AEO, GEO, AI SEO: Insights from SEMRush’s Spotlight 2025

Semrush’s Spotlight 2025 event brought together leading SEO specialists and digital marketers for a day of learning and workshops on the key developments in search marketing.

As the Head of SEO at Huble, here are my highlights from the event, and, six months on, what happened when Huble applied those lessons to its own site.

In October, I was fortunate enough to attend the Spotlight 2025 conference hosted by Semrush in Amsterdam. Unsurprisingly, “AI” was the disruptive term on everyone’s agenda that day.

The theme of practically every talk, covering the impact of AI search on SEO performance, how to fit AI into your SEO strategy, the evolution of buyer behaviour, the KPIs we need to add to our reporting, and the new features Semrush is launching to address this new reality.

At Spotlight 2025, Semrush’s CMO Andrew Warden summed up the feeling in the room best when he called the launch of ChatGPT in November 2022 “the most disruptive event in the SEO industry since the creation of Google in 1998”.

Disruption can cause chaos, but it is also an opportunity — one that aligns with Huble’s Answer Engine Optimisation (AEO) service, which we’ve been rolling out for clients (and, as you’ll see, for ourselves) ever since.

AEO, GEO or AI SEO: what should we call it?

There’s no industry consensus yet. At Huble, we use “AEO” (Answer Engine Optimisation) because it matches the terminology and tools inside the HubSpot platform — but AEO, GEO, AI SEO, AIVO and LLMO all describe broadly the same discipline.

Answer Engine Optimisation (AEO), Generative Engine Optimisation (GEO), AI Search Engine Optimisation (AI SEO), AI Visibility Optimisation (AIVO) and Large Language Model Optimisation (LLMO) were all mentioned on the day, used near-interchangeably.

Personally, I think GEO has too much crossover with geography and geology. Run keyword research on it and your search volumes will be inflated with irrelevant variants.

We chose “AEO” so that we speak the same language as HubSpot does: as HubSpot’s 2024 Global Partner of the Year, that consistency matters to us. But I also liked Semrush’s Leigh McKenzie’s case for “AI SEO”: the core fundamentals remain the same: great content that is understandable, relevant, trustworthy and solves the user’s problem.

And the name makes clear that an SEO team’s experience should drive these strategies. That the industry hasn’t settled on a term yet shows just how early we all are on this journey.

How should SEO impact be measured in the age of AI search?

Traffic alone is no longer a valid success metric. Measure the contacts, pipeline and revenue your content influences, which requires a CRM that connects content to outcomes, and add AI visibility as a KPI in its own right.

James McCormick of IDC shared research that since March 2025, businesses have seen a 25% drop in organic traffic across the board, driven by a completely new customer experience: information is brought directly to users in Google’s AI Overviews, ChatGPT, Gemini and other answer engines.

One study showed 78% of businesses still use clicks and traffic as their primary KPI for content, described on the day as “a wake-up call about the true value of organic traffic”.

I’ve spent the past decade working almost exclusively with companies using HubSpot to track leads, sales and revenue,  feeding those insights back to marketing.

Relying solely on traffic as an indicator of content and SEO effectiveness is no longer valid — especially when comparing performance to the pre-AI era. You need a system that tracks the contacts your content generates and their journey to revenue, and you need a way to understand how your brand performs inside AI answers before users ever reach your website.

What AEO tools do you need?

You need two capabilities: prompt-level visibility tracking (is AI citing you, where, with what sentiment?) and site-level intelligence (can AI engines crawl, understand and cite your content?).

Several tool categories now cover this.

At Spotlight 2025, Semrush One was announced: an all-in-one toolset combining SEO and AEO reporting. HubSpot’s AEO tools give brand-level visibility, grading and CRM-connected recommendations, and specialist vendors like ELELEM analyse how AI engines interact with your site and what attributes content needs to be easily digested and cited.

We’ve since tested this tooling landscape in depth — including on our own site. For the full breakdown of how HubSpot AEO compares with standalone platforms like Profound and Peec AI, read our comparison: HubSpot AEO vs standalone tools: which approach fits your business.

If you’re a HubSpot customer wondering what to prioritise, start with AEO for HubSpot: what enterprise teams should be doing now.

Why do AEO insights benefit your entire business? 

An AEO visibility and sentiment report works like a continuous customer survey: near-real-time insight into how your brand, products and pain points are discussed; useful to product, support and leadership, not just the SEO team.

AI tracking software costs more than basic keyword tracking, and the overhead can be off-putting. But these tools aren’t just for SEO reports. Visibility and sentiment on a product can surface FAQs about its specifications, or pain points and troubleshooting themes, which feed straight back to product development and customer support.

Semrush’s Andrew Warden calls this focus “Digital Brand Visibility”, something SEO can lead, but every part of the business contributes to and benefits from. Budget for it accordingly: from the company pool, not the SEO line alone.

Why are content and AEO intrinsically tied?

AI engines value freshness far more highly than traditional search engines, and they can only cite what exists. If you’re not producing content that controls your narrative, AI will tell your story in someone else’s words — or not at all.

Marcus Tober’s talk made the point that content is still what feeds AI its information.

This is the answer SEO and content teams should give stakeholders who see falling traffic numbers and question the value of content production in the age of AI: the channel has changed, the need for authoritative content has not.

 Can you trust what AI engines cite? 

Not uniformly. Studies presented at Spotlight showed ChatGPT cites spammy sources more often than Google’s AI Mode, which leverages decades of authority signals. Each engine cites differently; if you’re weaker on Gemini than ChatGPT, that’s a signal to build trust, credibility and authority.

Tilman Harmeling of Usercentrics also showed that users are comfortable adopting AI but don’t fully trust it yet, neither the answers or the handling of personal data. It can be frustrating to see spammy citations chosen over your well-researched page, but as each engine evolves, I’d expect the more trustworthy content to rise to the top.

Six months on: what happened when we applied this to Huble  

We ran the Spotlight playbook on huble.com from April 2026. In the first six weeks, mean AI visibility across our 100 tracked prompts rose from 22.2% to 26.5%, median visibility nearly doubled from 11% to 20%, and the number of prompts where we were completely invisible fell from 42 to 31.

It’s one thing to come back from a conference energised; it’s another to prove the advice works.

So in April 2026, we did to ourselves exactly what we do for clients: we established a measured baseline and started closing the gaps.

First, we measured

Using Peec AI, we built a library of 100 prompts a senior buyer might realistically ask about our space — partner selection, migrations, GTM strategy, data readiness, change management — and tracked whether Huble appeared in the answers across ChatGPT, Gemini and Perplexity.

The baseline (15 April 2026) was humbling. Mean visibility was 22.2%, but the median was just 11% — half our prompts barely registered. On 42 of the 100 prompts, Huble was completely invisible.

The one consolation: when we did appear, we averaged position 1.92, near the top of the answer. The problem wasn’t how AI talked about us; it was how often.

Then we found the embarrassing part

A technical audit turned up something genuinely ironic for an agency selling AEO services: our own robots.txt was blocking AI crawlers — including ChatGPT’s, Perplexity’s and Claude’s bots.

We were asking AI engines to cite us while locking the door. We also had no llms.txt file. If there’s one lesson to steal from this post, it’s this: before you invest a penny in AEO content, check that AI engines can actually reach your site.

It takes ten minutes, and it’s where more businesses than you’d expect are silently failing.

Then we did the work

With crawlers unblocked and llms.txt in place, we prioritised the 42 zero-visibility prompts — concentrated, not coincidentally, in the areas most aligned with what our ideal clients ask about: migrations, data and AI readiness, and AEO itself.

New and refreshed pages were structured the way this post now is: question-format headings, an answer-first paragraph under each, FAQ blocks with FAQPage schema, and internal links with descriptive anchors.

A technical audit turned up something genuinely ironic for an agency selling AEO services: our own robots.txt was blocking AI crawlers — including ChatGPT’s, Perplexity’s and Claude’s bots. We were asking AI engines to cite us while locking the door.

We also had no llms.txt file. If there’s one lesson to steal from this post, it’s this: before you invest a penny in AEO content, check that AI engines can actually reach your site. It takes ten minutes, and it’s where more businesses than you’d expect are silently failing.

What moved

Six weeks is a short window, and single-month AI visibility data is noisy, but the direction and the breadth of the movement (11 prompts off zero) came from exactly the pages we’d restructured.

Freshness, crawlability and answer-first structure: the Spotlight speakers were right.

And what didn’t (yet)

 Honesty is the point of this section, so: not everything went up. 

AI answers are probabilistic and competitive; visibility is a share you defend continuously, not a ranking you win once. That’s also why we report on a like-for-like prompt basis whenever the tracked set changes. If you can’t trust your own baseline, you can’t trust your trend.

Where should you start with AEO?

Start by asking, “What is AI saying about our business, and where is that information coming from?” Your answer determines your priority: build visibility, correct the record, or amplify what’s working.

If your visibility in answer engines is poor

Focus on increasing brand visibility and defining your niche. Next steps: brand strategy, content creation, local SEO, digital marketing, social media, and backlink strategy. (This was Huble’s own position in April — see above for what we did about it.)

If sentiment is poor or information about your brand is inaccurate

Focus on correcting it. Next steps: brand marketing, PR, website messaging updates, schema markup, product marketing, customer service, and community management.

If AI references you positively, accurately and often

Focus on amplifying your message. Next steps: paid media, social promotion on Reddit, LinkedIn, and YouTube.

No SEO team has infinite capacity, so identify the actions that deliver the best results at the right time. And remember: we’re all still finding our way. No one has all the answers yet, but two decades of SEO experience is a reliable compass for overall direction.

 Frequently asked questions 

What is the difference between AEO, GEO and AI SEO?

They largely describe the same discipline: improving how a brand appears in AI-generated answers. AEO (Answer Engine Optimisation) is the term HubSpot uses; GEO (Generative Engine Optimisation) and AI SEO are common alternatives. The fundamentals are shared: content that is understandable, relevant, trustworthy and structured so AI engines can cite it.

Does traditional SEO still matter for AI search?

Yes. Technical SEO, metadata, sitemaps and content optimisation remain the foundation — AI engines rely on much of the same infrastructure. The additions are AI-specific: allowing AI crawlers in robots.txt, publishing an llms.txt file, answer-first content structure, and schema markup.

How do you measure AEO performance?

Track a fixed library of prompts your buyers realistically ask, across multiple AI engines, and measure visibility (how often you appear), position and sentiment over time. When the prompt set changes, compare on a like-for-like basis so trends stay honest. Pair prompt tracking with site-level intelligence showing how AI engines crawl and cite your content.

How long does it take to see AEO results?

Faster than traditional SEO in our experience: on huble.com, meaningful movement appeared within six weeks of unblocking AI crawlers, and restructuring priority pages: 11 previously invisible prompts began citing us. But AI visibility is volatile month to month; judge trends over quarters, not weeks.

What should you check before investing in AEO content?

That AI engines can reach your site at all. Check robots.txt isn’t blocking AI crawlers (ChatGPT, Perplexity, Claude, Google-Extended), publish an llms.txt file, and fix broken links and crawl-depth issues. Content investment is wasted if crawlers can’t reach the pages.

Summary

The Spotlight event reassured me that the approach to AEO we’d been developing at Huble aligned with the industry’s best thinking. Six months of our own data says it works, and that it’s never finished.

If you want to know what AI is saying about your business, and what to do about it, our AEO services team can show you. [CTA link: /marketing/aeo-services]