eBay's Hugh Williams Explains Cassini Search, Part One
By Ina Steiner
eBay rolled out a new search engine to users in North America this month, and sellers have lots of questions. What exactly is "Cassini"? Will shoppers notice a difference when searching for items? How will it impact the upcoming holiday shopping season, if at all? Vice President of Experience, Search and Platforms Hugh Williams joined EcommerceBytes Editor Ina Steiner to answer these questions and more.
Our readers have lots of questions about Cassini, thanks for joining me to talk about eBay's new search engine.
Hugh Williams: I thought maybe it might be useful to start with the framing of why Cassini would be necessary and why it's something that we've gone to the effort of building at eBay. I thought that might be a good place to start and then talk about what we're doing with Cassini and what we'd love to do with Cassini in the future.
Probably the place to start, and I know you've listened to the webinar where I talked a little bit about this, but I think as a technology guy, I'm pretty aware that software ages. And systems age. Most of us aren't using our ten year old copy of Microsoft Office. Most people have got off Vista, the old Windows product, and moved to new things, most people don't have the computer they've had ten years ago. They have upgraded since.
You can think of our search technology in that way. We'll look back at those products as having served our needs and met our needs, but I think we all want to move on to the latest generation and to have all the new features that become possible through using the latest hardware and software. I think our search technology had gone into that state. Our Voyager search engine is a wonderful workhorse. It's served the company very, very well. But it's not state of the art technology.
I think sellers wouldn't argue with you that the search could always be improved but I just wanted to jump in and ask a question. When we're talking about Cassini, are we talking about how you decide what listings are going to appear in the search results, or how they're ranked, or is it both?
Hugh Williams: That's a great question. As I say, it's very much time to upgrade the technology, and when I say technology, I mean that in the broadest sense.
- Most modern search technology runs not on large reliable expensive computing equipment. It runs on thousands of lower cost machines and scales. So that's something that's embraced under the Cassini umbrella.
- The way we operate those machines in our data centers which, as you can imagine, is a really complex task, that falls under the umbrella of Cassini.
- The search technology itself - those systems that take the user's query and process that query and give back results fall under the umbrella of Cassini.
- Then of course the ranking algorithms, the way we power the best match technology, in particular is enabled by this Cassini technology and the possibilities of things we can do in best match are opened up much wider by using the Cassini technology.
An analogy might be, Voyager is a really nice toolbox. You open up the toolbox and get out a few tools and build. I feel like Cassini is more like a fully featured garage with walls of tools that you can use to get out and build great solutions for our customers. It's is a stack, if you like, it's all of those things. It's really a complete ground up rebuild of the whole search technology stack.
Hugh Williams: You might ask why we want to undertake this. There's reasons of, wanting primarily to be able innovate on behalf of our customers to deliver a great experience for them, and in particular, the team - and I include myself in that - are very passionate about ensuring that our buyers find what they want on eBay and are successful in purchasing things on eBay, and we're very passionate about making sure our sellers are successful on eBay.
One of the wonderful things about eBay is that if buyers are able to find what they want, then buyers are happy, and sellers, of course, are happy too because they are being connected to those buyers.
Cassini is a piece of technology that is really going to enable us to do that even better than we do today.
Can you share any metrics that you look at, because increased sales is a very broad kind of a measure. Do you look at exit rates and those kinds of metrics?
Hugh Williams: We don't share our metrics publicly for obvious reasons. In my mind, our goal is really to make sure that we're successful in connecting buyers to sellers, and we look at that in lots of different ways to make sure we understand the full picture.
eBay as a whole - clearly you see the big picture, but an individual seller who wants to compare their listings so they can improve their listings, how would they know when Cassini launched? So they can say, "under Cassini, now that we have this new search platform, I want to look at my listings and see how they're performing. Am I doing everything right? Is this actually enhancing or not?" When did Cassini fully launch on active listings on eBay.com?
Hugh Williams: I think "fully launched" is a difficult question to answer because I think as you know from the webinar from the things we've been talking about, we've been rolling Cassini out for different scenarios for over six months.
We did some really great work in upgrading our completed search last year so customers were able to do better price research both buyers and sellers and that was Cassini-powered. We did some work that I'm really proud of with null and low recovery, so when our buyers weren't able to find the items that they want, we were able to send those queries off to Cassini to try and dig deeper, to give them great results. I know that feature has been pretty well received. And so on and so forth.
Over the past few weeks, though, we have slowly ramped up taking traffic - I don't know how you describe it - for mainline search, where customers type a query in the search box and get back results on our search results page, and we hit 100% with that. Over the last couple of weeks, I guess.
How do you feel about what's happening in terms of shopper behavior? Are you feeling pretty good about it, or is it too early to tell?
Hugh Williams: Our goal in launching Cassini for that mainline search scenario was that customers wouldn't notice. I know you probably read a blog post that I wrote a little while back about how I was thinking about changing platforms and some of the lessons that we learned in doing that. I think our experience can be summarized as, when you're thinking of making a platform change, you should see it as a two step process.
The first thing you should do is try and replicate the behavior of the old platform as you bring the new platform on board. Then you should begin to innovate on that new platform to take the experience to the next level.
Our goal in bringing Cassini on line for the North America business over the last few weeks was to make that smooth sailing for both our buyers and sellers.
So searchers shouldn't see any difference in the type of results they're seeing?
Hugh Williams: I think clearly, you can find specific queries or specific items that might appear slightly differently if you went looking for the details. But overall at a macro level that is absolutely the goal.
An analogy is useful here. You can imagine if you bought a 2013 model car, brand new from the showroom, it's not trivial to make that perform exactly like the 2002 model car you've got in the garage, which is what we're trying to do - make the new platform behave like the old platform as we replace it, it's certainly not a trivial task. I don't want to trivialize all the great work our team's done in replicating the behavior of the old platform. It's a very challenging task to do that, as you can imagine.
That sounds like a really good analogy. I think the question people might have is, when are you going to take it out on the highway and really let it go? How much tweaking are you going to be doing to Cassini and how can sellers keep up with those changes, and also, is there going to be a moratorium during the holiday shopping season?
Hugh Williams: I wouldn't use the word tweaking. I would say I see very central to our goal, the search team, is constantly innovating and improving the search experience on behalf of our sellers and our buyers.
One thing Cassini is going to enable us to do is to do that faster and make progress more quickly. We very much look forward to doing that.
In answer to the second part of your question, I very much appreciate that there is a holiday season, and it is a significant season for both our sellers and our buyers, and I want to make sure that as we've always done every year, we minimize the change throughout the site as we go through the holiday period. But what I would say, though, it is a great period for us to be running small experiments and really learning about how customers might behave differently through the holidays.
I do that on behalf of both our sellers and our buyers, and it's a great time for us to learn as well. As you can imagine, the behavior of sellers and buyers during the summer months might be quite different from their behaviors during the holiday months as they are shopping for occasions that are on fixed dates, and we owe it to our customers to learn about that and make sure that every year we continuously improve that shopping experience.
In Part 2, eBay's Hugh Williams discusses five factors eBay will use to improve the search experience under the new Cassini platform - including the use of behavioral data. He also talks about search refinement and explains what sellers should be thinking about in order to get better exposure in search. Part 2 is now available in Tuesday's Newsflash newsletter.
Leave a comment on the EcommerceBytes Blog.
About the author:
Ina Steiner is co-founder and Editor of EcommerceBytes and has been reporting on ecommerce since 1999. She's a widely cited authority on marketplace selling and is author of "Turn eBay Data Into Dollars" (McGraw-Hill 2006). Her blog was featured in the book, "Blogging Heroes" (Wiley 2008). Follow her on Twitter at @ecommercebytes and send news tips to firstname.lastname@example.org.
You may quote up to 50 words of any article on the condition that you attribute the article to EcommerceBytes.com and either link to the original article or to www.EcommerceBytes.com.
All other use is prohibited.