eBay CEO Jamie Iannone Touts AI Advancements At Morgan Stanley Conference
eBay CEO Jamie Iannone and CFO Steve Priest stopped by the Morgan Stanley Technology, Media & Telecom Conference yesterday for a chat with small mid-cap interim analyst Lauren Schenk.
SA Transcripts
Most of the conversation trod a well worn path through familiar vertical focus and advertising revenue territory - if you tuned in to the Q4 2022 earnings call, you've heard these jokes before.
However, Iannone made a particular point to "lean in" to eBay's magical tech-led reimagination efforts, specifically highlighting how AI is transforming the marketplace.
Lauren asked:
Okay. Maybe I'll end on one that I feel like we're going to hear a lot about over the next two days. And it's a bit of a longer-term question, but how do you think recent advancements in AI and large language models could impact eBay? And what's your willingness to invest behind some of these new technologies?
Jamie's answer was a bit long, so I'll break it down into parts.
Yes. Some of my favorite days of the days I meet with our AI teams because they are just working on some amazing stuff. So AI isn't new for eBay. When you think about the scale of eBay, we use AI in almost everything and machine learning in almost everything that we do on the platform.
So take, for example, our advertising products. So when you look at advertising on eBay, we have over 2 million sellers and over 700 million listings. Well, to drive that amount of complexity, especially against a listings-based platform, you need very advanced AI to figure out the relevancy, how to move people across categories, how to get them to the right next product. So, that's a huge emphasis for us for AI.
The use of AI for ad relevance is an interesting concept, but actual experience shows eBay is far from having this dialed in - in fact one of the biggest complaints I hear from buyers is that Promoted Listings ads are often extremely irrelevant either due to keyword spamming or just very poor targeting.
The problem has only gotten worse as eBay stuffs more and more ads on listing pages and reserves more of the top search slots for cost per click ads.
I'm sure big name brands like Nike love seeing these kinds of results for their trademarked brand name.
When you think about the search challenges we have, last quarter, I talked about $0.5 billion in GMV through using AI in a different way in search and structured data to really help us present better results to customers. So, we use it a lot there.
I'll admit I laughed when reading that first sentence. "When you think about the search challenges we have, last quarter"... yes Jamie, there were a lot of search challenges last quarter and in quarters before that and in the current quarter and I suspect for many quarters to come.
For example, there was almost entire day where search was completely broken, returning 0 results for almost every keyword search.
Liz Morton ~ Founder
And apparently the tech-led reimagination has not included cleaning up legacy product pages leftover from ex-CEO Devin Wenig's failed catalog and structured data initiatives, resulting in buyer confusion and seller stress as incorrect and misleading product data is shown on these pages.
Liz Morton ~ Founder
Last quarter, we just used AI to figure out shipping times with more accuracy, and we were able to take 1.5 days off of shipping times on the platform and about two days off for an authenticated item.
This isn't exactly the positive statement that Jamie thinks it is. As unsuckEBAY pointed out on Twitter, historically missing transit calculations by 1.5 - 2 days is a huge deal.
$EBAY CEO Jamie Iannone at Morgan Stanley Technology, Media & Telecom Conference
— unsuckEBAY (@unsuckEBAY) March 7, 2023
Boasting about #AI finally fixing eBay's history of grossly inaccurate transit miscalculation isn't the flex Jamie thinks it is🔥🚛📦
(1.5 - 2 day transit table miscalculation is a *lot*😲) pic.twitter.com/6OML5nLU8f
I'm also not at all convinced the AI assisted transit calculations are going to be significantly more accurate, unless they are also changing other aspects of how they determine those timeframes.
eBay's shipping calculations have not only been historically very wrong, they've also history been a source of stress for sellers as eBay uses some opaque algorithm or AI powered calculation based on "historical data" rather than seller-input handling times.
Liz Morton ~ Founder
Jamie does not say whether eBay will continue to use that historic data instead of trusting that sellers actually know their businesses & set specific handling times for a reason.
So when you think about what that can do to conversion on the platform when you can use it in these capabilities, the reason I say that those are most exciting meetings is that nobody has the level of data that we have against that AI capability.
So just take images. So every day, we have 1.7 billion images on the -- or sorry, listings on the platform. So, we were able to train our image technology on over 10 billion images on the site and use that to be able to do things like image search.
So, we just rolled out a brand-new image search on the platform I could hold up anything here and find it on eBay or if I want to see how much I could get for selling something around the house, just hold my phone up and figure it out. And that's an example of the massive wealth of data on the platform that we have and how we can use AI in different places.
Now this one really caught my eye - what "brand-new" image search Jamie?!
eBay has had image search functionality since 2017.
There was about a 3 month period between December 2021 - March 2022 where it was disabled and removed from the eBay app due to "some inconsistencies that popped up" (yes, that's really what they told users).
When it was re-introduced in March 2022, the accuracy and relevancy were even worse than the previous version in my experience, despite reports of advanced machine learning and image recognition enhancements.
Liz Morton ~ Founder
So obviously image search on eBay is not a brand-new thing - maybe he meant they've come out with yet another new version that is even better now that we have magical AI abilities?
I updated my eBay app for Android to make sure I'm using the most up to date version of image search and....nope sorry, the results have not gotten any better.
I can understand maybe not getting all the details right (a wallet made out of recycled racing sails is a fairly niche product 😂) but I would have expected at the very least for it to be recognized as a wallet and/or to pick up on the Harken name and logo as one of the most well known sailboat hardware manufacturers in the world and show me some boat parts.
Instead, the AI missed it by a nautical mile.
If there really is a "brand-new" image search for eBay, it must be in very limited beta release because I'm not seeing it and have not heard anything about it from buyers, sellers or official eBay announcements.
But hey, at least both Jamie and Chief AI Officer, Nitzan Mekel-Bobrov have stopped trying to use that 3D sneaker gimmick as an example of eBay's amazing technology advancements....perhaps they finally got my message that it's a bit silly for them to take credit for simply bolting on a third-party solution created by RestAR. 😉
Liz Morton ~ Founder
Mekel-Bobrov did a recent interview with Forbes about eBay's AI initiatives where he touched on some interesting points about that vast amount of data Jamie referred to.
Shelley E. Kohan
Transparency around AI
As more and more retailers expand the uses of artificial intelligence in various capacities, there is a growing effort among companies and consumers to ensure that data collection and data usage are responsibly handled. Consumers are increasingly empowered and vocal about data transparency.Retail is not as heavily regulated as medicine and banking regarding data protection. Still, there is an urgent need for the retail sector to catch up with the foundational shift in consumer preferences toward more regulated responsibilities regarding data protection.
“EBay understands the cultural shift with consumers and has started to build a scalable approach to provide a framework for employees who are building out AI initiatives. We are fully aware of our dimensions of responsibility,” said Mekel-Bobrov, adding, “We are great at staying ahead of what the customers expect.” As more AI initiatives are rolled out globally across many retailers, providing transparency is critical, and companies should create plans to ensure that consumers are well protected.
I couldn't agree more that transparency is absolutely critical, but can we trust eBay to be transparent about AI when time and time again we have seen a lack of transparency on so many other issues?
So, I'm excited to see what we'll continue to do with it, but I would tell you that the AI team really works across the whole company, advertising, search, selling, et cetera, to really unlock these new capabilities for us. And it's pretty exciting with the volume of information that they get to canvas.
I agree with Jamie - it's very exciting to see how AI will continue to ripple across the ecommerce industry in new and innovative ways...like Shopify's recent addition of a ChatGPT powered assistant in the Shop app.
Liz Morton ~ Founder
Considering where we are with image search on eBay, I won't be holding my breath for conversational AI powered search any time soon.
