This am, I logged into my eBay account (buying focused account) and found that I had been opted into the ‘new search experience’. This is Finding 2.0, which is based on the Magellan search engine (the engine behind eBay Express). This is the engine that has been in the playground for a while so nothing new to report other than my guess is after rolling this in the UK and other countries, they are now starting to turn the dial on the USA. At eBay Live, I heard they had 10% of users in the new experience, so my guess is they have bumped that up to 20-40% at this point.
There’s some text at the top right that says: “opt out of the new search experience | FAQ”. If I click opt-out, it takes me to a Finding 1.0 experience. The FAQ is here for those interested.
What is Finding 2.0 vs. BestMatch
I’ve been asked by many eBay observers that are confused by the difference of BestMatch (BM) and Finding 2.0 (F2). Here’s my attempt at explaining it based on my understanding. BTW, another acronym I’ll use is Search Engine Results Page (SERP).
- BM – BM is a new way to change the ORDER of SERPs, not the content. BM takes into account many factors to hopefully provide a better buyer experience and generate more conversions than the historical ‘time ending first’. For example, before BM, you would search on ‘merrell size 10’ and get 84 results ordered by time ending first. After BM, the result set doesn’t change, BUT the order does. I still get 84 results, but now they are ordered by what eBay’s BM algo thinks will be best for me based on more criteria than ‘time’.
- F2 – F2 adds some more intelligence to the search engine and changes the result set, not the order of the result set (that’s BM’s job). It does this in many ways, first and best IMO is the ability to ‘pick out’ those pieces of text in my query that could be item specifics and use that to either widen or narrow my search and hopefully improve my buying experience. To continue the example, in F1 (which is a blunt text search engine), I enter “merrell size 10” and get 34 shoes because that’s what matches my text query. Actually many are “bad matchs” because the sellers put “size 10.5” and the raw text match wasn’t smart enough to know that I’m not looking for a 10.5, but a 10. Now, in F2 when I enter the same string, I get a narrower (19) set of results, but this is a much better experience, because they are all REALLY are size 10 and there aren’t any 10.5’s or other false hits that as I buyer I have to figure out and slug through.
Think of these two system as lego blocks. F2 works to improve (narrow or wider) the result set and BM works to optimise the conversions of the result set.
What does this mean for eBay?
I’ve spent a good bit of time over the last year with the Finding team (led by Jamie Iannone and Jeff King) and they definitely are sharp folks. So let’s assume they have done some really smart stuff here (tested the hell out it and proven it does improve conversions), then what we should see as this rolls out is an improvement in conversions overall (this can be distorted by the changes to listings), and ultimately from a public-data perspective, an increase in GMV/active buyer. Internally, I’m sure they measure converts/search and the effective CPM of each search, etc., but unfortunately we don’t have access to those metrics. Because the active buyer metric has a year tail, it could also take several Q’s to see any movement there. Also, you may see a decrease in pageviews on the site which if you put your ‘finding hat’ on, is what you want. If I put myself in Jeff King’s shoes, you never want a buyer to have to go to page 2/3/4+ of the SERPs because that means you failed to deliver the right result on the top of page 1. As F2+BM get better over time, you should see a 1:1 correlation of searchs/pageviews (much like google has). Before F2/BM, you may have a 1:2-3-4 kind of ratio which generates more pageviews, but fewer conversions as the buyer isn’t finding what they are looking for or maybe they deactivated totally because shopping on eBay was so time consuming.
Seeking Alpha disclaimer: I am long google.