272 of the 316 pages ChatGPT cited on specific questions were read, all from small providers' own sites. The median page had 928.5 visible words (n=272), 19% had a table and 5% showed an updated or published date. This is descriptive only, with no control group of uncited pages.
Pages read
272 of 316
ChatGPT specific cited URLs that were fetched OK.
Median visible words
928.5
Median across 272 pages. Google specific pages: 1779.5. ChatGPT general pages: 1591.
Has a table
19%
52 of 272 pages.
FAQ heading or FAQPage schema
40%
110 of 272 pages.
Visible updated or published date
5%
14 of 272 pages.
Price near the top of the page
30%
82 of 272 pages, checked in the first 300 words.
This is descriptive only. There is no control group of uncited pages, so none of it says what makes a page get cited.
ChatGPT specific pages had a median of 928.5 visible words (n=272), against 1779.5 for Google specific (n=318) and 1591 for ChatGPT general (n=137).
Of the 272 ChatGPT specific pages, 19% had a table (52/272), 40% had an FAQ heading or FAQPage schema, 78% had any JSON-LD and 5% showed a visible updated or published date (14/272). 30% had a price in the first 300 words.
By page type, ChatGPT specific citations were service or product pages for 39%, home pages for 26%, pricing pages for 8% and blog posts for 6%. Cells with fewer than 10 URLs are marked thin.
Bars show the share of cited pages in each page type, for ChatGPT specific and Google specific. Page type comes from rules, not human labels.
| Group | ChatGPT specific | Google specific | Observations |
|---|---|---|---|
| service or product page | 106 (39%) | 71 (22%) | 272 / 318 |
| blog post or guide | 17 (6%) | 146 (46%) | 272 / 318 |
| pricing page | 22 (8%) | 15 (5%) | 272 / 318 |
| comparison or alternatives page | 5 (2%) (thin) | 34 (11%) | 272 / 318 |
| documentation | 17 (6%) | 3 (1%) (thin) | 272 / 318 |
| location page | 21 (8%) | 32 (10%) | 272 / 318 |
| home page | 70 (26%) | 6 (2%) (thin) | 272 / 318 |
| other | 14 (5%) | 11 (3%) | 272 / 318 |
The same numbers are in the table below.
| Set | Citations (unique URLs) | Fetched OK | Share of URLs |
|---|---|---|---|
| ChatGPT specific | 316 | 272 | 86% |
| Google specific | 359 | 318 | 89% |
| ChatGPT general | 160 | 137 | 86% |
| Google general | 164 | 140 | 85% |
| Measure | ChatGPT specific | Google specific | ChatGPT general | Google general |
|---|---|---|---|---|
| Median visible words | 928.5 (n=272) | 1779.5 (n=318) | 1591 (n=137) | 1915 (n=140) |
| Median h2 count | 6 (n=272) | 9 (n=318) | 10 (n=137) | 9 (n=140) |
| Has a table | 52/272 19% | 114/318 36% | 77/137 56% | 47/140 34% |
| Has FAQ heading or FAQPage JSON-LD | 110/272 40% | 187/318 59% | 77/137 56% | 69/140 49% |
| Has FAQPage JSON-LD | 87/272 32% | 112/318 35% | 58/137 42% | 32/140 23% |
| Has any JSON-LD | 213/272 78% | 270/318 85% | 122/137 89% | 117/140 84% |
| Price or currency in first 300 words | 82/272 30% | 84/318 26% | 43/137 31% | 32/140 23% |
| Visible updated or published date found | 14/272 5% | 43/318 14% | 41/137 30% | 23/140 16% |
| Median age of visible date, days | 34.5 (n=14) | 87 (n=43) | 49 (n=41) | 86 (n=23) |
| Visible date within 180 days (of dated) | 13/14 93% | 33/43 77% | 41/41 100% | 18/23 78% |
| Any date incl. meta and JSON-LD | 71/272 26% | 210/318 66% | 97/137 71% | 98/140 70% |
| Median age of meta or JSON-LD date, days | 50 (n=66) | 95 (n=204) | 66 (n=94) | 107 (n=97) |
| Median question words in title | 2 (n=272) | 2 (n=318) | 2 (n=137) | 2 (n=140) |
| Median question words in h1 | 1 (n=272) | 2 (n=318) | 2 (n=137) | 2 (n=140) |
| Median share of question words in title | 17% (n=272) | 18% (n=318) | 50% (n=137) | 67% (n=140) |
| Median share of question words in h1 | 8% (n=272) | 17% (n=318) | 50% (n=137) | 67% (n=140) |
| Title has at least 1 question word | 223/272 82% | 303/318 95% | 121/137 88% | 139/140 99% |
| h1 has at least 1 question word | 175/272 64% | 271/318 85% | 105/137 77% | 127/140 91% |
| Page has no h1 | 19/272 7% | 11/318 3% | 4/137 3% | 4/140 3% |
Each cell shows the median or the count and share of pages with data.
| Type | ChatGPT specific | Google specific | ChatGPT general | Google general |
|---|---|---|---|---|
| service or product page | 106 (39%) | 71 (22%) | 4 (3%) | 17 (12%) |
| blog post or guide | 17 (6%) | 146 (46%) | 41 (30%) | 69 (49%) |
| pricing page | 22 (8%) | 15 (5%) | 4 (3%) | 3 (2%) |
| comparison or alternatives page | 5 (2%) | 34 (11%) | 44 (32%) | 10 (7%) |
| documentation | 17 (6%) | 3 (1%) | 1 (1%) | 1 (1%) |
| location page | 21 (8%) | 32 (10%) | 10 (7%) | 14 (10%) |
| home page | 70 (26%) | 6 (2%) | 31 (23%) | 22 (16%) |
| other | 14 (5%) | 11 (3%) | 2 (1%) | 4 (3%) |
ChatGPT citations were collected 2026-10-10 through DataForSEO's ChatGPT scraper, cited pages only. Google rows are earlier SERP data. The pages are from small providers' own sites. There are 981 unique URLs across four sets: ChatGPT specific, Google specific, ChatGPT general and Google general. 564 were reused from the first pass and 417 were fetched new for this run.
Pages were fetched with a plain HTTP client, no JavaScript. Word count is visible body text with script, style, nav, footer, aside and form blocks removed. A visible date is one next to a label such as Updated or Published. Meta and JSON-LD dates are reported separately.
The price check looks for a currency symbol before a number, or a number followed by a currency code, in the first 300 visible words. Page type comes from rules. An analyst check agreed with 33 of 40 (83%) of a random sample.
The CSV files hold the figures behind the tables. Units are in the column names.
Peak Answer, "What Pages ChatGPT Cites Look Like", https://peakanswer.com/research/cited-page-anatomy.