside of the World of eBay in the wonderful world of the Internet, google's algorithmic search engine is constantly being gamed by those with good intentions (white hat), unknown intentions (grey hats) and devious intentions (black hats).  This is broadly known as search engine optimization or SEO.  Google and the bad guys are locked in a never-ending arms race where google works constantly to stay one step ahead of linkfarms, cloning, cloaking, linkspam and a whole dictionary of strategies designed to confuse the google bots and increase a website's organic traffic.

Back in our eBay World, as you probably already know in March, eBay switched the search engine from a reverse chronological (aka 'ending first') system to a black box algorithm known as BestMatch.  BestMatch has many positive aspects and seems to be working as we've noted.

However, it only took about a month, but now we're hearing from sellers that are upset that other sellers are significantly gaming the system.  I received a great example of this from a long-time reader and video game seller today that blew me away.

Here's the example of the term "nintendo wii" with BestMatch results.  I've numbered the top 6 results for us to discuss (you may need to click on the below to see the details):

Bestmatch_seo1

What do you notice about results 1, 2 and 6?  Well, let's just say these are very new systems.  Why on earth would a seller put the word new into a title 5-10 times and use that valuable space on a repeated word instead of something like "includes 5 controllers and 4 games"?

Yep, you guessed, it, doing so appears to increase your BestMatch score significantly.  Let's assume that these sellers have simliar performance histories as they all have around 300 feedback and that's a level that's hard to mess up.

Now look at the prices and the ending times - two other heavily weighted bestmatch inputs.  Item 2 ends in 13 minutes (the longest time in these 6), yet is rated second. It's price is clearly higher than item 5.

So if it isn't performance or price or time, then why would item 2 be ranked above the less expensive, more temporal listings 4 and 5?

It must be the title!  By including a whopping 12 NEWs in the title this seller has been able to get much more exposure for the item and has cleverly filled in the missing title information occupied by all those NEWS in a subtitle.

In the world of computer science we call this ability to take advantage of an unforeseen/anticipated aspect of an algorithm a hack.  Now we are officially seeing the kick off of what I think will be a long and arduous (sometimes very profitable) saga of BestMatch hacks.

How can YOU hack BestMatch to your advantage
The best thing about this is eBay has given us the tools to hack BestMatch.  In their research lab area, as we've mentioned before they have a very simple utility called the Bayestimator.

Here's how to leverage this tool:

Step 1:  Enter in your boring old original descriptive title that a buyer would understand and get your scores (here we've tested "New nintendo wii includes controller and 5 games:

Bayestimator_1

Overall this title earns a 40% score and the words that eBay likes are: "new 5 and games"
eBay gives zero juice to nintendo, wii, and so those are somewhat wasted words in the world of BestMatch.

Now that we know new is popular let's try a new title, let's say something like: "nintendo wii console new new new new new new" (ok i controlled myself to 6 'news').

Bayestimator_2

How about that we went from an overall score of 40% to a 100% - eBay's BM algo loves this new title and in fact I have 11 more characters available.  Guess which word would be best there?

White hat, grey hat, or black hat?
As far as know, aside from being a) annoying to look at+read and b) highly repetitive, these titles don't violate any eBay policies so I think that automatically would classify this kind of gaming as grey hat.

The BestMatch race begins, can you afford not to play?
I'm sure eBay has already noticed this and is working on ways for BM to be smart enough to actually discount the repetitive use of potentially high scoring words (just like google does when you spam the heck out of metatags).  But, by the time they roll that, sellers will be on to the next hack and so the game continues.  Here's how it's different.  In the google world, you may have one or two results pop in above those that are more relevant so it's a minor annoyance but doesn't really impact anyone that negatively.  Also on google you can always use adwords to get your message out there.

What's tough with these eBay BM hacks is that many sellers are going to feel economic pressure to rip off the white hat and put on a grey hat.  What if one day you notice your top selling item's conversion rates are dropping?  You do some searches on eBay and realize all of your hard work on DSRs, title optimization, great pictures and low prices are being nullified by someone doing some BM hacks.  You can report them to eBay and they will ultimately fix it, but in the interim, your only strategy is really a 'if you can't beat them join them' approach.

What about the buyers?
Well, I think we can all agree that the results for 'nintendo wii' with the BM hacks have actually become much lower in quality than in the 'ending first' world (you can't hack time), so it will be interesting to see what kind of hack-curve eBay faces here and how fast can they get to a BM 3.0 /4.0 that eliminates these kinds of things and truly offers a long-term better buyer experience.

Report your BestMatch hacks.
Have you seen/utilized any BM hacks? Do you feel pressure to go grey hat?  If so, let me know.  While this is a new (new new new new new new new!) eBay issue, I'm sure we haven't seen the last of it.


TrackBack

TrackBack URL for this entry:
http://www.typepad.com/services/trackback/6a00d83451d7ed69e200e551c656df8833

Listed below are links to weblogs that reference Sellers are learning how to 'hack' BestMatch - is this really good for eBay Buyers?:

Comments

I can vouch that best match does not work well enough yet to be live.

We have 100% feedback and DSR of 5 or 4.9, we are a good ebayer... We use to spend £15 on all our ads to have them highlighted and featured, and its just money down the drain now, adverts that cost 20p do just as well if they are optimized, so we are learning the hacks and not going to spend the money on featuring any adverts anymore... i really hope ebay feel the pinch too from this, with standards as low as they are at ebay they dont deserve to do well.

what is type pad all about

wow, too funny. eBay really have their head up their ass don't they? i have no faith in their algorithm if they didn't anticipate this kind of issue.

wow, you dont normally write this sorta stuff Scot. I am more looking at you with my jaw open than that crummy best match. Do you think ebay labs is all in a huff with their pouty notice on their best crapper tool? They seem a little miffed. You may have caused them to come into work early! hugs

Bayestimator has now been pulled by eBay!

Bayestimator has now been pulled by eBay!

wow, they should really go back to the reverse chron.

wow, they should really go back to the reverse chron.

lol, it seems maybe I can recycle my 1999 search engine optimization tricks. here comes the invisible text and the doorway pages!

What I have notice is sellers abandoning their poor fb "disadvantaged in best match level" DSR ids for new ids. A long time media seller at almost 100,000 just disappeared...but some looking finds their inventory being listed on a different id while negs pile up on the old one for "no response/no item". The whole seller performance thing seems to be a farce as long as one can just sign up for unlimited ids.

Hi Scott,

I figured this out about 3-4 weeks ago a couple of days after BM was rolled out and used it in other categories without much fanfare but with much success.

For the Wii listings I decided to use it maybe 2-3 weeks ago.. People didn't seem to notice it until I used it on the Wii listings and it caught like wildfire then variations such as 5 5 5 showed up.

I kept hush hush about it to use it to my advantage, but the trick worked all too well. I agree it is slightly underhanded, but so is grading us on a 'match' criteria which at times makes absolutely NO SENSE at all to the seller. Sellers are forced to do whatever works to get ahead of the competition.

This is our lively hood!!!

Who's great idea was it to roll out BEST MATCH without a full rollout of the seller dashboard? I equate it to driving around at night with my eyes closed. Its terrible. Ebay needs to be more accountable and not shove changes down our throat that can turn your business upside down overnight. Sellers are shaking in their chairs wondering if the next ebay 'policy change' will be the final blow which squashes their business.

As far as the previous comment I agree, I was amazed that coding against duplicate words wasn't thought of. (I was once a programmer myself). Not sure how the system is setup but my feeling is that parsing the title string and eliminating duplicate keywords would be a simple subroutine or module.

I also find it interesting how much people pay attention to what the competition is doing and how quick they are to exploit it. You truly have to be on the cutting edge because as soon as you find a product/title/etc that works, 50 people are on top of it and crowding you out of your space. Very cut throat.

Is ebay hiring consultants? I'd like to apply.

LOL

Look forward to see ebay's response to this.

You did a good job of explaining the trick, by the way. Kudos.

Scot,

This is the second time I have noted in your blog that you state best that match is working. Working for whom? Working for Ebay or working for the average seller? Can you clarify? I have heard many sellers say their sales actually dropped, and not only dropped, dropped considerably since Best Match was implemented.

And the term "working" is sticking in my head. You say it is working but yet you write an article about how it is being gamed. In my opinion it is not working and your article cites only one but one very valid reason why it is not working. It is supposed to be BEST MATCH, not BEST MATCH FOR THOSE WHO KNOW HOW TO GAME THE SYSTEM.

Don't get me wrong, I realized kinks need to be worked out, but you would think someone who is smart enough to write a highly technical algorithm would have anticipated this (the gaming of their system). I buy and sell on Ebay, and I am quite amazed at some of the returns I get when using Best Match. These are returns that have absolutely nothing to do with what I am searching for, and it is not an improvement over the previous means of searching.

My purpose of writing to you is not to criticize, but in hopes that your readers will get some detailed information about how best match is working and some success stories, not just data.

Does best match factor subtitle into the equation?

Hey Bart,

Well you'd think so, but I'm sure hack-proofing this was something got pushed in the haste to get it out. Now they get real-time QA on what hacks will come out.

Here's another one that I saw tonight:
http://search.ebay.com/search/search.dll?from=R40&_trksid=m37&satitle=nintendo+wii+new&category0=

Looks like putting numbers in there works well 2 2 2 2 2 2 2 2 2 2 2 2

Scot

Wow. You would think that as soon as you came up with the idea to assign a point value to a keyword and then rank according to the cumulative points in a title, the very next thing you would do is code against duplicate words. It seems sort of obvious doesn't it?

Verify your Comment

Previewing your Comment

This is only a preview. Your comment has not yet been posted.

Working...
Your comment could not be posted. Error type:
Your comment has been saved. Comments are moderated and will not appear until approved by the author. Post another comment

The letters and numbers you entered did not match the image. Please try again.

As a final step before posting your comment, enter the letters and numbers you see in the image below. This prevents automated programs from posting comments.

Having trouble reading this image? View an alternate.

Working...

Post a comment

Comments are moderated, and will not appear until the author has approved them.