Skip to content

Insights

SEO Didn’t Die. The Click Did.

Ali McDonald, Associate Director, Organic PerformancePublished Updated 8 min read

Most Google searches now end without a click. What 18 months of AI search changed for SEO, what it didn’t, and how we measure visibility heading into 2027.

A search bar surrounded by floating answer cards, with a cursor that never reaches a link

What 18 months of AI search actually changed, what it didn’t, and where we’re putting our clients’ effort heading into 2027.

Search didn’t shrink. The click did. People are asking search engines and AI assistants more questions than ever, but a growing share of those questions get answered before anyone visits a website.

It happens every few years: someone declares SEO is dead. This year, that claim finally has some data behind it, just not the data people think.

Search is now bigger than it has ever been. At I/O in May, Google said AI Overviews now reach 2.5 billion people a month and AI Mode about a billion. ChatGPT passed 900 million weekly users in February. People have never asked machines more questions.

What collapsed is the click: the moment a searcher leaves the results page and lands on your site.

A recent SparkToro clickstream study found that 68% of US Google searches between January and April ended without a click. Ahrefs found the top organic result loses roughly 58% of its click-through rate when an AI Overview sits above it.

So the question for brands has changed. It used to be “how do we rank?” Now it’s “when someone asks a machine about our category, does our name come up, and is it described the way we’d want?” That question is what we’ve rebuilt our organic practice around. It’s worth unpacking, because a lot of the advice in the market right now is either panicked or wrong.

Your rankings are fine. The scoreboard is broken.

The most common conversation we’ve had with clients this year starts the same way: “Our rankings held, but traffic is down. What happened?”

Usually, nothing is wrong with the rankings. The page in position one is simply sitting under an answer that satisfies a growing share of searchers before they scroll. News publishers felt it first. Reuters Institute research using Chartbeat data from more than 2,500 sites found Google organic traffic down 33% globally and 38% in the US between November 2024 and November 2025.

Informational content takes the hardest hit because it’s the easiest thing for a model to summarize. Transactional and local searches hold up better, since people still want to see the product, the price or the provider before they commit.

The old KPI stack of rankings, sessions and CTR now describes a shrinking slice of reality. If organic traffic is your only success metric, you’ll report decline every quarter even while your brand earns more visibility than ever. We now report in three layers:

  • Visibility. Are we present for the prompts that matter across Google’s AI features, ChatGPT, Perplexity, Gemini and Copilot?
  • Representation. When we’re named, is the description accurate, current and favorable compared with competitors?
  • Outcome. Branded search lift, direct traffic, assisted conversions and pipeline. The clicks that do arrive tend to come later in the journey and convert better, so conversion rate matters more than raw volume.

Google says AEO is “still SEO.” Google is right, about Google.

In May, Google published its first official guide to optimizing for AI Overviews and AI Mode. The message was blunt: from Google Search’s perspective, answer engine optimization and generative engine optimization are “still SEO.” It went further and named tactics site owners can ignore for Google, including llms.txt files, chunking content into tiny pieces, rewriting pages in a special “AI style” and AI-specific schema. It also cautioned that chasing inauthentic brand mentions across the web isn’t as helpful as it might seem.

We welcomed it. A cottage industry grew up in 2025 selling AEO packages built on unverified tricks, and plenty of brands paid for them. If someone pitches you an llms.txt file as the key to Google’s AI results, you now have Google’s own documentation to push back with.

But Google’s guidance covers Google’s surfaces. It doesn’t describe how ChatGPT, Perplexity or Claude choose their sources, and those systems draw from different places. Even within Google, an Ahrefs analysis found that only about 38% of AI Overview citations come from pages in the organic top 10. Visibility in a generated answer is related to ranking. It isn’t identical to it.

Our position: the foundation is SEO. Crawlable, fast, well-structured pages, written by people who know the subject, answering real questions better than anyone else. Nothing replaces that, and AEO without it is decoration. What has expanded is everything that happens off your website.

Your brand is now what the internet says about you

This is the finding that should change budgets. Ahrefs studied 75,000 brands and found that branded web mentions correlated with visibility in AI Overviews at 0.664, roughly three times the correlation for backlinks at 0.218. A follow-up study across ChatGPT, AI Mode and AI Overviews found that mentions in YouTube titles, descriptions and transcripts showed the strongest single correlation, at about 0.737.

That makes sense once you think about how a language model forms an opinion. It doesn’t rank your page. It builds a consensus from everything it has seen and retrieved: reviews, Reddit threads, comparison articles, trade press, video transcripts, forum answers. If the web consistently describes you as the best option for a specific job, the model will too. If the web is silent or contradictory, so is the model.

The practical consequence is that SEO, PR, content, social and customer experience now feed the same system. A review in a niche publication, a founder’s appearance on an industry podcast, a well-moderated community answer: these used to be “brand” activities with fuzzy attribution. Now they’re inputs into the answer your next customer reads.

The link-building playbook needs rethinking too. Links still correlate positively and still drive classic rankings. But a placement nobody reads, on a site nobody trusts, does very little for how a model describes you. Coverage that real people discuss does.

Being cited is not the same as being chosen

Most AI visibility dashboards blur an important distinction. A citation is a link in the source list. A mention is your brand named in the answer itself. Semrush research published in June found that 62% of AI citations don’t come with a brand mention. The model used the page, then recommended someone else, or no one at all.

For a publisher, a citation might be the goal. For a brand, the goal is the mention, ideally with an accurate description and a reason to choose you. We’ve watched teams celebrate citation counts while a competitor was the one actually named in the answer. When we audit AI visibility, we track both, weight mentions more heavily and check each engine separately, because a brand can dominate one assistant and be invisible in another.

This changes how content should be written. Pages built purely to be extracted, a definition here and a stat there, feed the model without giving it a reason to name you. Content that pairs a clear answer with a distinct point of view, proprietary data or first-hand experience gives the model something it has to attribute. If an insight could have come from anyone, the credit goes to no one.

Your next visitor may not be a person

This is the shift most brands haven’t planned for. At I/O, Google introduced search agents that monitor information and complete tasks in the background, along with a redesigned search box built to hand queries off to AI. OpenAI expanded its Agentic Commerce Protocol to power product discovery in ChatGPT. Google and Shopify co-developed the Universal Commerce Protocol, and select brands are already selling inside AI Mode and the Gemini app. Adobe reported AI traffic to US retail sites up 393% year over year in Q1 2026.

The road hasn’t been smooth. In March, OpenAI stepped back from its original Instant Checkout and handed checkout back to merchants; Shopify merchants can now sell inside ChatGPT, with Shopify running the checkout. The direction is still clear: more of the research, comparison and even the purchase will happen inside an assistant, with an agent reading your site, your product feed and your reviews on the customer’s behalf.

For e-commerce, product data is now a visibility factor in the broadest sense. That means complete, accurate and consistent attributes across your site, your Merchant Center feed and your marketplace listings, with pricing, availability, shipping and returns stated plainly. For B2B and services, your pricing logic, integrations, use cases and proof points need to be findable and unambiguous. An agent building a vendor shortlist won’t fill gaps generously. It will move on to someone whose information is complete.

What we’re telling clients going into 2027

  1. Keep investing in the fundamentals. Technical health, crawlability, page experience and genuinely expert content are still the price of entry for every AI surface, Google’s included.
  2. Stop buying hacks. If a tactic’s main selling point is that it’s built for AI rather than for people, be skeptical. Google has now said as much in writing.
  3. Measure visibility, not just visits. Build a prompt set that mirrors how real customers ask about your category, track presence and sentiment across engines every month, and connect it to branded search and conversion data.
  4. Treat earned media as an organic channel. Reviews, community presence, video, podcasts and trade coverage shape how models describe you. Plan them with the same rigor as on-site content.
  5. Publish what only you can say. Original data, firm opinions and real experience earn a mention instead of an anonymous citation.
  6. Get your data agent-ready. Clean product feeds, structured information and consistent facts everywhere your brand appears.
  7. Reset expectations on traffic. Fewer clicks is the new baseline for informational content. The goal is to be the brand the answer names, so when someone is ready to act, you’re who they look for.

The bottom line

The brands losing ground right now mostly don’t have bad SEO. They’re measuring 2026 with a 2019 scoreboard, or chasing tricks that promise a shortcut. Search didn’t get smaller. It got more selective about who gets named.

The work now is making sure that when your customer asks, the answer includes you and gets you right.

Curious how your brand shows up in AI answers today? Talk to us. We’ll run the prompts your customers are asking and show you where you’re named, where you’re missing and who’s taking your place.

Now, let's connect and continue the journey together

Contact