AI Search

Direct Answers Win AI Citations: 193,507 Answers Studied

A TJ Digital study published on October 28 analysed 193,507 AI answers from ChatGPT, Google AI Mode and AI Overviews, and found that pages which open with a direct answer get cited far more often than top-ten pages that do not. The gap is real, but the study's own author calls it a correlation, not a cause.

How the study was done

TJ Digital, an SEO and generative engine optimisation agency, analysed 193,507 answers collected weekly from August 27 to October 21, 2026, for 78 client businesses, tracked through the Peec AI platform. The answers came from three sources: ChatGPT, Google AI Mode and Google AI Overviews. The researchers then used AI raters in blind comparisons, so the raters were not told which page in a pair had been cited, to judge whether pages opened with a direct answer to the question. They compared cited pages against two different groups: pages the same answer had retrieved but did not cite, and pages from Google's top 10 that the AI never pulled up at all. The study also measured other page traits and their odds of being cited, adjusted for each site's Ahrefs Domain Rating and page type, across 21,874 cited and 4,507 passed-over pages for 434 questions.

The direct-answer gap

The headline result comes from the smaller sets of paired pages. In blind ratings of 195 page pairs, 20% of cited pages opened with a direct answer, against 5% of Google top-10 pages the AI never pulled up. In the earlier round, comparing cited pages with pages the same answer retrieved but did not cite, the split was 23% to 14%. That second comparison matters because it is the stricter test: both groups of pages were found by the AI, and only one group got cited. Even so, the gap is modest, and the study is honest about that. Three out of four cited pages in the best group still open without a direct answer, so most cited pages do fine without one. The wider numbers add context: ChatGPT cites about 27% of the pages it finds, while AI Overviews and AI Mode cite 85 to 92%. Almost a quarter of the cited pages did not answer the buyer's question at all in the part the graders looked at, so a link does not guarantee an answer.

What else separates cited pages

Direct answers are not the only trait that tracks with citations. The study reports that cited pages more often had the year in the title, carrying more than twice the odds of being cited, and a recent publish or update date, with a date within the last 12 months giving 88% higher odds. Tables carried 69% higher odds, first-hand experience 45%, and prices 28%. These are associations, not proof that any single trait causes a citation, and the study says so plainly. TJ Robertson, the analyst behind the research, calls the direct-answer finding a correlation rather than a cause, and the agency itself sells SEO and GEO services, which is a conflict worth keeping in mind when reading vendor-published research. For the 23% versus 14% comparison, neither the number of pages nor the grader model had been disclosed as of October 28, so that part of the study is harder to check from the outside.

What to do with this

None of that makes the finding useless. It points in the same direction as most AI search advice: put the answer first, then the explanation. A page that opens with a clear, direct answer to the question is easier for an AI to quote, and the blind ratings suggest the models do notice. But the numbers also argue against rewriting a whole site around one trait. A 20% versus 5% gap across 195 pairs is a signal, not a law, and the fact that most cited pages do not open with a direct answer means the rest of the page still does most of the work. Freshness looks like the stronger lever in this dataset: a recent date and a year in the title both moved the odds more than the direct answer did, and tables and first-hand experience helped too. The sensible reading is to treat the direct answer as a cheap, low-risk improvement to make on pages you already want cited, alongside keeping dates current and adding tables where they fit, rather than as a rewrite programme.

Why it matters for SEOs and link builders

For SEOs, this study is one of the largest looks yet at what actually separates a cited page from a passed-over one, and its most useful message is the modest one: small, concrete page traits move the odds, but no single trait decides anything. Chasing citations means doing the unglamorous work of answering the question early, keeping pages fresh, and showing real experience, not gaming a single signal. For link builders, the citation angle is becoming as important as the ranking angle. A link from a page that AI answers cite carries visibility that a link from a page nobody quotes does not, so when you earn a link, the citing potential of the linking page is part of its value. Prioritise placements on pages that answer their question directly, stay current, and use tables and real examples, because those are the pages this study found the AI actually pulling into its answers.

Sources

Sources: TJ Digital, October 28, 2026.

Related reading on this site: AI Overviews Appear on 70% of Everyday UK Searches, Sayu Study Finds, Google Tests an AI Overview Local Pack Without Buttons, Reviews or Photos, Gemini 4 Carbon Arrives in AI Mode for Google AI Ultra Subscribers, Google's Search Volume Is About 5x ChatGPT's, OpenAI Says, Google: Technical SEO Remains Key for AI SEO, and SEO Split Testing in 2027: How to Prove a Change Actually Worked.

More from Linkbaits

Turn this playbook into links

Linkbaits helps you create content that earns citations and backlinks. Start with the toolkit and build your first linkable asset this week.

Start free trial