Welcome to the Surfaced newsletter, outlining important insights relevant to digital marketers.
SMX Advanced happened in Boston (hometown!) the other week, and I went through a summary on LinkedIn. I grilled vendors about their capabilities, who left me answerless, and the talks in the halls all confirmed we’re in the nascent stages of AI search and not to get attached to technologies or tactics. As digital marketers, we’re all trying to make sense of our new environment, and you’re not alone if your head is spinning a bit.
The other fascinating session I didn’t talk about was about AI operations for search, which went over the AI infrastructure necessary to win at enterprise. Slamming tactics are not a strategy, and to win in this environment, it’s about proper enablement and systems. Very aligned with where I am taking Surfaced.
Scaling AI without systems isn't growth, it’s more like chaos, is how I’d describe it.
Another random tip as we all lean on AI to become a little more streamlined,
Onward!
Google recently started rolling out AI Search insights into Search Console, giving insight into how you are showing up on Google’s surfaces. Typical Google only giving impressions and is directional at best as a high-level indication of how often you’re being cited.
It almost feels like an April Fools’ joke, and I hope they give more data over time, including clicks. First-party measurement for AI visibility is finally here.
Additionally, “Conversational Attributes” for products were announced, which are basically schemas to accompany Merchant Center Feeds. As with most recommendations, start testing to see the impact on live queries, but current Merchant Feed optimization is still crucial, and these attributes seem more like a topping than an ingredient.
Claude released Opus 4.8 almost 2 weeks ago, and before I can even write about it, just the other week Fable/Mythos was released. I have been testing it, and it seems more of the same to me, despite tokens costing 2x as much as Opus.
The key here isn’t the features…it’s about getting closer to artificial general intelligence. Also, the better the results, the faster the adoption and trust with consumers.
Out of the 60 subreddits I’ve been monitoring, about 60% are still asking foundational questions ("what is GEO," "does SEO still matter") while 40% are past that and debating methodology.
To me, this shows that Reddit is a lagging indicator of what’s happening in search if someone like me is mining it for gold to share with you, but there is quite a bit of astroturfing happening there due to its prominence in AI citations. But it does bring up good questions from humans looking to have problems solved.
Picking one and answering it: Is my JavaScript site invisible to AI crawlers?
Absolutely yes. LLM retrieval does not parse or render as Google does (they have MUCH better crawling capabilities). It misses dynamic content and largely will make your content invisible. If you’re curious about how this might affect you, download a plugin to look at the page without firing JS and see what happens.
Most modern CMS’s know this, and developers’ knowledge of SEO has improved over the years. Looking cool is not as important as being crawlable. Use LLMs or Lighthouse to audit your code if you’re unsure.
On LinkedIn this week, some highlights:
Dan Hinckley posted a super practical way to visualize your search console data with topics, giving a great alternative to looking at tables and charts for performance.
Quite a few posts about using AI to sound more authentic. Ruben Hassid gives some nice tips on how to converse better and how to get less flowery or agreeable responses. I think content creators need to really double down on this to get better output. If I hear “quietly” one more time..
Lily Ray continues to bring to light bad practices leading to bad outcomes, and there’s quite a bit of noise around the negative effects of large-scale AI pages. Dozens of cases have flooded social media in the last few weeks. It’s no surprise to me, considering spam and unhelpful content are what are slowing down trust in using AI. If you’re thinking of being black hat, don’t.
A study from Siege Media showed that in B2B, “X vs Y” is the winner when it comes to mid-funnel comparison pages. Though a fairly small sample set, a good data insight to clock if this is relevant to you.
Really clean signal here despite a small data set, and a look at what comparison content performs better on search surfaces is good to know.
Cloudflare released data showing the increase in bot traffic up to 61%, which shows the push towards agentic having serious implications on how your teams develop and code websites when it comes to the agentic web.
TechnologyChecker showed something else interesting related: Bytespider (ByteDance/TikTok) nearly doubled its AI crawl share from 5.7% in April to 10.2% in May, passing ClaudeBot to become the #4 AI crawler overall and the fastest-growing crawler on the list.
While Google dominates worldwide crawling by a significant margin, seeing such crawl growth from a social platform means they are investing in compute and holding market share.
What does all this agentic crawling mean for brands?
Make sure you’re eligible for all the crawlers you intend to serve content to
Comprehensive AI-readiness check for schemas, code, protocols, .txt’s, .md’s,
Google doesn’t own the internet, despite being a titan, and consider optimizations towards video if that’s how your audience connects.
They say timing is everything, and LLMs prefer fresh content. Coordinated efforts can help boost trustworthiness, and there is more of a need for communication & collaboration across channels than ever before.
The way LLMs validate through query fan-outs and ground answers will have a stronger signal if you show up on multiple surfaces.
Think of projects more like campaigns. Certain trigger actions should happen sequentially within days, if possible. This is extremely important for brands that care about timeliness (publications, emerging technology research, etc.). It’s not as relevant for evergreen content creation due to the long, slow value it’s meant to provide.
So talk, organic, plan. Good activations will lead to a much stronger authority for your brand in the age of agents and live content retrieval.
“Consultancy must be good for SEOs today, what are you doing different?”
I got asked this at SMX, and I laughed. I told them that while there’s definitely a lot of hype and FOMO that people are selling, there’s also more access to quality research and data being shared than ever before, which means people are empowered to do things themselves. Resource-strapped teams without dedicated management though, beware.
Companies have the ability and access to create cheaper and lazier strategies than ever before. They also have technology to supercharge the execution side of optimization. I do not advise following LLM-created strategies, despite how confident it might sound. The number of horror stories being posted these days from blind faith is unreal.
Tactics are a dime a dozen, and anyone can ask ChatGPT the top 10 things that work in 2026. Some of them might be great, and some answers might lead you down an exhaustive path toward no reward.
There’s whiplash from the number of things being told on what you can do, but it’s hard to prioritize the most leveraged things to do. Cheap and fast inputs lead to short-sighted outcomes.
This is ultimately what I told this person, put another way and more poetically now:
there’s wisdom in an experienced human, and there’s no wisdom in an LLM's responses.
If something in here sparked a question or there’s something that moved you, let me know. Happy for the feedback to improve future sends.
Appreciate you all. Until the next time,
Andrew