I just got back from SMX Advanced in Boston this week, and there were a lot of great speakers and a lot of great stuff for marketers. So these are my biggest takeaways. Let's get right into it.
Purna Virji: AI Has to Prove Its ROI
The opening speaker was Purna Virji, and my takeaway from her speech was that AI has to prove its ROI. She said time saved is the ultimate vanity metric, and that's definitely something we give AI a lot of credit for, being really fast. But is it actually making things better? And is it actually impacting the bottom line? That's really what you should be focusing on, not just speed gains.
One of her great quotes was, "Better comes before faster." It's fine to be fast, but make sure it's actually making things better. And how should it be making things better? She gave three areas: it should be improving the quality of output, it should be increasing your scope or your competitive advantage, and it should be increasing capability, but sustainably. That was the kickoff of SMX, and it was a great mindset speech.
Kelsey Libert: Become a Source, Not a Summary
The next speaker I attended was Kelsey Libert. Search is expanding across channels. We know this is happening, and marketing is going to have to get more expansive and go beyond just websites. So yes, SEOs, we're going to have to crack the code on socials. This is in line with everything Rand Fishkin has been talking about for a while.
AI search had trust early on but has since lost a lot of it. People want disclosures if you're using it. On my LinkedIn, I've talked a lot about human sensitivity to AI and how many people are out there rejecting it when they identify it. So if we're using it for our clients and the brands we work for, we have to be really careful. You have to govern it. The best thing you can do with AI is give it a lot of context so the outputs aren't average or generic, and you want to supervise it every step of the way so it's not getting off course, hallucinating, and degrading trust in the brand.
She had some stats here. There's a quality problem with AI: 53% of marketing work uses it, but only 26% say it's faster and better, and 48% say it's making the work more average. That's right in line with what Purna was talking about. Adoption pressure is really high for analytics and SEO, so we actually have a lot of responsibility to use it appropriately.
"Become a source, not a summary."
— Kelsey LibertIf you're thinking about Google's AI Overviews, don't just let your content dissolve into the answer without your brand mentioned. You really want to be doing the long-term hard work to build a moat: build a brand, build yourself as an authoritative entity on the web. That's tough, but she gave some tips on how to do it: lean into original data, proprietary studies, and digital PR. We do a lot of this at the agency I work at, where everything we do is expert-led. We schedule regular Q&A sessions with our subject matter experts to keep things as unique as possible, make it hard for competitors to copy us, and make it more likely we're going to be selected by AI. Because if you're writing to the average, what compels AI or any other system to select you?
She talked a lot about repurposing to help build yourself as an authoritative entity, and that's something our agency is big on too. We love starting with video, because video can become long and short-form video, and you also have the transcript to turn into long and short-form text. All of those formats can go to a lot of different platforms, so it really helps with your spread across the web.
Her framework for success in AI search:
- Monitor platforms and choose the ones relevant to your audience.
- Build entity authority by getting your subject matter expertise out there.
- Triangulate visibility with a good spread of platforms where you're publishing.
- Govern for trust when it comes to AI.
Dave Davies: Can This Text Stand on Its Own?
Dave Davies was a more technical speaker, but there were some great takeaways we could apply to our work. When it comes to copywriting, his question was, "Can this block of text stand on its own and answer a specific question?" I just had an example of this the other day with one of my copywriters who wrote a key takeaway that was really dependent on the context from the paragraph that preceded it. You have to be careful with that, because it was a great key takeaway, but it couldn't stand on its own, so it might have a little more trouble being selected for an AI answer.
He also talked a lot about how powerful third-party mentions are for AI visibility, and that was a through line with a lot of speakers. It's not surprising, but there was a lot more talk of mentions being just as powerful, if not more powerful in some cases, than links. And he emphasized that freshness is important when it comes to AI, but if you're going to update your content, make sure the updates are significant. You can't just go in and change the last updated date and think that's enough.
Andrew Beckman and Andrew Shotland: Extracting the Gold From Local Owners
Andrew Beckman and Andrew Shotland held a local SEO seminar that was like an open forum to share your issues. It was really good. I actually went up to Andrew Shotland afterward and talked to him about how local business owners often have content that could grow their business locked away in their mind, and how our job is to extract it and get it out onto the web.
I shared some of the ways I do that. Our monthly meeting is sixty minutes: the front thirty minutes is reporting and strategy, and the back thirty minutes is an SME Q&A session. Another way we try to extract subject matter expertise is to send questionnaires, whether there's a new page we want to create, a page we want to update, or a PR request for expertise. Then I asked him, "What other ways can we extract this gold from our busy local business owners?" And he said, "Why don't you ask if they record sales calls?"
I thought that was a great idea and wanted to share it. Sales call transcripts and recordings: you can mine those for subject matter expertise and customer pain points. There's a lot of gold in there. You can also start to understand the language customers use, and that might alter your copywriting a little bit.
Amanda Farley: Conversational Queries and a Local Video Tip
Amanda Farley leaned more into the e-commerce side, but she had a local tip I'll share too. She had a system where she took in data on search behavior, product intent, and audience intelligence, and gained insights she could then apply to product pages.
What we're finding with conversational queries is that there's a lot more detail about what people are looking for. They average about 24 words, from what I heard in another presentation, whereas traditional SEO keywords average about four words. So there's a lot more data in these prompt-based queries, and we have to account for that on our pages. One example I've given is someone might prompt for men's Nike shoes, size 11, that are good for both running and hiking and also for narrow feet. There's a lot of detail in there, and if our data is telling us that's important to people, we want to update our pages accordingly so the AI system can make those connections. It doesn't mean we have to make our pages a mess of fine details, but you do want to do this research so you can gain those insights and then intelligently and tastefully update your page content.
Her one great local tip, which harkens back to what I talked about with Andrew Shotland, is to capture local video in the field using Meta glasses and build a video bank for your clients. Then again, you've got video you can use on multiple platforms, and if there's any talking, it can become text. Meta glasses are not as expensive as I thought they were, so this is one I'm going to try.
Dawn Anderson: Lost in the Middle
Dawn Anderson gave quite a technical talk breaking down how AI systems and large language models read text. One of the big things was a paper called "Lost in the Middle": if you have an enormous amount of text or context, the large language model will tend to read what's at the top and the bottom of it. So the takeaway is to put your most important stuff at the top of the page so it doesn't get missed.
Then she came in with a great point: a lot of this is just good journalism and common sense, isn't it? That's definitely something for SEOs and marketers to take away. With AI coming into the fray, things are getting really technical. We're looking under the hood and trying to understand what's behind these systems, and I think that's great to do. But at the end of the day, it boils down to principles of good copywriting and SEO fundamentals. Get into the detail and understand it, but don't get lost in it and lose the bigger picture.
Monica Ho: Treat Reviews, Search, and Social as One System
Monica Ho shared a presentation based on a large data study of businesses with multiple locations. Here are some cool takeaways:
- Reviews are a gate. With AI especially, she found you don't even get in unless you have 4.3 stars.
- One in four people use social as primary search. That harkens back to what Kelsey and Rand Fishkin have been talking about: search is fragmenting, and you want to be everywhere your target audience is. SparkToro is a great tool for that. You type in some information about your target audience, and it shows you the social platforms they use, the AI platforms they use, and a ton of other great data, so you can get your content in front of them.
- One in three people use generative AI in over half their searches. A lot of people are using AI search.
She emphasized auditing your data, and this is tough to do. I think we're going to have to be more thorough as SEOs.
Heads Up
One example she gave was a local business that showed up in an AI search where people were seeing one of their offers, but that offer had long expired. It turned into a nightmare for them: they had to manage all these inquiries about an offer that had long since passed, and letting these people down might not have been helping the brand much.
This is the kind of stuff we have to monitor, because the AI systems are grabbing onto it, putting it in front of people, and confidently telling them it's current information.
She said to treat reviews, search, and social as one system, not as separate fragmented teams, and I think that's a great mindset.
Beth Nunnington: Build Something Worth Finding
For AI search, pretty much everything Beth Nunnington said was great. First of all, stop chasing GEO or LLM summaries and build something worth finding. This is sort of what I was getting at with not getting lost in the detail or pulled into the hype of getting into an AI response. Think bigger than that.
So lean into your experts. From Beth: "Your experts are your strongest asset." I totally agree, especially in the AI era. They always have been, but now especially. It's not about manipulating outputs. Do the long-term hard work. Quick wins are fine for SEO, but they tend not to be lasting. You want to do things that are going to be sustainable despite how many Google algorithm updates are happening. Build a strong foundation, then build on top of it with good SEO fundamentals and strong PR.
This SMX definitely focused a lot on third party: the wider web pointing to you with something, a link or a mention. That is really how to build yourself as an entity, how to build authority, and how to get into AI outputs. Nothing new, but good to reinforce, and I think it's even more important when it comes to AI. She also emphasized that true brand building needs multiple teams working together, which is the same cohesion point Monica was making.
Grant Simmons: A Quick Word on EntityMap
Just a quick mention of Grant Simmons. He's involved in entitymap.org, and I encourage you to visit the site, see what this proposed protocol is about, and try the tool where you can experiment with it. It's quite similar to Schema, but it's a different kind of thing. I think it's a supportive layer for getting your content and subject matter expertise out there on many platforms with a lot of third-party support, confirmation, and additional context. Those are the primary drivers of AI and search systems understanding you as an entity, and the stuff at entitymap.org can reinforce that even further.
Kyle Risley: AI Will Work Around Our Mess (To a Point)
Kyle Risley was from Shopify, and there was a lot of talk about AI agents coming and how they're going to do a lot more of the navigating people used to do. You need to make sure that from a technical, foundational standpoint, you're as sound as possible. But at the end of the day, he emphasized that you just need to nail the basics. AI will work around our mess.
He invited us to raise our hand if we thought people were going to make perfect websites. Of course they're not. Google deals with this already: there are websites with multiple H1s on the page, and Google can figure out the actual primary heading through other clues. Google learned to work around our mess, and AI is going to as well, but only to a certain extent. You do need to make sure you get those basics right.
The Four Bigger Takeaways
That's a little from each individual speaker. Here are the four larger takeaways.
My Four Big Takeaways from SMX Advanced Boston
- AI accountability. AI needs to prove its value, and it doesn't do that by showing it can do things faster or generate larger outputs. Take that story all the way to the end, to ROI. It has to earn its keep through expansion: quality, scope, capability, and ultimately ROI. Those are the percentages that should be increasing, not just speed.
- Trust. We have a real responsibility as marketers to use AI in a way that improves our work but also keeps brand trust intact. That means deep context and very careful supervision.
- Empathy. When it comes to audience research, SEOs and marketers are going to have to get a lot more thorough. We need to be even more empathetic to our audiences to understand all the nuances of what they might be looking for and the types of searches they might do. If we don't mend those gaps, we may not be chosen by AI search systems, and we'll miss out on business. So go deeper on audience research, map every search behavior and intent, and make sure our content reflects it.
- Authority. Build it by doing the difficult long-term work to create a moat that's original, hard for competitors to copy, and stable against whatever comes next, whether it's algorithm updates or a new technology. Don't just be part of the summary. Become a source.
So there you have it, all my takeaways from SMX Advanced Boston this year. I hope that was useful. If you have any questions, drop me a comment. I love hearing from you, and I'll see you in the next one.
