B2B AI Search: What Actually Works (and What to Stop Doing)

Adam: Welcome to another episode of How About Some Search? Today we're talking about B2B AI search and are joined by three expert panelists. Please welcome Dayna Lucio, Associate Director of SEO at Amsive; Brian McLeod, AI SEO Director at Level Agency; and Sara Vordermeier, Content and Search Strategist at Uberall.

Adam: Dayna, would you mind introducing yourself?

Dayna: Hi everyone. Dayna Lucio, Associate Director of SEO at Amsive. A lot of my background is in multi-location and franchise businesses, but I've also worked a lot with B2B businesses across different industries: software, manufacturing, professional services, cybersecurity, and more. Excited to chat with everyone today.

Adam: Thanks, Dayna. Sara?

Sara: Thanks for having me. I'm Sara Vordermeier. I'm a content and search strategist at Uberall, a location marketing platform for multi-location businesses around the world. I've been working in the B2B space for about five years, ad tech first and now martech, and before that I worked in various editorial and content roles. So that's where my focus lies, on the editorial side of search. It's my bread and butter, and I can get pretty opinionated about it.

Adam: Awesome. Thanks, Sara. Brian?

Brian: Hey everybody. My name is Brian McLeod. I'm an AI SEO Director at Level Agency. I've been involved with SEO, AI, and all things digital for over 10 years and have had the great pleasure of working with some amazing brands throughout my career. In this case, with B2B brands, I'm really excited to talk about all things SEO, GEO, or whatever we like to call it these days, and how we can help improve visibility.

Adam: Awesome. Thanks, Brian. Let's jump right in. How does AI search for B2B differ from CPG, consumer packaged goods, or D2C, direct to consumer? Dayna, let's start with you.

Dayna: When we're talking B2B, it's really a longer research process, because we're not getting that immediate consumer buy-in. It's somebody doing the research and bringing it back to their leadership, their board, or another entity like investors. What's really important is having that trust and third-party validation and credibility, and being able to find that easily, whether that's on the website with third-party mentions, product roundups, or just a general overview of the brand.

Dayna: I think that's really important with B2B. And once people actually get to the site, find you, and have already vetted you, being able to clearly and quickly communicate value props, especially for those stakeholders at the decision level.

Sara: I completely agree with everything Dayna said. In B2B you're rarely optimizing for, or speaking to, the real decision maker, so sales cycles tend to be a bit longer than with CPG or D2C. Closing deals and influencing deals takes a lot of trust and a lot of time.

Sara: When I think specifically of AI search within B2B, the most interesting thing we've come across is the data sources that AI uses to recommend businesses. For example, LinkedIn this year has been one of the big data sources that AI models pull their information from. So a lot of B2B brands have been leveraging LinkedIn, whether through posts, newsletters, or articles, to win that trust on those channels. It's not just going through the website, but also publishing content in these spaces.

Brian: It's funny, because throughout my background I've seen this from all different angles. What I found in the D2C space is that the prompt within AI typically resolves into a purchase. In the past I ran an executive audit on a Shopify website, a supplement brand, against their direct competitors, and almost every tracked prompt ended with a product, a price, and a place to buy. Within the AI space, citations are the number one goal today in terms of how to stay relevant in search, and in that case the citations came from retail and review aggregators. That's typically what happens in a D2C space.

Brian: For B2B, it resolves into a shortlist. I had a client who runs corporate training seminars, and we used their insights to run 210 prompts loaded into Profound and a proprietary tool we use, across 10 topics like corporate sales training, leadership development, and government training. What we found is that none of those end in a checkout. They end in a vendor set.

Brian: The sharpest version of this is a financial services account with two services under one domain: consumer wealth planning on one side and advisor recruitment on the other. Same brand, same site, completely different AI behavior. So what I found is two different strategies for one client. B2B will give you lower prompt volume and a much higher value per prompt. It's pretty much the inverse of D2C, where prompts lead to a purchase. Within B2B we're playing more of the long game in this new age of search.

Adam: A couple of things resonated with me there. You talked about multi-phase sourcing, or a multi-phase process, especially when you're looking at sources down to the citation layer. So when we think about how we're working with the different types of B2B brands, what have you noticed? Brian, this question is specifically for you: how have you noticed the multi-phase B2B buying process has changed from, let's say, SEO to GEO?

Brian: Even if we're just talking GEO and SEO, there's quite a bit of overlap between the two. Some may say it's 100% overlap, but we're not going to debate that here. What we found within B2B is that it typically doesn't end in a purchase right away. It always starts with an inquiry, learning more or requesting more information, before the user makes a purchase.

Brian: Within traditional search we had really strong ways of tracking that, whether through tagging events in GA4 or using Salesforce to track the pipeline. Within AI search, it makes things complicated, because now all of that is foggy. So what we have to do is look at how brands are showing up for certain prompts using tools like Writesonic or Profound, and also look at it from a query fan-out perspective and identify two things. One: is there tangible search volume that is trackable and measurable? And two: is there a way we can attribute it to the birth of a sale?

Brian: One client we partnered with is in the B2B cybersecurity sector, and we wanted to track queries based on content we'd been producing. We can't necessarily track how conversions appear within AI, but the best way we could attribute it was by correlating the increase in citations and AI visibility with the increase in sales within Q2. Once we made that connection, even though it isn't one-to-one as it was previously, you could easily make the case that these sales are increasing based on your AI efforts. But as we go into 2027, it's just going to get more complicated. With how AI is advancing, it's going to make our jobs harder and, dare I say, maybe even more fun.

Adam: I agree. I think attribution is muddy at best, even with current channels. You mentioned something, and I'm putting this out to the broader panel: deciding what kind of B2B buyer questions to track. How do you decide what's worth tracking, and how do you know those prompts represent real buyer behavior? Sara, let's start with you.

Sara: I want to point out that these prompt trackers are an excellent data source for understanding how we can optimize our content to match the prompts, the queries, the long-tail questions that we're seeing in these platforms. But I see the buyer questions that appear there as just one data source. I will always combine and validate the questions we see in these tools with the real buyer questions we see in webinars, client calls, or events like brightonSEO, which just happened last week.

Sara: For me, it's combining those two data sources that really helps you understand buyer questions better. It's enriching what you see in the prompt tracking tools with the real-life conversations you're having. The more data you have from those sources, the better your content output and the better your strategy. It'll be more customer oriented, and at the end of the day, that's what builds trust, generates sales, and builds pipeline, which is the most important thing for B2B businesses.

Dayna: I was thinking along the lines of real customer data and where you can mine that from: sales calls, CRM notes if you have them, customer interviews. Mirroring that with the prompt data, or even some of the search query data, could at least be directional to give you an idea of what to start tracking. Then I would break that down throughout the customer journey, as well as the use cases that would impact the decision: capability topics, pricing, integrations, implementation, depending on whether you're doing software or a service.

Adam: Interesting. So you talked about real customer conversations and also using search data directionally, so that's keyword research as well?

Dayna: There's still a little bit of typical keyword research. If you have nothing to start with, that's a good starting point. Then you build on it and validate using customer data or what's available in the prompt tools, as well as query fan-outs. Even Bing Webmaster Tools, with their grounding queries, gives you a thought starter for some of that data.

Adam: That makes sense. So know your customer, and still do research. Awesome. Brian?

Brian: I agree with everybody here, but we have a three-phase approach, and it doesn't involve keywords initially. The first thing comes down to the discovery calls you have with clients or prospects: they have to validate the set you're tracking. For one client, we share our view based on the competitor set and the subject matter the client is looking at, and we need them to approve, remove, or revise what we're tracking. We want to make sure we're representing the brand accurately, that we get the product conventions right, and that we use the right verbiage so the right customer gets the right product. If they won't approve it, we don't track it and we move on.

Brian: Second, cut hard for intent. Most approved keyword lists are half junk, in my view, once you read them as questions a buyer would actually ask out loud. I have cut half a list in one pass: consumer intent dressed up as commercial, named venues, academic word matches. Word-match demand is not buyer demand.

Brian: Third, we actually have to check the data before we build on it. I've seen click-through rates that look like a content quality problem, but it turns out to be an acronym collision inflating impressions. So you have to build a prompt set that is well aligned, and that's why that communication very early on, whether through an SEO discovery call or early in the sales process, is critical. Plenty of prompts return a generic explainer with no brands mentioned, and they are worthless no matter how good the question sounds. It's really about the conversations you have at the very beginning, way before you start importing keywords into Semrush or Ahrefs or whatever tool you have.

Adam: Interesting. You mentioned intent, and I'm going back to January, when Rand published his seminal piece on AI search and how intent matters more than frequency for tracking. Looking at intent, the actual syntax, the linguistics, the structure of a question matters less than the intent behind it. How are you gauging that intent? Sara, if I was looking for noise-cancelling headphones and you were looking for noise-cancelling headphones, but you wanted them for a train ride and I wanted them for jogging, that particular piece of information matters less than the fact that we're both looking for a particular product with a particular purpose. Does that make sense the way I phrased it?

Sara: I think so. In AI search, sentiment matters a lot more than the literal, word-for-word prompt or long-tail keyword. The query fan-out method you see in these prompt tracking tools is an excellent way to check that you've covered all the different phrasings of a question coming from your potential buyers. For example, if the prompt is something about AI search visibility through Instagram for restaurants, you have all the different query fan-outs there, and you can use them to guide your content strategy and your process. I think that is a bigger thing in AI search than it was for traditional search.

Adam: Dayna and Brian?

Dayna: I'm thinking again along the lines of the query fan-out, because it gives you more context on the different ways a question can be phrased, or the secondary question behind it. I'd also look at what kind of results are being yielded, because sometimes you'll get really clear results from a specific question or prompt, where others will be mixed on what's showing up, especially between the different platforms. That can be a good way to see how Perplexity is answering a question versus something like Google AI Overviews, which might be more informational and direct, where Perplexity brings back more transactional answers.

Brian: I do think the query fan-out method has been an absolute game changer, especially for those who are new to the space, since it has been the main topic. It also gives tools like AnswerThePublic a bit of a comeback, because that has essentially been the query fan-out approach for quite some time, and now it's having its moment in the spotlight.

Brian: But there's a delicate balance. If we're going to look at the linguistics of a question or a prompt, we could go down a rabbit hole of how a user is going to interact with an AI, or how we should be tracking this versus that. The one thing I think is important for us in search is to avoid going down that rabbit hole, because it's going to cause a lot of unnecessary internal friction. If you're worried about certain misspellings or certain punctuation, you're going to miss out on something. These LLMs are pretty darn smart, and if you're worried about the little things, an LLM can autocorrect itself. Of course, pay attention to this stuff, but I want to throw a bit of caution out there, because if there is a listener hitting the panic switch, we have to keep in mind that while it's not perfect, AI and LLMs are pretty good at separating the little things.

Adam: I think we can all agree that when we log into LinkedIn, there's always a new strategy, a new tactic. I'll speak for myself: every time I log in, there's always something going on, and it feels almost overwhelming. There are so many levers we can pull, and the discipline is still in its infancy. In your practice, what tactics do you find work best, and which haven't yielded great results? Dayna, let's start with you.

Dayna: With any marketing, there has to be a multi-pronged approach. It's not just one tactic. A lot of tactics work together to create a strong strategy. Entity alignment and understanding of your brand is important in general: how your messaging goes from the website to social channels, listings, and third-party mentions, and having that be accurate and consistent across all of them. I think that's really important for AI search and LLMs, for their understanding of your business and what you do, and to grow that credibility.

Dayna: Then obviously technical. If your site's not crawlable by these crawlers, if there's too much JavaScript or they can't understand it, that needs to be sound and accessible. And then strong content that answers questions and is clear about what you're offering. A lot of times when working with creative, you get marketing language and jargon. Being as clear as possible about what the offering is really matters now, because some systems might not be able to interpret the made-up name you're using for your product. Just be clear about what it is.

Dayna: For the second part of your question, the methods that don't yield the best results are the ones that work in a silo, where people are looking for shortcuts. A couple of months ago schema was the big thing, but schema has always been important. That's not the flash in the pan for AI search. What I'd be careful of is choosing one tactic and going all in on it.

Adam: That's fair. And we've got to be careful where we say schema. We're going to start a debate online again.

Dayna: I know. I was scared about that one.

Adam: Brian?

Brian: Dayna has pretty much said everything I was going to say. It's funny, because a year ago it was all about content: make your content unambiguous, make sure it's crawlable, structure it in a way that, if we're using the acronyms, fits E-E-A-T. Just write good, honest content. But in recent months, and I've been telling my team this, the real strategy that can lift the levers for any website is your technical work. There are so many sites blocking ClaudeBot, GPTBot, CCBot through Cloudflare and other blockers, and nine times out of ten prospects think it's the content. What I've been finding more often than not is that if you lift those barriers, it's going to really help your AI visibility.

Brian: There's also schema markup, which I know has been a topic of everyone's conversation. Just put it on and be done with it. Interestingly, a week ago at brightonSEO in San Diego, the keynote speaker gave us a profound thought: instead of incorporating schema onto the website just to have schema, look at a page, look at what schema could be on there, and write the content in conjunction with the schema, so you give yourself a boost in visibility. It sounds simple, but for many of us it was mind blown. Schema is a net positive, but is it going to bring in the citations? Likely not. But every step helps.

Brian: Another interesting thing: we've been incorporating llms.txt files onto client sites, and I know everybody says it's useless, don't do it. I'm in the camp of two things. One, just because Google says it's not working doesn't mean you should trust Google. Who's to say a week from now they won't flip the switch and start reading the files? Also, I've seen Claude Code firsthand looking at that type of file on a website. It may not move the needle, but it's one of those things you set and forget. We've also been testing adding a page with the header "Hey AI, learn about me, come right here," a page made specifically for LLMs.

Brian: Those are some strategies that have been moving the needle a little here and there, but always in conjunction with the technical work we've been doing. I still think it's the Wild West in terms of what works best and what doesn't. But in recent months I've come to the conclusion that you've got to get the technical foundation taken care of first before you start worrying about what your content is doing to influence LLMs.

Sara: Completely agree. Brian, you've covered all the technical stuff, so for me, in terms of content strategy, I actually haven't overhauled much specifically for AI search. Topic clustering has still been super important. It's good for building authority and it's good for internal linking, and we know that's important for traditional and AI search: making sure that all your content across the website, from the blog to the product pages to the case studies, connects in a way that not only humans but also search systems can follow. That lends itself to a really good strategy for AI search that I've continued with very happily.

Sara: Something that hasn't been working well, like Dayna said, is going all in on one strategy. For example, publishing layers and layers of AI slop. It looks great when a business is active and publishing content, but when there's no strategy behind it, the internal linking gets super messy. How are you, and how are search systems, supposed to know how everything connects? Updating all that content is also messy, and we know freshness is a big factor for AI visibility, so the content you produce needs to be fresh and citable by all these AI models.

Sara: One underestimated strategy, which I happen to have been working on this year, is localization of website content. Maybe I'm biased because of my languages background. Last year we had a website publishing in English, with German and French only sporadically, and this year we decided to really localize for our clients and prospects in those markets. We've seen more net sales come from those markets, from people who heard about us through AI search. So this is also worth tapping into for B2B businesses, because ultimately it's about building trust and optimizing your content for AI visibility, not just in English-speaking markets but in the other markets you serve.

Adam: Interesting. Going back to the last episode, Veruska was saying much the same about localization: make sure you work with translators in the actual language so you get the nuance and context that goes with it. I think that's brilliant.

Adam: One thing you were saying, Brian, about llms.txt. Yoast integrated an llms.txt generator, I think in late 2025 or early 2026. I'm curious what the panel thinks. I've read a lot about AI poisoning and AI cloaking, where we have public-facing content, but different content in markdown files for LLMs. We only have to look back a couple of months to the great listicle debacle. I imagine there's going to be a lot that happens with gaming the system, and then an overcorrection. Do you think AI cloaking is a legitimate threat, and is there a possibility of being penalized for having different content for humans versus different content for machines?

Brian: I think you could get penalized, maybe not by Google, but by the public. Who's driving those LLMs? Users. If people are typing into ChatGPT, "Is this fastener business reputable?" and ChatGPT says yes, but you hop onto Google Reviews and see a one-star review saying they never got their fasteners, that's an extreme example, but it's a straight tell. In conjunction with all these metrics of share of voice and citations, sentiment is important in the new age of search we're in right now. If you have negative sentiment within AI, it's going to have a ripple effect.

Brian: So if you are AI cloaking, I think of it as black hat SEO in a sense. You're trying to game the system, and for what? To avoid being a good business? Consistency, for human eyes and for LLM eyes, is highly important.

Dayna: I agree with Brian on sentiment, because search is multi-touch now more than ever. Even if people are getting information and doing their comparisons and research in AI platforms, they inevitably go back to the search engine and look at you. If what they see is completely different, that confuses the user and puts bad sentiment out into the universe, and that makes your business automatically untrustworthy. That's where you lose the buyer. Either way, having that validation not match the real-world experience is going to hurt the business.

Adam: We've talked a lot about different tactics. In my practice, I've seen that digital PR plays a pretty disproportionate role. For yours, how does digital PR and third-party coverage play into your strategy for B2B clients? Sara, let's start with you.

Sara: I've been excited about PR for years. We've spoken about building trust in the B2B space, and skepticism is a natural element you have to contend with in B2B marketing, especially in saturated markets like martech, the one I'm in. So you really have to get your name out there beyond your website. It's non-negotiable. Looking at Uberall's AI citations and mentions at the moment, I can see a whole mix of third-party and even competitor websites mentioning or citing us: LinkedIn, G2, YouTube, Search Engine Journal, Reddit, the usual culprits.

Sara: PR has always been important, but now I see it as an extra validation layer on top of your website that emphasizes to AI systems, to Google, to whoever, that you are a trustworthy, credible, authoritative business. Your digital footprint across all of these platforms, especially industry-specific ones, builds that trust with buyers and also that citation safety, so AI systems feel confident recommending your brand to buyers using ChatGPT to find the solutions they need.

Adam: Brian?

Brian: The way I think about it is that a buyer shortlist is going to be built from sources you do not control. Your site can only tell your side of the story. So the real question is: which third parties are the models actually pulling from when someone asks about your category? That's a citation question more than an authority question.

Brian: What I tell people is this. You pull your referring domains and rank them by domain authority, and you have a link-building plan. But if you rank them on how often each domain shows up in the citations for the prompts you're tracking, you get a totally different list, and that list is your PR brief. Some mid-authority trade publication that cites consistently is worth way more than a national placement the models never touch. A high domain authority site that never gets cited for your category is basically a vanity backlink with a PR budget attached.

Brian: The same machinery works when a model is saying something wrong about you. You can't edit a model. That's something we all have to come to grips with. So you starve the bad inputs, you own the facts on the properties you control, you get the accurate version published in the places where models actually read, and then you monitor everything.

Brian: Within SEO, PR has always been in the background. I've worked with clients who had minor to major PR crises. In AI right now, PR is having its own moment of glory, because everybody wants to know why ChatGPT is saying this about me, why Claude isn't recommending me, why Gemini is portraying this business in a certain light. PR and SEO have always had a bit of a crossover, but now it's intertwined, and it's become its own little product within the SEO community.

Adam: Dayna?

Dayna: I agree with the intertwining of PR and SEO always being there. Now, to what Sara said, it's a validation method, especially when you're looking at specific topics and what's being cited. To Brian's point, what product roundups or comparison content is out there that you're being mentioned in, and how you're being mentioned, can give you an idea of sentiment. And if there's somewhere you're not being mentioned and want to be added, that gives you a shortlist of what to go after.

Dayna: If it's an important topic you're tracking, and a certain publication or niche site is showing up consistently for those prompts, then even if your website isn't showing up, maybe that's not the intent behind the prompt, and that's why the comparison content is showing up instead. Being listed there is helpful, not only for the people who read that content, but also for the systems' understanding of your brand and your product in the industry.

Adam: How do we balance between earned and paid? I'll preface this by saying some of the agencies we've worked with have told me that some of the sources they see in their reporting will charge up to three to five thousand US dollars for a single listicle, which seems asinine to me. But I only have to think back ten years, maybe even five, to buying backlinks. So how do we weigh the importance of earned PR versus paid PR?

Sara: I'd go back to what Dayna was saying about looking at where your competitors are being cited that you're not, and filling those gaps. Is it worth paying for PR on these platforms, or are there other industry-specific platforms you maybe haven't explored that are also coming up a lot in these analytics as important AI data sources? I'd look at that one with some skepticism, as you said, because it is the Wild West, buying the backlinks of the AI search era.

Sara: I still think earned PR is a great way for B2B businesses to build trust. Proprietary data is so important: the unique insights into products and solutions that buyers are looking for. With AI search creating imitation content, repeated across websites, especially listicle content, that's the content I hope dies out quickly. Earned PR initiatives are what I think will really perform well in AI search, and buyers are going to appreciate proprietary data sources a lot more.

Adam: Brian, Dayna, anything to add?

Brian: I agree with everybody here. There's a delicate balance between buying PR and the expected ROI from it. I think that's more of a paid media conversation, I'd make that argument. Organically speaking, you want to be cited in those publications. But it comes down to building the foundation. It harkens back to everything we've been talking about: build a good website, be 100% transparent about who you are as a business, and be a really good business. That goes a long way.

Brian: From a PR perspective, if there is a crisis, there's a PR expert I follow, Molly McPherson, and the simplest advice is: own everything you've done, apologize, and put out an action plan. Of course, no PR crisis is linear. But strictly for SEO purposes, buying a PR citation is a bit of a gamble, and I think that's more of a paid media conversation. In my practice, you focus on the organic stuff, and you can control a huge bulk of it just by doing simple, good SEO practices from the start of the engagement.

Adam: Dayna, anything to add?

Dayna: Building off what Sara said about earned PR, come up with ways to showcase your thought leaders. LinkedIn is a pretty good place to post articles, and that kind of thought leadership can get picked up somewhere. Proprietary data and case studies are also good practice, and that's where I'd put my efforts when trying to get earned media or PR. There are a lot of ways to spin that kind of data if you have it, whether in content or pitching, and it can be really valuable in showing the differentiators of your business.

Adam: I've heard everyone mention LinkedIn. From what we're seeing, we're not seeing LinkedIn as one of the main sources at the retrieval layer. Lots of Wikipedia, lots of YouTube, a variety of industry periodicals or publications depending on the industry. Are you noticing LinkedIn coming up more for a specific industry or vertical, or is it blanketed across the board as one of the main sources AI relies on?

Dayna: I think it's a credible source that AI relies on, especially when we're talking about professional services, subject matter expertise, and being able to make that alignment for your business through case studies and thought leadership. As far as social channels go, in the B2B space LinkedIn is important because of that. In the consumer space, you might see more of Instagram, Facebook, and TikTok, whereas LinkedIn shows up more in B2B market spaces, particularly professional services and maybe software.

Brian: In my perspective, most of LinkedIn sits behind a login wall, and it can be tricky for retrieval bots to get past a post, an article, or even a comment thread. So even when there's a great piece of thought leadership sitting there, models have a hard time pulling sentences or any type of content that needs a citation. Compare that with a trade publication, Wikipedia, or a YouTube transcript, where the text is out in the wide open.

Brian: Where I do see LinkedIn surface is mainly on entity prompts, not category prompts. Who is this person? What do they do for their company? Who runs it? The public page and the public profile header get pulled by an LLM. But when a buyer asks how to evaluate vendors in a category, or which platform handles a specific compliance requirement, LinkedIn basically vanishes from that citation set. And that's the prompt set that matters most in B2B.

Brian: The one place it leans more vertical is professional services, consulting, and recruiting, categories where the person is the product. You'll see a bit more of it there, but it's nowhere near the trade press or review platforms. So my typical advice is: don't treat LinkedIn as a citation channel. Treat it as a demand channel. It drives branded search. It gets your point of view in front of the people who matter most, including the editors and journalists who write these publications, and you can get quotes there. It feeds the sources, but it's not the source.

Sara: It always depends on where your target buyer is. A lot of our activities are via LinkedIn, so that's where a lot of my data comes from. You can be on various platforms that function as AI data sources, and the best practice is to diversify, but ultimately it's about where your target buyer is based. If that's LinkedIn, there's really great value in publishing, maybe not company posts, but definitely personal posts, personal articles, and newsletters. There's great value in that, especially in the industries I've been working in.

Adam: I like what you said about personal services. We're seeing the solopreneur this year more than any other year, where the Paul Boags of the world are really coming up. That might be a niche reference for usability and UX research, but it's interesting to see how much LinkedIn in particular is, as you said, Brian, a demand channel, and, as Sara and Dayna said, a cornerstone for sourcing when it comes to professional services.

Adam: One thing we mentioned earlier is that attribution is difficult, to say the least. How are you currently connecting AI visibility to an actual business outcome, whether that's being part of the vendor set, tracking branded searches, website traffic, which is also extremely contentious in AI search, pipeline, or revenue? What metrics are you tracking? Sara, let's start with you.

Sara: We're still working on this one. It's the Wild West. In general, I've never been one for fixating on traditional metrics like impressions and rankings. I connect those more with indications of how healthy a website is, and it's the same for me with AI share of voice, citation rate, mention rate, and all these new metrics we're seeing in AI search. I monitor them to check on the health of the website, but I focus a lot on sales-related or engagement-related metrics that come from AI referrals.

Sara: Sales, obviously, because that directly impacts business outcomes, and engagement, because B2B is a lot about building trust. Even if these AI referrals aren't connecting to business impact today, maybe seven months down the line there will be a conversion or pipeline generated, and it's still important to measure how your website content is engaged with from those AI referrals. Separately, because attribution is so messy, we're also measuring through self-attribution: how customers heard about us. That way we can actually see whether they came from ChatGPT, Perplexity, Gemini, and so on. It gives us that extra bit of granularity.

Dayna: I agree. Attribution is very messy at this point. It's a lot of directional data and correlations. If you're seeing an increase in AI visibility metrics or citations, and you can correlate a tactic you've done with an increase in website traffic in general, that might be a way to connect visibility to people getting to the site.

Dayna: It's hard because search isn't one touch. It never really was, but now more than ever. It's hard to know whether that person went to ChatGPT, typed something in, went to validate on Google, went back and forth, clicked on an ad. So look at your website metrics, your on-site engagement, referral traffic to see where people are coming from, and your general business metrics like CRM data. Then ask customers how they got there, as Sara mentioned. Even then, they might not give you the starting point; they might give you the last touch. It's hard, but as with everything, you're going to have to look at a bunch of different metrics to understand the impact of anything you do.

Brian: I agree with everybody here. It's incredibly difficult. What we've found and been testing at Level is that we have to build a chain where every link is measurable. Visibility and citation share are the first moves. Then AI referral traffic, segmented in GA4, measured out as small volume, high intent. Then branded search, which is the best mid-funnel proxy, because if this is working, people leave the chat and then go search your name. Then lead source via self-reporting: include a dropdown in the RFI form that lists ChatGPT, Gemini, Perplexity, and Claude as options for how did you find us. That's the cheapest attribution fix for most companies, and a lot of them don't do it. The pipeline sits at the end of that chain, not in front of it.

Brian: Before any of it, we have to fix the measurement layer. Look for lost-lead events configured as key events in GA4, so conversion reporting isn't counting losses as wins. Look for organic conversions undercounted badly enough that skepticism about a program is built on a broken number. Find those before you promise anybody revenue attribution. Then set goals and win conditions for six, nine, and twelve months against a documented baseline. That's how you keep a client from expecting pipeline in a month or two. It's incredibly hard, especially with how multi-touch AI search is, but it's not impossible. We just have to think outside the box from a tracking perspective to tie our work to increased revenue.

Adam: All right, everyone. Something we like to do on every episode is a lightning round: quick-fire questions, light and breezy. We'll go Sara, Dayna, Brian. Sound good?

Adam: One free thing every business owner can do this week to improve their search presence. Sara, go.

Sara: Internal linking. I'm going to say it. I'm a big one on internal linking. Explore what your internal linking options are, because I'm sure most brands aren't doing enough of it.

Dayna: I would check messaging across channels: your social channels, your owned channels, your About Us page. Make sure it's aligned.

Brian: Make sure you have Google Analytics and Search Console. Most businesses don't have GA4 set up. From that point on, use the data to your advantage. If you need assistance, there are a lot of free webinars and YouTube videos out there to get the ball rolling. It all starts with having the right tools in place, and Google Analytics and Search Console are 100% free.

Adam: Most overrated or underrated GEO tactic. Sara?

Sara: What Dayna said before: making sure all of your information is consistent and accurate, with no conflicting information to confuse humans or search systems. Making sure everything's aligned is super important.

Adam: So underrated: entity management. Dayna, she stole yours.

Dayna: I know. Internal linking, then, for crawlability and content alignment.

Brian: I'll give you two. Underrated: your technical prowess. Everybody's focused on content, and while it's not the most attractive thing to talk about, everything under the hood matters. Overrated: Reddit. With the recent updates on how Reddit is cited, you don't need to create a Reddit profile and try to game the system. Reddit's still going to be cited, but there are more important things than your presence in a subreddit nobody's ever heard of.

Adam: First thing you look at when auditing a new client. Sara?

Sara: From a content perspective, how messy their content is in terms of topic clusters, whether anything's overlapping, and whether there's actual expertise in the content or it's just surface-level, top-of-funnel content.

Dayna: Poking around in Search Console, understanding the crawlability of the site, and the robots file to see what they may or may not be blocking. Those are always the first things I look at.

Brian: The first thing I look at with a new client is whether they take search seriously and whether they're willing to do the work. I know it's not a website thing, but if a client is not serious about updating their site, or they're very picky, or not open to change, they may not be a good client. That's the number one thing I look for: their willingness to change and think differently. They're reaching out to increase their visibility in organic search, and if they're not willing to change anything on their site, they're probably not a good client to work with.

Adam: Fair enough. One thing everyone is talking about that you think will matter a lot less a year from now.

Sara: I don't have to think about this one. The endless listicles. There's a good way of doing listicle-style content, and it's not what most brands are doing. There's absolutely nothing wrong with comparison content, but these listicles are driving me crazy.

The recording continues with the final lightning-round answers.