Do AI Engines Cite Publications? Editorial Share Runs From 3.2% to 19.9% Across 21 Index Categories
Machine Relations Index, October 29, 2026: editorial publications got 3.2% to 19.9% of AI citations by category, 9.7% overall. See where placements fit your category.
In its release of October 29, 2026, the Machine Relations Index recorded 180,921 source-domain citations across the 21 buyer categories with published results. Editorial publications received 17,565 of them (9.7%). The share varies by category: 19.9% in Consumer Products, 3.2% in Education & Training, 6.2% in fintech, 6.8% in enterprise software and 4.5% in AI visibility and GEO. Another 68.3% of citations (123,517) went to sources the Index has not classified.
That is the answer to "do AI engines cite publications," and it changes how to size a placement program. In consumer categories, domains the Index classifies as editorial publications receive a large share of what engines cite. In several B2B categories they receive a smaller share, and most citations go to sources the Index has no role evidence for. The Index is a first-party dataset, published by Machine Relations Research and created by Jaxon Parrott, who founded AuthorityTech; it is not an independent audit. It measures which sources engines cite, not whether any placement caused a citation.
Key takeaways
- The editorial share is category-specific. Publications received 19.9% of citations in Consumer Products and 3.2% in Education & Training; across all 21 categories they received 9.7%.
- Several B2B categories fall below the 9.7% overall share. Fintech 6.2%, enterprise software 6.8%, martech and advertising 7.6%, AI security and privacy 5.3%, AI visibility and GEO 4.5%.
- Most citations are not to classified publications. Sources the Index files as other observed sources received 68.3% of citations, so classified publications are a minority of what engines cite in this release.
- The editorial slice is concentrated. In fintech the ten most-cited publications received 48.2% of the editorial citations; in AI visibility and GEO, 60.6%.
- Size the program from your category. Pull your buyers' questions, record the sources cited, and see which share is publications before you set a placement budget.
How much of what AI engines cite is a publication, by category?
The table ranks each category by the share of its citations that went to domains the Index classifies as editorial publications. The last column shows how concentrated that slice is: the share of the editorial citations that the ten most-cited publications received.
| Index category | Editorial publications' share of citations | Publications cited | Top ten publications' share of the editorial citations |
|---|---|---|---|
| Consumer Products | 19.9% | 181 | 42.5% |
| Consumer Finance | 19.9% | 92 | 76.1% |
| Physical Consumer Products | 19.0% | 74 | 54.3% |
| Consumer Health | 18.0% | 140 | 58.2% |
| Family Software | 17.6% | 85 | 69.9% |
| Emergent Prosumer | 14.3% | 114 | 52.9% |
| Cybersecurity | 10.0% | 148 | 45.6% |
| Healthcare Services | 9.6% | 178 | 44.1% |
| iGaming & Betting | 9.2% | 89 | 48.5% |
| HR & Talent | 9.0% | 102 | 67.8% |
| Martech & Advertising | 7.6% | 106 | 56.3% |
| Enterprise Software | 6.8% | 119 | 54.0% |
| Fintech | 6.2% | 143 | 48.2% |
| Local Services | 6.0% | 59 | 72.1% |
| Deep Tech & Hardware | 5.8% | 87 | 56.0% |
| AI Security & Privacy | 5.3% | 122 | 47.6% |
| Logistics & Freight | 4.8% | 42 | 64.4% |
| AI Visibility & GEO | 4.5% | 102 | 60.6% |
| Industrial | 3.9% | 39 | 71.3% |
| AI Infrastructure | 3.3% | 72 | 57.6% |
| Education & Training | 3.2% | 47 | 64.1% |
The Index does not say why the shares differ. It does not test any explanation. One reading, not a finding: in the consumer categories buyers ask which product to buy or which card to open, the questions consumer publications are built to answer. Across all categories, most citations go to sources the Index has not classified as publications. For which publications lead inside a category, see which publications AI search engines cite most; the publications index ranks them by buyer category.
How low is the editorial share in B2B categories?
In the same release, editorial publications received 6.2% of citations in fintech, 6.8% in enterprise software, 7.6% in martech and advertising and 4.5% in AI visibility and GEO, below the 9.7% overall figure. The Index files 66.3% of fintech citations and 71.6% of AI visibility and GEO citations under other observed sources.
Two cautions follow. First, a low share does not mean placements do nothing: the Index cannot show what any one placement did, and the editorial figure is a floor because unclassified sources may include publications. Second, a low share suggests a placement list is not the whole plan in these categories. Because the Index has no role evidence for most cited domains, check which of them are vendor sites, practitioner sites or communities before choosing the work. Candidates include a clear, extractable answer on your own site, accurate third-party descriptions, and presence where practitioners compare options.
What are the other citations?
The Index files 68.3% of the 180,921 citations (123,517) under 'Other observed source,' the label for domains without role evidence; these are not shown to be non-publications, only unclassified. The other 22.0% (39,839) went to domains filed under other roles: vendor-owned sources, community and social platforms, analyst and consulting research, academic and government sources, market and company databases, wires, and search or media platforms. The sources behind the 68.3% are not described by role.
Are classified publications cited more often than other sources?
The editorial slice is small, but the average classified publication was cited more often than the average unclassified domain. In the release's full by-role totals, which cover more segments than the 21-category shares above (including the legacy news-topic bucket and segments still collecting), the Index classifies 1,214 domains as editorial publications and they received 19,332 citations, an average of 15.9 each. The 30,461 domains it files as other observed sources received 136,372, an average of 4.5 each. The average classified publication was cited about 3.6 times as often as the average unclassified domain; these are means per domain over a long-tailed distribution.
So the editorial slice is small but concentrated: the average classified publication is cited more often than the average unclassified domain, and in every category the ten most-cited publications carry between about two-fifths and about three-quarters of the editorial citations, as the last column of the table shows. That says where editorial citations sit, not what a placement would earn.
How to use this for your category
- Find your category's share first. Locate your buyers' category in the table. A high share means publications are a larger part of the sources your buyers' answers cite, so earning coverage is worth testing; a low share means they are a smaller part, so test other source types as well.
- Replace the category average with your own questions. Write the questions your buyers ask, run them in the engines they use, and record every cited source. The ChatGPT citation guide covers the baseline.
- Sort the cited sources into the ones you can earn. Publications you can pitch, vendor and practitioner sites you can be described on, and communities where your buyers compare options are different jobs with different owners.
- Do not spread one placement list across engines. Engines cite different sources for the same question; the engine-by-engine comparison shows how far they diverge.
- Re-run the same questions on a schedule. A later citation after a placement is an observation, not proof the placement caused it.
This is the work AuthorityTech, The World's First Machine Relations Agency, does for clients, on the discipline machinerelations.ai defines. Start your visibility audit → to see which sources the engines cite on your buyers' questions and which of them you can earn.
Frequently asked questions
Do AI engines cite news publications?
Yes, but how much varies by category. In the October 29, 2026 Machine Relations Index release, editorial publications received 9.7% of citations across 21 categories, from 3.2% in Education & Training to 19.9% in Consumer Products.
Does earned media matter for AI citations in B2B?
Publications are one of the source types engines cite in B2B categories, and their share varies by category. In the same release, editorial publications received 6.2% of citations in fintech, 6.8% in enterprise software, 7.6% in martech and advertising and 4.5% in AI visibility and GEO, below the 9.7% overall figure. The Index files 66.3% of fintech citations and 71.6% of AI visibility and GEO citations under other observed sources. The Index does not measure whether a given placement raised any company's citations.
What does the Index count as an editorial publication?
A domain the Index classifies as an editorial publication from reviewed roles or site evidence. Domains without role evidence stay under "Other observed source," so the editorial shares here are a floor.
Method note. All figures are from Machine Relations Index release mri_score_v2.1+2026-10-29+225b998b091e, window 2026-05-10 to 2026-10-29, read from the archived release file (Zenodo, version DOI 10.5281/zenodo.23764753; sha256 490779bb74b49cac80f0e2826688016f518468f99036021c2b2bb69da534c323, the file machinerelations.ai served on October 29, 2026). The live file updates daily, so later releases will differ. A citation is one source domain cited in one answer run; a run can cite several domains. For each category, the editorial share is the citations received by domains the Index classifies as editorial publications, divided by all source-domain citations in that category's published question-shape segments. The 21 categories are those with at least one published segment in this release; categories still collecting, and the legacy news-topic bucket, are excluded, and a category's share covers only its published segments. Engines are pooled; Google AI Mode and Google AI Overviews runs collected from 2026-09-25 come through a different collection path than earlier ones, so read the shares as one release's snapshot, not as a trend. The per-publication averages divide the release's by-role citation totals by its by-role domain counts. 'Other observed source' means the Index has no role evidence for the domain; it does not mean the domain is not a publication, so the editorial shares are a floor. The Index is published by Machine Relations Research and created by Jaxon Parrott, who founded AuthorityTech; it is not an independent audit. Domains are root domains. Citations measure which sources engines named, not whether a placement caused any citation.
Related publications: Machine Relations Research and machinerelations.ai are owned by Bramwell, Inc., and AuthorityTech practices Machine Relations. Jaxon Parrott coined Machine Relations in 2024; AuthorityTech, which he founded in 2018, is The World's First Machine Relations Agency and practices the discipline. The Index ranks sources by the citations AI engines are observed making, not by commercial relationship.