2026, eTail Palm Springs -SEARCH_Keynote_ Next-Gen SEO Metrics_What Ecommerce Marketers Should Track (and Drop) for Maximum Impact in 2026

Announcer: Um, for our first session here, I am so excited to welcome, um, Dana Tan, who is, um, the director of CE, uh, SEO at Under Armour. She will be talking about next gen SEO metrics, what e-commerce marketers should track and drop for maximum impact in 2026. Please welcome her up on stage

Dana Tan, Director of SEO, Under Armour: Hey, good morning everyone Thanks for getting up early and coming to the first session. I am not a morning person, so this is a stretch for me. Um, I'm gonna talk a little fast, and it wasn't lost on me that I'm speaking about AI and my notes look like this. So clearly AI had nothing to do with writing these.

Um, yeah, we're gonna talk about new metrics for AI. Anybody talked about AI yet this morning? I'm pretty sure you probably have, and it's not... It's barely AM. So scary statement, right? Your traffic is going away and it's not coming back. It's truth. Might take a while, but the world is changing, right? Um, do I need a clicker? Here we go

So we're gonna talk about how the landscape is changing. Um, Gartner has data that says that you're gonna lose about 50% of your organic traffic between now and 2028. It might be sooner than that. Um, one of these, uh, data points is older, uh, about the zero click. So one of the most recent studies I saw was by Semrush, and results in AI mode, Google AI mode, result 93% of those searches do not result in the click.

So 93%, not 50, which is where this was at when I pulled this data a l- a few months ago. So we're in a new landscape, and we need new KPIs because traffic isn't gonna do it, right? Um- I'm gonna skip through the agenda slide 'cause y'all can read, and I got limited time. So we're gonna talk about how LLM, LLMs work, some new metrics, where to get data, and some tools, and there'll be references in the, in the appendix for y'all on some further reading if you wanna do it.

This is old data too, but it's interesting still because it took ChatGPT half the amount of time as it took TikTok to get to four hundred million users. ChatGPT is way over that now. It's roughly a third of the world's population. Um, but interesting data, just the lightning speed, right? This you all know, right?

Um, I think... Yes, it works. Awesome. Uh, this is the old experience. You know, you put in a search, you get a list of blue links. You click on the blue link. Maybe there's products and information there that you like. Maybe you don't find everything you need. You go back to Google. You click another blue link.

You go there. You read some more. You do this over and over again, maybe for half an hour, depending on what you're looking for, right? And the new experience, you all know this too, right? Now you just say, "Hey, I'm looking for a great pair of long-distance running shoes for women, uh, high cushioning, size eight, purple, and my budget's a hundred and twenty dollars."

And you get a robust answer. You might get some nice product recommendations. And probably by the time you go to a website, if you have to, if you can't do a one-click buy already inside the, the search platform, the AI search platform, you do-- You go to the site. You already know what you want. You put it in the cart.

You buy it. You leave. You haven't navigated the site. You haven't searched the site. You haven't looked at anything on the site. You just went straight to the product page that you wanted. So when I first started researching this, um, I honestly thought that SEO was going to transition to AEO, and that's probably gonna happen at some point over the next three years.

But for now, this is really important. If you remember one thing from this talk this morning, it's that SEO is your ticket to the audition Without it, you're not getting in. You're not even getting invited to try out. So your SEO fundamentals are what get you into the AEO pool. Um, how many people, show of hands, how many people think that Google is, is the company that sort of is the library of the internet?

Me too, right? Three, three months ago. I mean, I've been doing this, I've been doing this for 22 years. I've been doing it since SEO was dead the first time. Uh, anybody heard of a site, a search engine called Common Crawl? Ah, one. I'm, till recently didn't know it either. So AI platforms when, when they build their systems, it's not like they just woke up and didn't know anything and had to learn it, uh, you know, from day one.

They go and look at everything that already exists. They look at everything that's indexed in Google, everything that's ranking and indexed in Bing. They look at everything in Common Crawl. Common Crawl is a non-for-profit search engine that documents the web for research and AI systems, right? So they're gonna look at all this stuff that already exists.

Well, how do you get there? How do you get in Google? SEO, right? You have to get crawled, indexed, and ranked. So let's talk a little bit about how AI systems work. I built this visual specifically for technical marketers. I wouldn't show this to an executive. Maybe a CTO. I mean, if you've got a really technical executive, cool.

I made this for myself just to visualize for myself how these things work, right? So the big box, number one, is the search engine rankings, right? So that's your ticket to audition. Doesn't mean you're gonna get retrieved, doesn't mean you're gonna show up, but it means you're eligible, right? Number two, machine-validated trust.

When I first started putting this together, I thought that retrieval happened first and that machine-validated trust happened after. Not so. Machine-validated trust happens first, and that involves a lot of things like looking at the landscape around your brand, sentiment around your brand, right? If you pass that level, then you're eligible for retrieval, which means can an AI system retrieve your content?

If it's text embedded in an image, probably not. If it's JavaScript, no. You know, flat out no. So retrievability is step three, right? Step four is after the AI system has put together the whole answer, right, and it's ready to serve it up to the, the person who did the prompt, it's gonna check it again, and it's gonna...

It's looking for fact verification, third-party sources, author information, anything. What, what's important to know about AI systems is they're inherently risk-averse I know, hallucinations. If, if they're that risk-averse, why are there so many hallucinations? Um, there are not as many hallucinations as, as there used to be, for those of you who've been using it for a year or more, right?

It's gotten... They've gotten better. Um, but they're gonna, they're gonna put this last layer of checks, and it's kinda like if you're doing a project at work, and you finish, um, a, a publication, a post or something, and it has to go to legal, right, for legal compliance. It's like the last check, and if you don't pass that, you don't get put in the response.

So that's the order of things. Universal verifiers are not all in place yet, but they're coming. So making sure that your content is concise, trustworthy, accurate, verifiable, all important things. This is kind of the same process. If you wanna share with an executive, this might be a good visual to share.

Feel free to steal it, use it, whatever. It's kind of the same steps, only simplified, right? You've got that bottom layer of SEO, right? Machine learning from content online, that's SEO. Without that, you're not gonna show up in AI systems. So really important to know that. So traffic. Traffic was only ever a vanity network, uh, um, metric to begin with, right?

It was useful because it led to a desired outcome. So let's talk about new metrics that will lead to desired outcomes for your business, and you're gonna have to get really comfortable with the word chunk. I'm sorry, it's not a pleasant word. It's-- But it's the word for now until some of the naming conventions of these metrics change or evolve or maybe a good marketer can give them better names, right?

Um, this is an eye chart. This is really meant for you to take home, print out, and put it on your wall above your desk. So I'm not gonna go deep into these 'cause we haven't got a lot of time. But basically, on the left, all new. All new metrics. On the right are things that you can do or are doing or you should be doing today.

And there are just a couple red dots on the right for agent behavior, right? For any of you who are trying to track, um, in analytics agent behavior, good luck. It's, it's a, it's the Wild West right now. Um, we'll get there eventually. Uh, but eventually we're gonna need to know. What's important about that is for Adobe Analytics, Google Analytics, what do they track?

They track humans. They track human behavior, right? Adobe is really working on being able to track agentic behavior. They're gonna have to get there because people are gonna wanna, you know, know where their marketing spend is going, right? And determine how to optimize it. So those are challenges in the, in the web analytics space right now.

Um, I'm gonna breeze through this slide. Sometimes when you're trying to explain to executives, uh, new things, it's kinda like introducing a food, a new food to a picky eater, right? What's the best way to do that? Find something familiar and hide the new food in, in, the, the unknown food in the familiar food, right?

So if you had to sort of say, "Hey, what, what... These new metrics that you're looking at, do they roughly equate to something that I already know or that I'm familiar with?" So I kinda did an equivalency chart here where, you know, traffic, chunk retrieval rate. We're gonna talk about what that is in a second.

Um, share of voice, citation share, right? Um, gonna talk about- The fifth one here, relevance. Relevance is everything. Um, I, I have this phrase I, I like to use called perceived need. That is what AI search is really about. So they're going to determine what your perceived need is, not your persona, not your demographic, not your age, not your...

Perceived need might be related to location, right? If you're looking for basketball shoes in Seattle, right? Your perceived need is that, that area, right? But your perceived need in an AI system is the center of the universe, right? And then what they're gonna do is look at your prompt and say, of all the stuff that's out here, mathematically, they're gonna do all these e-equations that we won't, we won't go into.

They're gonna do math to determine what topics and answers are closest to your perceived need by math, and the ones that are closest are the ones that are gonna get served in the prompt. So relevance matters a whole lot. All right. I picked five. There are more. There's, there's about a dozen, but I picked five KPIs for executives, how to explain them, how to define them, what they mean for the business. Chunk retrieval rate. This was the first one in that last chart.

This is really just about how often your content appears in responses. Now again, tracking some of this is hard. There's a ton of tools out there. When you're looking at tools, considering tools, ask them, "Can you track chunk retrieval rate for me? Show me. Show me how you do it." Right? Some can, some, some not so good, right?

Um, so this is about frequency. How often is your content showing up? LLM answer coverage. This is really about relevance, right? So if your business is skincare, I was talking to some nice ladies here who have a skincare business. That's your topical universe, right? How completely can your information cover the topics that you and your customers care about, right?

Comprehensive, unique, factual, verifiable, concise, right? All very important. Prompt tracking, not a fan. Prompt tracking can be useful, but promp-prompt tracking all of it in, in and of itself, eh. 'Cause n-how many people type in the same question the same way, right? You might get a similar answer, but prompts-- What you really can do with prompts, though, is sort of create that topical area.

So a, a cluster of prompts around a topic can determine how well you, you are covering LM, LLM answer coverage, right? So it's really about having a bunch of prompts in a topical or category area, right? And measuring how comprehensively in that category are we covering that particular topic, right? Versus competitors.

Trust signal scoring. This, this goes back to machine-validated trust, which we talked about a little bit, right? Um, and we, we've talked about like clear authorship, um, claims, backing up claims. If you say something like, "We tested this fabric for ten thousand hours," and that's it, and that's all you say, yeah, that's not g-- that's probably not gonna pass machine-validated trust, right?

What you can do is maybe go to a third party, a, a trusted publication, and get them to verify and talk about the ten thousand hours and say, "Who? Who tested it? Where did we test it? How did we test it?" Maybe you can't share proprietary information, but share some facts that can back up your claim that you tested your fabric for ten thousand hours, right?

That's trust signal scoring. If you don't pass that, you're, you're gonna be invisible, right? There's that word again, chunk performance, right? This is different from chunk retrieval rate, right? This is how well you appear. It's not whether or not you're there. It's how much of the result you take up. Where do you appear in the result?

Do you get just a mention with no citation? Citation would be a link, right? Just for those of you who don't know. A me- a mention with a citation, which would be a mention and a link. Just a citation. Do you get paraphrased? This happens too, with no mention and no citation, but they're using-- the AI is using your content.

So your brand has influenced the response, but you're not getting a citation or a mention, or you're omitted. You're just not there at all, right? So that's the difference between, uh, chunk retrieval rate, which is how often is your content showing up. This is how well your content is showing up versus competitors, of course.

Brand influence in a zero-click world. We talked a little bit about the ninety-three percent non-click rate in AI mode. Um, this is a hard one. So for those of you who haven't read it, there's a book called The Machine Layer by Duane Forrester. I'm not, I'm not a- an affiliate. I'm not, I'm not selling, I'm not selling anything.

Read it. Buy it. Read it. Read it again. It's gonna help you understand the landscape, and some of it's gonna be technical, and just let it wash over you because it'll help you understand the landscape that we're all playing in. I was saying at the table, the game has started, right? The game is happening right now, and we're playing it whether we know we're playing it or not, and building the trust in our brand, that influence and trust for your brand, it's being built right now.

And if you're not going to address it, if you're not actively addressing it, getting it back, earning that reputation back later is going to take a lot of time, a lot of effort, and a lot of money. So it's really something to think about. Your brand reputation, the sentiment, sentiment towards your products, right?

AI won't, won't just, um, mention you in positive ways, right? It'll mention you if there's problems with your, with your products. If you have a product page and it says, "This thing can go in the washing machine," and s- a bunch of people buy it, and they take it home, and they put it in the washing machine, and it falls apart, and they go and post a review online.

When people find that product in AI, the AI system's gonna tell them, "Hey, people are saying that-- not to put this in the washing machine. It'll fall apart." Right? So that might be something you either address as a brand, right? You can't really control it. You maybe can mitigate it a little bit, right? But to be aware of it is important.

Um, the reason I mentioned Duane's book is one of the rec- the recommendations he had, which was awesome, is to build an AI visibility journal And I'm gonna-- I highlighted it in red here. Um, it's manual. Sorry. Uh, but what, what it can help do is it can mitigate some things. It also can help you track whether or not your content is showing up in AI systems paraphrased with no mention or citation.

It's really important to know if your brand is influencing those responses. It also, when I say mitigate, can help you when someone comes to you from another team and says, "I went to ChatGPT, and I looked up, um, basketball shoes for point guards. Why aren't we showing up there?" Okay, it's like, pump the brakes.

I need to know exactly verbatim what your prompt was. I need to know exactly what system you put it in, what version of that system it was, what time of day it was, what, what the date that you did it, and then I need to know the exact output. I need a screenshot of the output, and you need to track those.

So then you can take that same prompt, put it in all of the platforms verbatim, and document what the output is. You'll be shocked at how different they are, right, from one platform to the next, and from one week to the next, right? So when you're reporting on AI visibility to your executives, don't do it in aggregate.

Don't do that. Don't say, "Our AI, AI visibility across all platforms is seventy percent." Tells you nothing because it could be-- and it could be going up, and you're like, "Great, it's going up." It could be going down in four platforms precipitously and going up in two platforms precipitously and giving you an aggregate that actually looks positive, and you're blind.

Now you're blind to the places where you need to go work, right? So very, very important. AI visibility journal. If a tool can do something like that, I w- I would ask, right? If you're interviewing pot-potential tool providers, ask them, "Can you do this thing for me?" Um, so many tools, right? I mean, I'm sure probably some of them are here.

I think Scrunch, Scrunch AI is here. Um, I'm agnostic as far as tools go, right? Um, oh, I didn't say where to get the data. Oh. So where you can get data today. Server logs. If you're not looking at your server logs, your SEO team's not looking at server logs Do it, right? It's very important so you know what LLMs are coming, what content they're accessing, all that stuff.

Your analytics, Google Search Console. Uh, Bing just rolled out a whole suite of AI tools. I have not ... It's in beta, I think. I haven't looked at it, haven't tested it yet, but it's there. Be on the look- have your SEO team be on the lookout for that. Um, thank you. We've got, like, two minutes left. I really wanted to have time for a, a question or two.

Um, and in the appendix there's some sources, some recommended reading, um, et cetera. But thank you. Thank you so much. I think we've got a couple minutes for questions, so

Announcer: We're getting a mic

Audience Member: Here. Oh, sorry.

Dana Tan, Director of SEO, Under Armour: Oh. Hi. Yeah, I got it. Yeah. So wherever it's relevant, right? So if you've got a blog, um, you can blog about topics that you know. If you're ... You've got specific unique features about a specific product, product page, right? The other thing to track is if your content... Track which content is getting retrieved more.

If you find that the content that appears at the top of the page tends to get retrieved more, that's important to know, right? How do you track that? Um, tools, right? Um, vari- various tools. I'm not gonna mention any 'cause I'm, you know, I'm agnostic that way. But, um, or you can use the AI Visibility Journal, right?

You can use that if you're a small company, and, you know, maybe you've only got 100, 200 pages, right? If you're enterprise level, obviously that's gonna be a little ha- uh, harder. I thought I saw another one. Any more? Oh, we're on time. Ah, there's one.

Audience Member: Video photo content versus written content today of shifting and resourcing- Mm-hmm ... um, different allocations. What are you guys doing from a written allocation of content production versus a visual allocation?

Dana Tan, Director of SEO, Under Armour: Oh, man. That's a great question. Really, it, it, you're better off being comprehensive, right? So if you're gonna do a blog post, for example, you're better having all those things.

You're better having some text, some video, some great images. If you have video, you need, ah, captions and transcriptions 'cause AI can't... they can't tell what's in that video, right? Very important. But, um, I will say I have talked to my PR team probably more in the last eight months than I ever did in the last eight years, right?

SEO and PR for, for AEO pur- purposes have to be tied at the hip, right? Um, 'cause that's really important in, in terms of how your brand is being... Everything matters, right, outside of your site. Everything everyone's saying about you, your social sites, Reddit, right? Not just your subreddit, ev- what people are saying on other subreddits about you, right?

All of that matters. All it, all of it plays into sort of how your brand is perceived in the ecosystem, right? Whether it's not negative, positive, neutral. Cool. Thank you so much, guys. I hope you enjoy the rest of the conference.