It doesn’t matter who the customer is, but what they want

Andrej Salner
Andrej Salner
Senior konzultant
July 16, 2026
11 min read
It doesn’t matter who the customer is, but what they want

At our Basta Stage event in May, we discussed the biggest shift in marketing over the past 25 years. Here is the essence of my presentation in an easy-to-read format, including data that should be of interest to anyone with a “performance marketing” line item in their budget.

We’ve built marketing on identity. That identity is crumbling.

All of modern digital marketing rests on a single idea: we figure out who our customer is and target them accordingly. Personas, demographic targeting, cookies, remarketing, lookalike audiences. For two decades, this worked brilliantly, and remarketing was arguably the most powerful weapon in our arsenal.

Today, this model is under attack from two sides at once.

We’re all familiar with the first attack: privacy protection. GDPR, the end of third-party cookies, Apple’s App Tracking Transparency. If we operate within the law, we know almost nothing about a large portion of website visitors today. We don’t know where they came from, where they’re going, or where they’re leaving to.

The second blow is less visible but far more significant. The new interfaces people use to search don’t ask who the user is at all. They respond to what the user wants. When someone asks ChatGPT for suitable running shoes, the AI doesn’t need to know their age, gender, or income bracket. It needs to understand their situation, and a conversation is all it takes to do that.

It’s worth noting that demographic targeting has been flawed for a long time. Google recently published figures that should be posted on the wall of every marketing department. Over 60% of people who buy sports equipment and watch related video content are women. Over 60% of people searching for cars on their mobile devices are women. Over 40% of people searching for household goods are men. A marketing model based on stereotypes and oversimplifications misses a huge portion of actual demand.

Source: Google

For large-scale brand campaigns, demographics and interests still make sense. But if we don’t have a massive budget and want to address a specific demand, we need to dig deeper

The monopoly on intent is over

For many years, we’ve been accustomed to the idea that information about a customer’s intent comes from a single source. It’s called Google. People faithfully revealed their intentions to it through the search bar, and we were able to target ads based on those intentions. It’s no coincidence that Google Search has become the largest advertising channel on the planet. Search advertising brings Alphabet over $50 billion per quarter.

And it’s no coincidence either that when someone reveals their intent to Google, Google is increasingly keeping that information to itself. And when it does send a visitor elsewhere, it’s usually to “foreign territory.”

The monopoly on intent is crumbling in three waves.

Vertical search. We look for products on Amazon, Alza, or Heuréka; tutorials on YouTube; and accommodations on Booking. According to data from Datos and SparkToro, Google now has “only” a 78.5% share of searches in the EU and the UK. Amazon has 5.35%, ChatGPT 3.68%, and YouTube 2.87%. For some categories, Amazon SEO or YouTube SEO is now more important than ranking on ChatGPT.

Source: Datos / SparkToro

Social media as a search engine. Younger users search for restaurants, fashion, and product reviews on TikTok and Instagram. You can’t “rank first” there. The algorithm decides who gets the answer.

AI responders. The latest and fastest trend. The question still exists, but the click no longer does. And the intent never reaches the advertiser as a signal that could be targeted.

All of this means that information about what we want is now fragmented across dozens of platforms. And both the top and middle of the funnel are increasingly taking place outside our website. The website is becoming primarily a place of conversion, where customers arrive having already made up their minds.

What AI does to clicks: hard numbers

We’ve written in detail about the impact of AI Overviews on Slovak websites, as well as what AI Mode is and how to optimize for it. The numbers have only gotten stronger since then.

In February, Ahrefs published an update to its study covering 150,000 keywords. The presence of AI Overviews reduces the click-through rate to the first organic position by an average of 58%. In a previous study based on data from 2023, that figure was 34.5%. The trend is accelerating and has been independently confirmed by Seer Interactive, Authoritas, and others. Incidentally, the first position is the hardest hit. For certain types of search queries, it no longer makes sense to push from the fifth spot to the first as it once did.

Source: AHREFS 2026

Even more interesting is how people behave when they shop directly through AI. Kevin Indig, working with Citation Labs and Clickstream Solutions, tracked 185 real-world shopping tasks (laptops, TVs, appliances, insurance) in both Google AI Mode and classic search. The results are worth noting:

  • 74% of users accepted the AI’s shortlist without any external verification. No comparisons, no second opinions.
  • 64% didn’t click on anything throughout the entire task.
  • The average rank of the final selection was 1.35. The AI’s choice becomes the customer’s choice.
  • For laptops, three brands accounted for 93% of all final selections. In traditional search, the distribution was significantly broader.

The last figure deserves its own sentence. AI search is an environment where the winner takes almost everything. If you’re not on the shortlist, you don’t exist.

And one bit of irony to conclude this section: Google trains its models, among other things, on Reddit content. In Indigo’s study, Reddit appeared in 19% of traditional search queries but only twice out of 149 queries in AI Mode. Users have stopped visiting the source because the AI serves them a summary of it directly.

Important: this does not mean that AI will replace search. The costs of AI search are still many times higher, and Search itself continues to grow. Therefore, the most significant AI outcome so far is not standalone chatbots, but rather AI Overviews in search.

Funnels are no longer helping us

See-Think-Do-Care, AIDA, and similar models share a common foundation. They segment people based on the stage of their relationship with our brand. However, AI interfaces and algorithmic feeds don’t see people in stages. They see moments of intent. The same person might be in the “See” stage for one category in the morning and in the “Do” stage for another in the afternoon.

Planning communication based on the audience’s status is like mapping a city based on who lives there. But people move around.

It’s more practical to plan using the “intent × platform” matrix. For each type of intent—that is, orientation, comparison, decision, and transaction—ask yourself where the brand should be present and in what form. In Google’s featured snippet, in an AI-generated answer as a quote, in a YouTube comparison video, or in a community as an authentic response.

Same words, different intent. The web as a filter.

Here’s something that almost every website we audit underestimates. The same search query can hide completely different intents. “Running shoes for flat feet” might be searched by someone with chronic pain, a beginner runner, someone buying a gift, or a competitive runner optimizing their training. Four different customers, one phrase.

Conventional wisdom says: one landing page, one assumed intent, one CTA. In an era when AI brings us visitors without any profile, that’s not enough. The page must function as a filter. It must offer multiple entry points at once and let the visitor identify themselves through their behavior—what they click on, where they pause, and what they expand.

Engagement with content is a new source of first-party data on intent. We don’t need to know who the person is. It’s enough to track what caught their interest and tailor further content, CTAs, and sales follow-ups accordingly.

Intent can be gleaned from reviews and discussions

Most companies view reviews, forums, and discussions under content as part of reputation management. Yet this is the cheapest form of customer research on the market. Real people write in their own words about what they’re dealing with, what frustrates them, and what they’re missing. With today’s AI tools, this corpus can be processed at a fraction of the cost it used to be. You can extract the language customers use to describe their problems—which is exactly the language they’ll use to prompt the AI. You can uncover unmet needs as well as real purchasing objections, which are often not what we assume them to be.

These findings apply to the online experience—that is, to texts, FAQs, and page structure—as well as to offline operations—that is, to processes, training, and product changes. And here’s a bonus to wrap things up. The very same content you’re learning from is the source from which AI models also learn. Those who understand their customers’ language and translate it into their own channels become a stronger signal for AI as well.

The new “SEO”: We optimize to be the answer

To summarize the points above, the very goal of optimization is changing. It’s no longer about getting a specific user to a page. It’s about being what the AI recommends when a user expresses their intent. The three classic layers of SEO are thus taking on new meaning.

Technical SEO is evolving into feed integrity. The bot doesn’t browse your website like a human. It queries your data. A structured product feed becomes your storefront.

On-page SEO is evolving into Information Gain. Language models already know from their training the consensus on what everyone has already written. They’ll cite whoever brings something extra to the table: your own data, your own tests, precise specifications, and comparison charts. In other words, marketing can’t gloss over inferior product specifications. If you claim to be the best running shoes for flat feet, the model will verify this against parameters, not against adjectives.

Off-page SEO is shifting toward verification and brand. Backlinks still play a role, but more as a source of reputation verification alongside reviews and mentions. And when AI offers three options, the brand is the deciding factor. In Indigo’s study, brand recognition was the second-strongest factor in trust, and 26% of people overruled the AI’s ranking based solely on the brand. In other words, brand identity is more important today than user identity.

Where this is all heading: agent-driven shopping

Today, AI gives us a list, and we confirm the selection. The next step is agent-driven shopping, where AI selects and purchases on our behalf. At that point, the deciding factor shifts. Today, range and price—the logic of large platforms and aggregators—hold the upper hand. When AI selects and purchases, the depth and specificity of the information it can gather about a product will carry more weight. We’re shifting from being “available at a price” to being “chosen for something.” Those who invest in Information Gain and in their brand today aren’t just building a better ranking for today’s interfaces. They’re securing a ticket to a layer that’s only just emerging.

What to do about it: first steps

  1. Find out if the AI recommends you at all. Take 20 to 30 real search queries in your category and run them through Google AI Mode, ChatGPT, Perplexity, and Gemini. Note whether you appear, in what position, how specifically the AI describes you, and who appears instead of you.
  2. Reevaluate your KPIs. Organic traffic will decline even if your rankings remain stable. Track citations in AI responses, the share of brand mentions, and the conversion rate instead of just the raw number of visits.
  3. Invest in content that AI has nowhere else to get. Your own data, tests, precise specifications, and comparisons. AI already knows generic content. It will cite those who bring something extra.
  4. Prepare your website for multiple search intents. Identify the main search intents behind your top queries and give each one its own path on the page.

Slovakia has a time lag compared to the U.S. and the U.K. AI Overviews don’t have as much reach here yet. Based on our observations, you have roughly 12 to 18 months to prepare before it starts to hurt just as much. Those who start now will have a double advantage: readiness and a head start over competitors who are still asleep.

In digital marketing, user identity is becoming less accessible, less reliable, and often less useful. Intent, behavior, context, and current need are better predictors of relevance and conversion. It doesn’t matter who the customer is. What matters is what they want.

We help major brands and ambitious companies prepare for search in the age of AI. If you want to know how your brand is faring in AI-powered search results and what to do about it, reach out to me for a consultation or training session for your team.

Related content

Local SEO: What Is It and Why Does It Matter?

Local SEO: What Is It and Why Does It Matter?

Natália Gibasová 06. 05. 2026
Read the article
Duplicates harm your positions: How do you remove them?

Duplicates harm your positions: How do you remove them?

Andrej Tomič 27. 02. 2020
Read the article
Why brands in 2026 will increasingly turn to performance content – and what you can gain from It

Why brands in 2026 will increasingly turn to performance content – and what you can gain from It

Veronika Kašiarová 30. 06. 2026
Read the article