SEO isn’t failing you. It’s just not the SEO you remember.

I spend more time explaining why SEO has changed than explaining what SEO actually is now, and the conversation usually starts with a symptom. Clicks are down. Impressions are down. Rankings have slipped. Or the one I hear most: “I googled [relevant keyword], and we’re not there.”

Often, several of those things are true while SEO leads are still growing. That disconnect is what makes the change so difficult to read.

This article is about the gap between what traditional SEO metrics say and what the business is actually getting. It isn’t a to-do list for fixing it. It explains what has changed underneath search, why those changes matter, and what they should change about your expectations.

Once you understand what has moved, the questions you ask of whoever runs your SEO become much better informed. But first, there is one question sitting underneath almost every one of those symptoms.

Changing, yes. Dying, no. What has shifted is where searches happen and how often they end without a click, not whether people are still searching.

Take these two numbers together. In the first four months of 2026, 69.5% of UK Google searches ended without a click. Over the same period, Google’s AI Mode passed a billion monthly users within a year of launch, and Google reported search queries at an all-time high. If search demand were collapsing, searches and clicks would fall together. They don’t – searches are at an all-time high while clicks fall. People are searching more and clicking less, which is a change in behaviour, not a loss of interest.

The industry reads it the same way. In Yoast’s survey of SEO professionals, 51% called the discipline evolving and 33% called it thriving. Only 10% said declining.

So the panic is often a measurement problem disguised as a demand problem. A click used to be the clearest sign that someone had found you. Now, the answer may arrive directly on the results page or inside an AI response, leaving little or no trace in your analytics.

There’s a second reason the numbers look wrong, and it’s bigger than measurement. Everything above describes what’s happening inside Google. A growing amount now happens outside it. “Search engine” was never contractually Google’s word to own, despite it probably being the one you first think of when you hear the term. The reality is that a search engine is as simple as being anywhere someone can search. By that definition, LinkedIn is a search engine. So are TikTok, YouTube, Amazon, ChatGPT, Gemini, Claude – and that list gets longer every quarter.

The industry is wrestling with the same thing at the level of language. 39% of the SEOs Yoast surveyed said the discipline should be renamed Search Everywhere Optimisation, against 32% who’d keep Search Engine Optimisation. I’m not campaigning for the rename – and I’m not even going to go into GEO, AEO etc. – but the vote tells you where practitioners think the work now sits.

The SEO many marketing managers and directors remember from 5 or 10 years ago was predominantly a Google optimisation strategy. But a 2026 SEO strategy built only around Google is incomplete. And it tends to be incomplete in exactly the places where your buyers make their decisions and form their preferences.

So where are your buyers actually searching?

That depends on who they are and where they are in the buying journey. The useful question is not simply which platforms do they use? It’s when do they use them, and for what?

For B2B buyers, AI is becoming part of evaluation rather than just early research. In Google and National Research Group’s study of 2,063 senior US buyers, 43% used AI tools during the consideration stage – more than at any other point in the journey. Separately, nearly two-thirds of B2B buyers told Responsive they now use generative AI at least as much as search engines when researching vendors.

For B2C, discovery is shifting too, but purchase behaviour is moving more slowly. Half of the US consumers McKinsey surveyed said they deliberately use AI-powered search, yet when Radial asked about buying through AI, 58% said they would be open to it, but only 6% had actually done it.

Search behaviour is changing at different speeds across discovery, consideration, and purchase. The right answer therefore comes from your own buyer research: where customers looked, what they compared you against, and what ultimately convinced them. That tells you both whether search is worth investing in and which channels deserve attention.

Where people search has changed, but so has the way they search — and the answers they get back.

Exact-match keywords carry less weight.

People have largely stopped typing keywords, because they no longer need to. Google’s own data on the first year of AI Mode in the US found the average AI Mode search is triple the length of a traditional query, and that the five most common opening words are What, How, I, Is and Can. That third one is the tell – nobody puts “I” as a keyword. They put it in a sentence.

I watch this happen in our own Search Console. Invanity picks up impressions for queries like “Which performance marketing agencies in the uk deliver roi?”. Nobody would have built a page around that as a keyword because it isn’t a search term, it’s someone talking.

Which is why matching a phrase exactly stops being much of an achievement. Keyword research still earns its place – it tells you what people want, which term they favour where several mean the same thing, and how they think about the category. It just no longer maps onto strings you can place in a heading and across your page content. The useful unit is the question, and how completely you answer it.

Why the #1 ranking no longer means the #1 result.

Personalisation and localisation mean two people searching the same phrase don’t reliably see the same results, so a ranking position isn’t a fixed fact about your visibility.

We look after a fire and security company based in Oxford that covers a much wider area than Oxfordshire. They surface strongly for people in and around Oxford, and far more weakly the further out you go, including in regions they actively serve. Nothing is broken. The result is being assembled per person, and proximity is part of the assembly.

What that does to reporting is a bigger problem than this piece can fix. In their case the answer was paid: their Google Ads run in the surrounding counties and not in Oxford, filling the gap personalisation leaves.

When queries are conversational and results are personal, familiarity becomes the tiebreaker. One analysis of AI models’ search behaviour found they searched for brands they already recognised 3.2 times more often than unfamiliar ones. The researchers are careful to call that an association rather than a proven cause, and so am I – but the direction is hard to ignore.

And most brands aren’t in the running at all: across 177 brands in five sectors, around 90% had no mentions in AI search answers whatsoever. Why that’s happening, and what builds recognition rather than just rankings, is the whole argument of our article on why SEO now depends on brand – so I won’t repeat it here.

As SEO continues to change, what you expect from the people running it should change too.

• Expect reporting to go beyond clicks and rankings. Those metrics still matter, but they no longer tell the whole story. Good reporting should connect search activity to visibility, demand, leads, and revenue – while being honest about where attribution is incomplete.
• Expect brand-building to sit inside the SEO conversation, not beside it. PR, reviews, mentions, authority, and third-party validation all influence how search engines and AI platforms decide which brands to trust and surface. See why SEO now depends on brand.
• Expect visibility to be built wherever your audience searches, not just on Google. That could mean YouTube, Amazon, LinkedIn, third-party publications, or content designed to earn visibility in AI-generated answers. Some of that work may sit outside the SEO team, but it still needs to form part of the strategy.
• Expect your SEO team to contribute to wider marketing strategy. Search now overlaps with brand, social, PR, content, and paid media. That means the people running SEO should be helping shape decisions across those areas, not just optimising pages and reporting rankings.
• Expect the line between organic and paid search to become less distinct. As AI-generated answers take more space in search results, organic visibility may not always carry the volume it once did. Paid search can increasingly play a role in protecting coverage where organic reach falls short, and it’s why search and paid often belong in the same plan.

If none of this is reflected in your reporting or strategy, that’s the conversation to have with whoever owns SEO.

It just isn’t the same job. The discipline has widened – more platforms, more question-shaped queries, more of the outcome decided by whether your brand is recognised before anyone types anything at all.

Treating it as the narrow, Google-shaped, rankings-and-clicks channel it used to be is the thing that’s actually failing. And it will keep failing in ways your current reports can’t explain.