In the previous article, I described why website click-through rates are declining for many sites, even though demand for their products or services isn't necessarily falling.

The short version: Part of the research process is shifting into search and AI systems. Users compare fewer websites, read fewer reviews, and spend less time manually evaluating alternatives. The actual pre-decision increasingly happens where information is aggregated — not where it was originally published.

Whoever understands this development soon arrives at an obvious question:

Why does an AI actually recommend one provider — and not another?

The answer is surprising. Because it has less to do with visibility than many believe. And more to do with a skill that was important long before the age of AI: The ability to make differences visible.

Most Websites Provide Information — But Not Orientation

When you look at typical corporate websites, you usually find a lot of information. You learn:

  • what services are offered
  • what products exist
  • how long the company has been on the market
  • what customers are served
  • how to get in touch

All of this is useful. But surprisingly often, something else is missing: orientation.

A potential customer doesn't just want to know what a company does. They want to know:

  • Who is this suitable for?
  • When should I choose this provider?
  • When should I not?
  • What makes it different from others?

These are exactly the questions that form the basis of every recommendation. And exactly the information many websites lack.

Why Recommendations Need Tension

In an earlier article, I argued that strategy ultimately lives on tension. Strategy doesn't emerge from wanting to be everything. It emerges from choosing. Every choice creates a contrast. Every focus excludes alternatives. And that's exactly how orientation emerges.

This applies not only to people. It surprisingly also applies to AI systems. Because a recommendation is ultimately also a form of selection.

Let's take two statements.

The first reads:

We support companies in digital transformation.

The second reads:

We support mid-sized manufacturers in implementing AI in development and quality processes.

Both statements describe a company. But only the second creates tension. It contains differences.

  • mid-sized manufacturers instead of all companies
  • mid-market instead of every organization size
  • AI instead of digitalization in general
  • development and quality instead of arbitrary processes

Suddenly a corridor emerges. And it's precisely within this corridor that a recommendation can take shape.

The Real Task of Humans

Many discussions about AI implicitly assume that AI takes on the actual value creation. In reality, the work is often distributed differently. AI excels at:

  • recognizing patterns
  • aggregating information
  • finding commonalities
  • formulating justifications

It's significantly worse at:

  • setting priorities
  • creating tension
  • taking positions
  • defining relevant differences

This is exactly where humans come in. Humans define the corridor. AI moves within it. Humans decide which target audience matters. AI can derive a recommendation from that. Humans define the focus. AI can explain it. Humans create the tension. AI makes it visible.

That's why positioning becomes even more important, not less important, in the age of AI.

The Five Questions AI Systems Look for Answers To

When you look closely at AI recommendations, you notice they almost always use the same kinds of distinctions.

1. Who is this intended for?

A recommendation needs a target audience. The clearer it's described, the easier the categorization.

2. What problem does it solve?

Users think in problems. Companies often think in services. AI needs the connection between the two.

3. When is this the right choice?

Recommendations always happen in a context. AI needs to understand in what context a provider is particularly suitable.

4. When is this not the right choice?

Boundaries also create orientation. Those who describe when they're not the best choice often provide more valuable information than those who want to be suitable for everything.

5. What makes this offering different?

Recommendations emerge from differences. When all providers are described the same way, AI can hardly prioritize meaningfully.

Why Case Studies Suddenly Become More Important

An interesting consequence concerns case studies. Many companies view them as marketing material. For AI systems, they serve a different function. They make differences visible.

A good case study automatically answers questions like:

  • Who was the customer?
  • What problem existed?
  • Why was this solution chosen?
  • Why not another?
  • What result was achieved?

This creates context. And context is ultimately nothing other than structured tension.

Why Third-Party Sources Become More Important

AI doesn't only learn from your website. It also learns from what others write about you. This includes:

  • trade publications
  • industry blogs
  • conference programs
  • podcasts
  • guest contributions
  • review platforms

Particularly relevant is something that could be called consistency. When different sources independently highlight the same differences, a stable picture emerges. AI then recognizes not just that a company exists. It recognizes what it stands for.

The Real Question Has Changed

Many discussions about AI visibility focus on the question:

Will I be mentioned?

The more important question is:

Why will I be mentioned?

That's a fundamental difference. Because visibility alone doesn't create a recommendation. Recommendations only emerge when enough differences exist to justify a choice. That's why the most important task for many companies in the coming years won't be publishing more information. There's already enough information. The real task is making better distinctions visible. Because that's where positions emerge. That's where orientation emerges. And that's exactly where the recommendations come from that AI systems later make.

A Simple Self-Check

There's a surprisingly simple way to find out whether your own website already provides enough material for recommendations. Ask yourself:

Could an AI use our content to convincingly explain why we're the right choice — and why not some other provider?

If there's no clear answer to that, the problem is probably not visibility. It's tension. Because language models are remarkably good at recognizing and explaining existing differences. They're much worse at creating those differences themselves. That's exactly why positioning remains a human task even in the age of AI.