Chapter 11: Content in the Age of AI

Last updated: November 1, 2026.

For years, content plans started with the same sentence: “People search for this keyword a lot, let’s write a blog post.” That logic produced thousands of “What is X?” and “How to do X” articles, and they were worth having because they brought traffic. Today a large share of those questions get answered on the results page or inside a chat.

The previous two chapters were technical: can bots reach your pages, and can they read and quote them? This chapter opens the book’s fourth part, content and authority, and starts with the most basic question. If AI writes the answer, what should we write on our own sites?

I suggest two ideas for this. Decision Support Content is content that helps a user make a decision. Click Resilience is a page’s ability to still offer something after the user already has the answer. Both start from the same place: the things AI can’t sum up in a single paragraph.

1. Why informational content alone isn’t enough

In Chapter 1 we sorted questions by what the user does after getting the answer. The group hit hardest was informational questions that can be answered in one paragraph. Since then, new studies have measured that effect.

In February 2026, Ahrefs looked at desktop data for 300,000 keywords. On queries that showed an AI Overview, the click-through rate for the top result was about 58% lower than it would have been without one. The effect shrinks further down the page; at position ten it’s around 19% (Ahrefs). Ahrefs notes that this is a correlation and that other search features, such as featured snippets, have been pulling CTR down for years.

A Seer Interactive study published in April 2026 points the same way (Seer Interactive). Across 53 brands and 5.47 million queries, organic CTR on queries with an AI Overview fell from 3.19% in January 2025 to 1.31% in December 2025. There was a partial recovery in January and February 2026, with the rate reaching 2.36% in February. The researchers don’t expect rates to return to pre-AI Overview levels.

These numbers are mostly US data, and results vary a lot by query type. Among our clients at İmza, none saw a serious drop in clicks after AI Overviews launched in Turkey; we saw small declines.

Still, the direction is clear. A page written for a question that can be answered in one paragraph is now more likely to become the source of that paragraph than to earn the click. That doesn’t make informational content worthless. It just isn’t enough as a traffic strategy on its own.

2. Information everyone has already written

Informational content has another problem: most of it repeats itself. The answer to “What is a filling machine?” appears on twenty different sites in almost the same words. When AI writes that answer, it doesn’t need any particular site. Whichever one it picks says the same thing.

Google’s guide to generative AI search makes this distinction directly (Google Search Central). According to the guide, commodity content “is often based on common knowledge, which could originate from anyone.” Strong pages offer an expert or experience-based view that goes beyond common knowledge. The guide recommends writing from what you know about the subject rather than repeating what others have said.

Here’s what that looks like for a filling machine maker. “What is a filling machine?” is common knowledge. The answers to these questions are not:

• At what speed can your machines fill a foaming liquid soap?
• How long does a format change take for each bottle size?
• What problems come up most often in the first month after installation?
• When do you recommend a submerged-nozzle system over a piston filler, and why?

Only the company that builds, installs and services the machine knows these answers. AI can’t learn them anywhere else; it has to learn them from you. Content that can’t be copied, such as original data, real tests and case studies, is the subject of Chapter 12.

3. Decision Support Content

By Decision Support Content I mean content that helps the user make a decision rather than learn a fact. Informational content answers “What is this?” Decision support content answers “Which one is right for my situation, why, and at what cost?”

The same topic, written both ways:

The questions on the left are common knowledge, and AI summarizes them well. The answers on the right depend on the user’s situation: capacity, product viscosity, budget, service expectations. AI answers these questions too, but only if someone has written down those variables and how they relate.

That gives decision support content two payoffs. It raises your chances of being cited and recommended when AI answers decision questions; the query fan-out we covered in Chapter 3 goes looking for exactly these sub-questions. And the user still has a reason to visit your page for details after getting the answer. I’ll come back to that second point in section 6.

4. Types of decision support content

There isn’t one format. Different steps in the user’s decision call for different content:

• Selection guide: a page that answers “which one, in which case?” by showing which model fits a given capacity, product type and budget.
• Comparison: a page that puts two methods, two models or two approaches side by side using the same criteria. The table rules from Chapter 10 are very useful here.
• Cost explainer: even if you can’t publish prices, a page that explains what drives the price and how each choice affects cost.
• Calculator or configurator: a tool where users enter their own numbers and get a result. For example, one that goes from daily output to hourly capacity, and from capacity to the number of nozzles.
• Checklists: “What to ask before you buy,” “What to prepare before installation.”
• Fit and limits: a page that says plainly what your product is suited for and what it isn’t.
• Proof: real installations, customer examples, test videos, references. Chapter 12 covers these in detail.

Fit and limits deserves a closer look. Brands are reluctant to say what their products aren’t suited for. But when AI gives a user a comparative answer, it looks for clear information it can turn into “X works in this case, not in that one.” If you don’t provide it, AI takes it from a forum comment or a competitor’s comparison page. A page that can say “This machine is not suitable for these products” becomes one that both users and AI trust.

5. Writing around the user’s constraints

Remember the question from Chapter 1: “We’re going to fill 3,000 bottles of liquid soap a day. The product foams, and our budget is around €25,000. Which manufacturers with service in Turkey should I look at?” That question has four constraints: capacity, a product property, budget and service.

Decision support content is written around constraints like these. In Chapter 6 I described how prompt research reveals the constraints people include in their questions. Now it’s time to turn that list into content. For each main constraint, your site should answer three questions:

• How does this constraint affect the decision? For a foaming product, which method, which settings, which speed?
• Where does your product sit on this constraint? Which model, which capacity range, which budget range?
• Where does the constraint run out? At what point is a different solution needed?

The trap here is opening a separate page for every combination of constraints. Producing hundreds of pages like “filling machine for 3,000 bottles of foaming soap a day” and “filling machine for 5,000 bottles of foaming shampoo a day” runs into the scaled content abuse risk I mentioned in Chapter 3. It’s better to bring the constraints together in one selection guide, one table or one calculator. Users find their own combination there, and AI reads the relationships it needs to build its answer in one place.

6. Click Resilience

Click Resilience comes down to one question: After getting the answer from AI, does the user still have a reason to visit your page?

It’s easy to test on your own pages. Open a page and imagine AI summarizing it in three sentences. The user has read those three sentences. What else can the page give them?

If the answer is nothing, the page is fragile when it comes to clicks. It may get cited and it may keep your brand visible, but it will struggle to bring traffic. If plenty is left, the page is resilient.

Here’s what a summary usually can’t give a user:

• A calculation for their own case. “How many nozzles do I need for 3,000 bottles a day?” is answered by a calculator that works with the user’s own numbers, not by a general reply.
• Something to see: a video of a foaming product actually being filled, installation photos, the machine’s dimensions and floor plan.
• Something to verify: the spec sheet, certifications, warranty terms and reference list a purchasing team will ask for.
• A full comparison table that won’t fit in a summary.
• Something to do: request a quote, send a sample, book a demo, buy.
• A file to use: a downloadable checklist, a specification template, a technical drawing.
• Reasons to trust you: real customer examples, who the team is, where the service network is.

Most of these are already part of decision support content. The two ideas work together: decision support content gets you into the AI answer, and click resilience brings the user to you after the answer.

Google’s guide points in the same direction. It says AI answers show prominent, clickable links to the pages that support the answer, and that users follow those links to dig deeper. The same guide notes that high-quality images and videos that support your content open additional ways to appear in AI features (Google Search Central).

7. A page that earns the click

Suppose a filling machine maker builds a page called “Choosing a filling machine for foaming products.” A version with high click resilience could include:

1. A short answer at the top: two or three sentences on which method works best for foaming products and why. This part is there for AI to quote, and being quoted is a good thing.
2. A selection table showing which model fits each product type, viscosity and capacity.
3. A capacity calculator where users enter daily output, number of shifts and bottle volume, and see how many nozzles they need.
4. A video showing the same foaming product filled with two different methods, side by side.
5. Which products this method doesn’t suit, and what’s recommended instead.
6. How choices like nozzle count, automation level and materials affect cost.
7. A list of “five things to prepare before requesting a quote,” with the form right below it.

The only part of this page AI can summarize is the first item. Everything else depends on the user’s own situation, on seeing something, or on taking an action. Even after AI tells a user that “a submerged-nozzle system reduces foaming,” they may still come to this page to calculate their capacity, watch the video or request a quote.

Click resilience doesn’t mean hiding the answer. In Chapter 10 I recommended putting the answer first, and that still holds. Give the answer, then offer everything beyond it. You don’t have to choose between being quotable and being worth a click; a good page does both.

8. Being cited still pays

The same Seer Interactive study shows what citations are worth while CTR falls. Average organic CTR on informational queries across 2025 looked like this (Seer Interactive):

• When the brand was cited in the AI Overview: 2.07%.
• When the brand wasn’t cited: 0.94%.
• On queries without an AI Overview: 3.35%.

A cited brand gets more than twice the CTR of one that isn’t cited, though it still trails the no-AI Overview baseline. The researchers note this is a correlation, and that stronger brands are more likely to be cited in the first place.

So on a question that triggers an AI Overview, these are your options: get cited and earn fewer but still meaningful clicks, or don’t get cited and earn far fewer. Going back to a results page with no AI Overview isn’t one of them.

Citations pay off beyond clicks, too. In Chapter 3 we saw that being mentioned and being recommended in an answer each have their own value, and in Chapter 8 we looked at how AI’s influence comes back through indirect paths like branded search and what customers tell you themselves. The more often your informational content is cited, the more likely users are to remember your name when they reach the decision stage.

9. Reviewing the content you already have

Before producing new content, look at what you have. Most sites have hundreds of informational articles piled up over the years. Deleting all of them would be a mistake, and so would leaving all of them as they are.

You can start by sorting your content into three groups:

For pages in the first group, the approach we took at Filmler.com is a good example. Out of roughly 450,000 cast and crew records, we opened only the pages that had content to indexing, not all of them. The same logic applies to informational articles: one strong page beats five short ones that repeat each other, both for readers and for AI.

Make this call based on data. In Chapter 1 I suggested splitting Search Console clicks by page type, and in Chapter 8 we covered Search Console’s AI report and the AI channel in Analytics. Put the three side by side and you can see which page group is losing clicks, which one shows up in AI, and which one brings in business.

10. Moves that backfire

Some of the ways people adjust their content for AI search don’t work, and some do real damage.

Hiding the answer is the first. Burying it at the bottom of the page or writing long intros to keep users on the page wears readers out. AI just goes to another source that states the answer clearly.

Using AI to churn out hundreds of generic informational articles is exactly the commodity content Google describes, and it carries a scaled content abuse risk. Chapter 19 is devoted to this.

Gating the spec sheet, pricing information or selection guide behind an email form also backfires. Neither search engines nor AI can read what’s behind a form, so they learn the same information from somewhere else. A form should be the next step for a user who has already decided, not a lock on the information they need to decide.

Dropping informational content entirely is a mistake too. Basic definitions and explanations are still needed. Decision support pages build on them, and citations of your informational pages keep your brand visible. The problem is treating informational content as the whole strategy.

Finally, don’t make decisions based on traffic alone. As I said in Chapter 8, traffic is the means and business results are the goal. A guide whose traffic is falling may still be getting cited and bringing in sales.

Before you write the page

AI summarizes common knowledge. The information that makes a decision easier, the kind only you have, it has to get from you. When planning a new page, two questions make a good filter: Does this page say something nobody else could write? And what will the user find here after reading the summary?

The first question, information no one else can provide, is the subject of the next chapter: original data, real tests, case studies and expert opinion.

Checklist

Sources

• Ahrefs: Update, AI Overviews reduce clicks by 58% (February 4, 2026)
• Seer Interactive: AIO impact on Google CTR, 2026 update (April 24, 2026)
• Google Search Central: Optimizing for generative AI search (last updated July 10, 2026)

Previous: Chapter 10, Readable and Citable Pages · Table of contents · Next: Chapter 12, Content That Can’t Be Copied