Brands get into ChatGPT answers when retrieval systems select their pages, or outside pages about them, as sources for a buyer's question. Seven methods have support in 5W's data: matching question wording, giving each service its own page, publishing original research, earning outside coverage, building community presence, measuring each AI platform and updating pages. The figures come from 5W's analysis using Profound's tools, and the limits are stated below.
Last updated: November 1, 2026. Data window: September 27 to October 26, 2026, with full-period figures from January 27, 2026.
Summary in six points
- ChatGPT builds answers from what the model learned in training and, when it searches the web, from the pages retrieved for the question.
- A study by researchers at Princeton University, IIT Delhi and other institutions found that adding citations, quotations and statistics raised source visibility by more than 40% in its benchmark.
- In 5W's data, dedicated pages that match a buyer's wording earned the most citations, and practice pages alone earned 54.6% of 5wpr.com's total.
- Earned media and social platforms supplied a larger share of citations than 5wpr.com itself on the tracked prompts.
- Each AI platform cites different sources, so results have to be measured platform by platform.
- The 5W figures are a company case study, and 5W chose every prompt.
How do ChatGPT answers choose which brands to name?
ChatGPT names brands from two places: what the model learned in training and, when it searches the web, the pages retrieved for the question. Researchers at Princeton University, IIT Delhi and other institutions described this design in a paper presented at the ACM SIGKDD Conference in 2024. A generative engine, they wrote, pairs a language model with a search engine that retrieves documents, then writes a response that cites them.
The same paper tested what makes a source more visible in that response. In a benchmark of 10,000 queries, adding citations, quotations from relevant sources and statistics raised source visibility by more than 40% across queries, and gains reached up to 37% on Perplexity. The effect varied by topic area, so the authors call for domain-specific methods. The full paper is available as GEO: Generative Engine Optimization on arXiv.
Being named in the answer matters because readers click through less. Pew Research Center analyzed the browsing of 900 U.S. adults in March 2025. Users who saw a Google AI summary clicked a traditional result link in 8% of visits, against 15% for users who did not see one, and they clicked a link inside the summary in 1% of visits. Pew studied Google AI summaries only, so the figures do not describe ChatGPT directly, but they show where the exposure moves when an answer is written out. Pew's full analysis is on its page about how AI summaries change clicking.
The practical reading is simple. A page has to be retrieved, and then it has to be used in the written answer. The next six sections cover what 5W's data says about each step, and generative engine optimization is the name for the work.
Why does question wording decide which pages get cited?
Pages that match the wording of a buyer's question get cited because retrieval systems select pages whose text answers that question. In 5W's data, 5wpr.com was cited on all 111 tracked prompts that begin with Which, Where or Who, on 40 of 43 prompts about case studies and on 21 of 22 prompts about hiring.
The gap shows the same thing from the other side. Prompts that begin with How ask for a method instead of a firm, and 5wpr.com was cited on 80 of 123 of them (65%). AI answers to those prompts may cite how-to guides instead of agency pages.
One caution applies to these numbers. Of the 396 tracked prompts, 139 used wording that matches how 5W describes itself, and 5wpr.com was cited on 93% of those against 76% of the others. Part of the wording effect is built into the prompt set, and a list written by an outside party would test it more fairly.
A brand can apply the method in four steps:
- List the questions buyers type for each service, in the words buyers use.
- Publish one page for each question.
- Answer the question in the first 40 to 80 words of the page, using the question's own wording.
- Repeat the subject's name in the first sentence under each heading, so any paragraph can be quoted on its own.
Does each service and industry need its own page?
Dedicated pages earned most of the citations in 5W's data. Between January 27 and October 26, 2026, 225 practice pages earned 58,605 of the 107,325 citations that AI platforms gave to 5wpr.com, or 54.6%.
| Site section | Citations | Share |
|---|---|---|
| Practice pages (225) | 58,605 | 54.6% |
| Homepage | 27,016 | 25.2% |
| Blog | 13,270 | 12.4% |
| Services pages | 5,242 | 4.9% |
This is an association and not proof of cause. It shows where citations landed. A likely reason is that a page about one service gives retrieval one specific question to match, while a post that covers five services at once matches none of them well.
Why does original research earn AI citations?
Original research earns citations because it holds data that exists on no other page. 5wpr.com's 94 research and AI Visibility Index pages earned 1,498 citations, and 832 of them (about 56%) arrived in the last 30 days of the data window.
This fits the Princeton and IIT Delhi finding that statistics and quotations from relevant sources raised source visibility. A brand can publish first-party numbers, such as a survey, an audit or a count of its own results. A citable research page contains:
- The headline number in the title or the first paragraph.
- The dates the data was collected and the number of items counted.
- A plain statement of what the data cannot show.
- A downloadable table, so other writers can check and cite the figures.
Which outside sources do AI answers cite most often?
AI answers cite outside sources more often than they cite a brand's own website. On the tracked PR prompts in the last 30 days, earned media sources earned 10.4% of citations and social and community platforms earned 8.0%, against 3.6% for 5wpr.com.
| Source type | Share of citations |
|---|---|
| Earned media sources | 10.4% |
| Social and community platforms | 8.0% |
| 5wpr.com (one site) | 3.6% |
| Institutions | 3.3% |
| PR wires | 1.1% |
Earned media and social platforms are categories made of many sites, while 5wpr.com is one site, so the table shows where citations come from and not which single page matters most.
The choice of publication also matters. On the same prompts, O'Dwyer's ranked eighth among all cited sources, PRovoke Media seventeenth, and PRWeek fell outside the top 25. In the last 7 days of the window, YouTube, LinkedIn and Reddit were all among the ten most-cited sources. A separate post on this site maps which websites AI answers cite most.
How should a brand measure results on each AI platform?
A brand should measure results on each AI platform separately, because each platform selects different sources for the same question. In the last 30 days, 5wpr.com ranked first by citation share on Google AI Overviews and ChatGPT but sixth on Perplexity.
| Platform | Citation share | Rank |
|---|---|---|
| Google AI Overviews | 5.18% | 1 |
| ChatGPT | 3.91% | 1 |
| Google Gemini | 4.62% | 2 |
| Google AI Mode | 4.17% | 3 |
| Microsoft Copilot | 2.28% | 4 |
| Perplexity | 1.47% | 6 |
Platform differences also show up in which brands get named. On Microsoft Copilot, Edelman had higher brand visibility than 5W (28.3% against 22.4%), while 5W led Edelman on the other five platforms. Brand visibility measures how often a brand appears in AI answers to the tracked prompts.
A brand can start by running ten buyer questions on three platforms and recording which sources and brands each answer names. Tracking those citations across a fixed set of questions is the basis of citation share, and a separate post explains how AI engines decide what to cite.
How often should pages be updated?
Pages should be reviewed monthly, because citation momentum differs sharply from page to page. The Consumer PR page earned 52% of its lifetime citations in the last 30 days of the window, while the Case Studies hub earned 6% of its lifetime total in the same period.
These figures are associations and not proof of cause. They do show that some pages gain citations while others fade, so a monthly check of which pages gained and lost tells a brand what to fix first. A visible "Last updated" date on each page tells readers and reviewers when the facts were last confirmed.
What should a brand do first?
A brand should first find out which questions its buyers ask and whether any page on its site answers each one. Services that carry out the seven methods are listed on 5W's generative engine optimization services page, and 5W's full chatbox report holds every table behind this article. The work also overlaps with GEO for PR teams and with AI communications.
This week, write the ten questions a buyer asks about your service, run them on ChatGPT, Gemini and Perplexity, and record which pages and brands each answer names. The first question that returns a competitor and no page of yours tells you which page to write.
Video: Getting a brand recommended by ChatGPT
A citation means an answer used a page as a source; it does not establish that ChatGPT recommends the company. The 5W analysis does not measure recommendations.
How was the 5W data collected, and what are its limits?
The 5W figures come from 5W's analysis using Profound's AI visibility tools for September 27 to October 26, 2026. 5W produced the analysis and uses Profound as a customer, so this is a company case study and not independent validation, and it is not a Profound study.
Method and disclosure
5W selected all 396 prompts, and no outside party reviewed them. A total of 190 prompts began January 27, 2026 in the US, UK and Canada on ChatGPT, Copilot, Perplexity and Google AI Overviews. The other 206 began September 14 in the US only, on those four platforms plus Gemini and Google AI Mode. The batches differ in topics, regions, platforms and run length, so totals should not be compared across them.
A citation shows that an AI answer used a page as a source and linked to it. A mention means the answer names a brand. A recommendation means the answer presents the brand as a good fit, and this analysis has no recommendation data. The prompts cover PR agency questions, so the source mix would differ for other industries.
The Princeton and IIT Delhi findings come from the paper by Pranjal Aggarwal and co-authors, published at the ACM SIGKDD Conference in August 2024. The Pew figures come from Pew Research Center's analysis published July 22, 2025.
Frequently asked questions
How do brands get into ChatGPT answers?
Brands get into ChatGPT answers by publishing pages that match buyers' questions, giving each service its own page, adding original research, earning outside coverage, building community presence, measuring each platform and updating pages. Each method is covered with data above.
Does success on one AI platform carry over to others?
No. In 5W's data, 5wpr.com ranked first by citation share on ChatGPT and sixth on Perplexity in the same 30 days, so each platform has to be measured on its own.
Do statistics and quotations really help?
In one benchmark they did. The Princeton and IIT Delhi researchers found that adding citations, quotations and statistics raised source visibility by more than 40% across queries. The result varied by topic and came from an experiment, so it is a tendency and not a promise.
What is the difference between a citation and a mention?
A citation means an AI answer used a page as a source and linked to it. A mention means the answer names the brand. A brand can be mentioned without being cited, and cited without being mentioned.
Does a citation mean ChatGPT recommends a company?
No. A citation shows that an answer used a page. A recommendation means the answer presents the company as a good fit, and the 5W analysis does not measure that.
Is the 5W data independent?
No. 5W produced the analysis, chose all 396 prompts and uses Profound's tools as a customer. No outside party reviewed the prompt list, and results for any other brand will vary by category, competitors and platform.