The Pages AI Cites for Category Queries Are Pages You Didn't Write

There are two types of AI search queries that behave completely differently, and most AEO advice only covers one of them.

Informational queries -- "what is [business name]," "how does [service] work," "what does a [job title] do" -- pull from brand pages, educational content, and direct entity sources. Optimize your page, structure your content well, and you can appear here.

Category recommendation queries -- "best [service type] in [city]," "what [tool category] should I use," "which [business type] do you recommend" -- pull from somewhere else entirely. They pull from the comparison articles, directory category pages, and industry roundups that have already evaluated options in your space. Your own page doesn't enter the picture unless you're already listed in those sources.

For businesses running AEO campaigns that focus on page optimization, this distinction is the gap between effort and results.

What Perplexity Actually Read in a Live Audit

In our May 2026 source URL analysis of a Sourcepull audit (Scout session 16, 2026-05-13, platform-citation-behaviors.md), we mapped the exact domains Perplexity cited when answering four category recommendation queries about AEO audit tools.

This is not modeled behavior or aggregate data. These are the URLs Perplexity's retrieval layer actually read to generate each recommendation response.

| Domain | Appeared in (of 4 queries) | |--------|---------------------------| | tryprofound.com | 4 of 4 | | meltwater.com | 4 of 4 | | conductor.com | 3 of 4 | | g2.com (AEO category page) | 3 of 4 | | aeoaudittool.com | 3 of 4 | | aiclicks.io | 2 of 4 | | hubspot.com | 2 of 4 | | stackmatix.com | 2 of 4 |

None of these are businesses being recommended. They are comparison articles, software review category pages, and industry blogs that have evaluated and listed AEO tools. Perplexity synthesized its recommendations from those sources -- not from the businesses' own pages.

A business not listed in those eight domains has essentially no path to appearing in Perplexity's category recommendations for this space, regardless of how well their own site is built.

Why This Is Structural, Not a Quirk

Perplexity runs live web retrieval for most queries. When someone asks for a recommendation, it's effectively asking: "which sources have already done the evaluative work on this category?" The answer is comparison articles, directory category pages, and expert roundups -- not the businesses themselves. Those sources have ranked and evaluated options. Perplexity reads their conclusions.

The same pattern holds across platforms. The Yext 2026 AI visibility study analyzed 17.2 million citations across ChatGPT, Perplexity, and Gemini (Scout's platform citation architecture investigation, 2026-05-12). ChatGPT draws 49% of citations from third-party directories and reference sources rather than brand-owned pages. Perplexity explicitly trusts niche expert directories for its category.

Both platforms, by different mechanisms, arrive at the same starting point: for recommendation queries, the sources they trust are sources that exist to evaluate and list businesses -- not the businesses' own self-description.

The Format of What Gets Cited

The format of comparison pages is not incidental. Practitioner data reported across multiple 2026 analyses (Scout's content format investigation, 2026-06-18, content-format-citation-signals.md) suggests that pages with three or more comparison tables earn roughly 25% more citations than structurally equivalent pages without tables. The mechanism is extractability: tables are parsed as discrete comparison grids that AI systems can read cell by cell, without needing the surrounding prose context.

Listicle-format content -- numbered or bulleted lists of recommended options -- is estimated to account for approximately 22% of AI citations across platforms, disproportionate to its share of total web content. These are unverified practitioner figures, not controlled study data. But the direction is consistent: the format AI trusts for recommendation queries is the format of comparison articles.

A typical business service page -- prose descriptions, FAQs, case studies -- is not this format. The comparison articles AI reads look like ranked lists with evaluation criteria. That's the format gap.

What This Means for the Standard AEO Checklist

The standard AEO playbook addresses page-level signals: schema markup, FAQ sections, service page content structure, E-E-A-T signals. Those optimizations matter -- but for a different query type.

A business owner who adds LocalBusiness schema, restructures their service page, and writes tight FAQ answers has improved their performance on direct brand queries and informational queries. They have done nothing for their category recommendation visibility, because that visibility is determined by which comparison sources list them, not by what their own page says.

For businesses whose primary gap is in category recommendation queries -- "which [business type] do you recommend in [city]" -- the fix is not page optimization. It's being present in the comparison and directory sources that AI reads for that query type.

The two problems require different work. Without knowing which query type is driving your visibility gap, it's easy to spend months on the wrong one.

Distribution Work, Not Content Work

Appearing in the comparison sources AI platforms read is a different kind of intervention than SEO or content optimization. It is distribution work: getting your business listed and evaluated in the external sources that AI treats as authoritative for your category.

For the AEO tool category in our audit, the high-priority actions would be: getting listed on the G2 category page for AEO tools (cited in 3 of 4 queries), appearing in aeoaudittool.com comparisons (3 of 4), and earning coverage from the blogs Perplexity treats as expert sources in this space. For a local service business, the equivalent is appearing in the local comparison directories and category review sites -- not just individual listing pages, but the aggregated category pages that AI reads when answering recommendation queries.

The specific comparison sources vary by business category, platform, and geography. What stays consistent is the pattern: for recommendation queries, optimizing your own page is secondary to appearing in the pages AI already reads.

Finding Your Comparison Gap

A Signal Check runs live queries across platforms for your business type and captures which domains appear in the source citations. That source data shows specifically which comparison pages and directories are driving recommendations in your category -- and whether you're in them.

If you're invisible on category recommendation queries, the fix is probably not on your website.