The buyer is already asking for the shortlist
A customer who used to search “best accounting software for a small business” may now ask an AI tool: “I run a five-person service company, I need invoicing and payroll, and I do not want a complicated system. Which of these three tools fits best?”
That is a different job for your content. The customer is not looking for a general explanation of accounting software. They are asking the system to evaluate alternatives against a situation.
This is why comparison pages are becoming more valuable in conversational search. They contain the raw material an answer engine needs: named options, decision criteria, trade-offs, use cases, limitations and a conclusion. A strong page can help the buyer make a choice even when the customer never visits it.
That last point matters. Pew Research Center found that people clicked a traditional result on 8% of visits when a Google AI summary appeared, compared with 15% of visits without one. Only 1% of visits with an AI summary included a click on a cited link inside the summary. (pewresearch.org)
The commercial implication is uncomfortable: a page may influence the decision without receiving the session. That makes visibility, citation and branded demand worth tracking alongside organic clicks. It does not make every comparison page a good investment.
Comparison pages win the decision job. Blog posts win the learning job.
| Criterion | Comparison page | General blog post |
|---|---|---|
| Matches a buyer weighing named options | Strong. It gives the system explicit alternatives and criteria. (better) | Weak unless the post naturally includes products or approaches. |
| Works for early research | Moderate. Useful when the reader already knows the category. | Strong. It can explain the problem before a shortlist exists. (better) |
| Easy for an AI system to summarize | Strong when claims, headings and evidence are clear. (better) | Variable. The answer may be buried in a long narrative. |
| Supports a direct commercial next step | Strong. The reader can move from trade-off to enquiry or purchase. (better) | Moderate. It usually needs an internal link to a service or product page. |
| Builds topical authority over time | Moderate. Fewer pages can cover fewer angles deeply. | Strong when the posts answer distinct customer questions. (better) |
| Risk of becoming thin or self-serving | High if it simply declares your offer the winner. | Moderate if it is generic or rewritten from other sources. (better) |
The evidence supports a shift in format, not a content purge
Google’s own guidance does not say that comparison pages receive special treatment in AI search. Its current guidance says the same foundational SEO practices apply to AI Overviews and AI Mode: pages must be crawlable, indexable, useful and eligible to appear in ordinary search. Google also says there are no additional technical requirements or special optimizations for inclusion. (developers.google.com)
That is the first correction to the popular advice. You do not need a new “AI page type.” You need pages that make the important answer easy to find, understand and verify.
Google’s review guidance does, however, describe several qualities that naturally favour good comparison pages. It recommends evaluating from the user’s perspective, explaining what separates alternatives, covering the important decision factors, discussing benefits and drawbacks, using quantitative measurements where appropriate and supporting recommendations with first-hand evidence. It also says ranked lists should contain enough useful material to stand on their own. (developers.google.com)
Those are not tricks for getting quoted by a chatbot. They are the basic ingredients of a page a human can trust. Conversational search increases their value because a model needs to assemble an answer from explicit, well-supported claims. A vague post about “choosing the right solution” gives it very little to work with.
The practical position is simple: comparison pages deserve a larger share of effort when the customer is choosing among known options. Blog posts still matter when the customer is learning what the problem is, whether it matters and what vocabulary to use.
What the current numbers actually tell us
Pew Research Center, “Google users are less likely to click on links when an AI summary appears in the results.”
Pew Research Center, same study.
Pew Research Center, same study.
Ahrefs, “Insights From 55.8M AI Overviews Across 590M Searches.”
Ahrefs, “AI Visitors Visit Fewer Pages and Bounce More Often Than Traditional Search Visitors.”
Ahrefs, same traffic-quality study. This is a study finding from participating traffic data, not a universal benchmark.
Practitioners are seeing the strongest use case in commercial questions
The practitioner reports point in the same direction, with an important qualification: most are vendor studies, case studies or analyses of their own clients. They are useful for spotting patterns, but they are not controlled evidence that one page format causes better revenue.
Semrush’s comparison of Google AI Mode, traditional Google results and other large language model search experiences analysed 5,000 keywords and 150,000 citations. Its report found that AI systems often cite deeper pages, including specific blog posts and help articles, when the query is nuanced. It also concluded that ranking highly in traditional search does not guarantee the same page will be cited in AI Mode. (semrush.com)
That finding argues against replacing blog posts with comparison pages. A detailed article may contain the one explanation, example or limitation that answers a follow-up question. In conversational search, the relevant unit is often not “the best page for the keyword.” It is the best passage for the next question.
HubSpot reports that it began creating software comparison articles and industry-specific “best” pages as part of its effort to improve visibility across ChatGPT, Gemini and Perplexity. The company describes these pages as part of a wider programme that also included measurement and optimisation. That is a practitioner report, not independent proof, but it reflects how larger teams are allocating effort: toward pages that help buyers compare vendors during evaluation. (blog.hubspot.com)
The most defensible conclusion is narrower than the hype. Comparison pages are especially useful for “which should I choose?” prompts. They are not automatically better for “how does this work?”, “what should I know?” or “how do I solve this?” prompts.
A comparison page only earns its place when it contains evidence
Many comparison pages fail because they are sales pages wearing a table as a costume. They list features, repeat vendor copy and choose the company that published the page. An AI system may still extract the product names, but the page gives a careful buyer no reason to trust the judgement.
A useful comparison needs a point of view and a basis for that point of view. For a local web agency, that might mean comparing a managed website service with a do-it-yourself platform across editing difficulty, response time, accessibility responsibility, local search support and total operating effort. The page should say who benefits from each option, where each option becomes a poor fit and what evidence supports those claims.
The evidence can be modest. Show the workflow you tested. Explain how pricing is calculated. Quote the service terms. Include screenshots from the actual interface. Use customer questions to identify the criteria that matter. If you have not used a competing product, say that the comparison is based on public documentation rather than first-hand testing.
This is also where your existing content should work harder. A comparison page can link to When Google Finishes the Search, Measure the Customer Action when explaining why a lower-click search result may still produce calls or enquiries. It can also point readers to Small Businesses Should Simplify Google Ads Around Intent, Not Abandon Keywords for AI Max when the choice involves paid search intent rather than only website features.
Google’s broader people-first guidance asks whether content provides original information, substantial value beyond other pages and clear evidence of expertise. A thin comparison page fails that test even if its headings are perfectly structured. (developers.google.com)
Build the page around the decision a customer is trying to make
-
Start with the buyer’s situation
Write the page for a defined use case, such as a five-person service business, a regulated clinic or a retailer with a small internal team.
-
Name the realistic alternatives
Include the options customers actually encounter, including the option to do nothing or keep the current system.
-
Choose criteria from customer friction
Use sales calls, support questions, reviews and lost-deal notes instead of inventing a feature checklist.
-
State the trade-off before the table
Tell readers what each option is good at and where it creates work, cost or risk.
-
Add evidence to important claims
Use tests, documentation, screenshots, pricing records, customer examples or clearly labelled public sources.
-
Give a conditional recommendation
“Choose A if…” is more credible and more useful than declaring one universal winner.
-
Create follow-up paths
Link to the detailed service, product, implementation and measurement pages that answer the next question.
The better content plan is a small decision library
For most small businesses, the answer is not to publish a dozen competitor pages or stop writing educational articles. Build a small decision library around the questions that lead to revenue.
Start with one comparison page for the most common buying decision. Then create the supporting pages that make it trustworthy: a cost explanation, an implementation guide, a page about who should not buy, a real customer example and a practical answer to the main problem the product solves. Those supporting pages may be blog posts, service pages or documentation. The format should follow the question.
This structure also protects you from over-reading AI search data. Ahrefs found that AI visitors in its study viewed fewer pages and bounced more often than traditional search visitors, while reporting a much higher conversion rate for those visitors. That suggests AI-referred traffic can be smaller but more commercially qualified, although the result should not be treated as a promise for every business. (ahrefs.com)
Measure the library by customer action, not just sessions. Track enquiries, calls, booked consultations, assisted conversions, branded searches and the questions salespeople say prospects already understand. If a comparison page earns few visits but helps prospects arrive with a clearer shortlist, it may be doing valuable work that last-click analytics will understate.
Comparison pages are becoming more valuable because conversational search has moved more of the customer’s evaluation into the search experience itself. They are not replacing blog posts. They are taking the lead at the point where a customer asks an answer engine to make sense of competing choices. Build for that decision, support it with evidence and keep the educational content that helps the buyer reach the shortlist in the first place.
