The Publications AI Cites for Your Industry News Are Not the Ones It Cites for Your Buyers

The pitch list your team built from industry news coverage does not carry into the answers your buyers get. In five categories where AI answer engines were measured on both, 104 publications hold a top-100 slot for news-driven questions and only 32 of them hold a top-100 slot on a published buying question. The other 72 — 69.2% — win the news answer and are absent from the buying answer.

SaaStr is the clean case. It ranks 4th of 1,375 cited sources on enterprise software news, cited in 48 of 596 observed answer runs (8.05%) across 53 run dates. On its domain profile that news segment is the only segment in the entire release where it has been observed at all. TechTarget, in the same category and the same release, ranks 1,234th of 1,375 on that news segment, cited in 1 of 596 runs (0.17%) — and ranks 3rd of 172 sources on "how buyers choose," cited in 18 of 99 runs (18.18%), and 3rd of 186 on head-to-head comparisons, cited in 14 of 114 runs (12.28%).

One of those two publications is on every enterprise software media plan. The other is the one a buyer's answer is made of.

What was compared

The Machine Relations Index measures which source domains six answer engines cite when real buying questions are asked. Release mri_score_v2.0+2026-09-21+d73e239e7c5b covers 2026-05-10 to 2026-09-21, 16,146 monitored answer runs and 22,377 cited domains.

It separates a subject category into question shapes. One of them, news-driven citations, is what an engine cites when the question is about what happened in your industry. Six others — best tools, how buyers choose, is it worth it, problem-first, top lists, comparisons — are what it cites when someone is deciding what to buy.

Five categories currently publish both a news segment and at least one buyer segment: cybersecurity, enterprise software, fintech, healthcare services and HR & Talent. That is what makes the comparison possible. The method is one question applied to every domain the release classifies as an editorial publication in each segment's published top 100: does the same publication appear in both lists?

Category	News segment evidence	Publications in news top 100	Buyer shapes published	Publications in any buyer top 100	In both	News only
Cybersecurity	623 runs, 54 dates	30	6	36	11	19
Enterprise software	596 runs, 53 dates	15	6	25	3	12
Fintech	616 runs, 53 dates	15	6	20	8	7
Healthcare services	613 runs, 53 dates	25	3	13	5	20
HR & Talent	608 runs, 53 dates	19	1	5	5	14
Total		104	22	99	32	72

It runs the other way too. Of the 99 publications holding a buyer-question slot, 67 — 67.7% — hold no top-100 slot on their category's news segment. Those are the outlets your media plan has no reason to be pitching and your buyer is reading anyway.

Widening it past the editorial class does not soften it. Counting every source role, 500 domains hold a news top-100 slot across these five categories and 144 also hold a buyer slot; 356 of them, 71.2%, do not.

HR is the newest and narrowest test

HR & Talent joined the comparison on September 21 when its how-buyers-choose segment cleared the evidence floor at 101 observed runs across seven dates. Its news segment is much deeper: 608 runs across 53 dates.

Nineteen editorial publications hold a top-100 slot in the HR news answer. Five hold a top-100 slot in the buyer-choice answer. All five buyer publications also appear in news; 14 of the 19 news publications do not transfer. Across every source role, only 21 of the two top-100 lists overlap. The Jaccard overlap is 11.7%.

The ranks show why a shared masthead does not make the lists interchangeable:

Publication	HR news rank	HR buyer-choice rank	What the gap says
People Managing People	8	11	Strong in both shapes
SelectSoftwareReviews	38	19	Stronger in the buyer answer
HR Executive	7	99	News authority barely transfers
TechTarget	41	83	Present in both, but not a buyer leader here
TechRadar	53	82	Present in both at the long edge
Unleash	1	No top-100 slot	The news leader does not enter the buyer list
HR Tech Feed	9	No top-100 slot	A top-ten news source does not enter the buyer list

This is the cleanest category-level warning in the study because the buyer segment is new. The absence of Unleash and HR Tech Feed from its top 100 may persist or may narrow as more dates accumulate. The current release supports one statement: a newsroom can dominate what happened in HR without becoming a source engines use to explain how to choose HR software.

Healthcare is the starkest version

Fierce Healthcare ranks 2nd of 1,264 sources on healthcare services news, cited in 117 of 613 observed runs (19.09%) across 53 run dates. That is a genuinely dominant position, earned across nearly two months of daily observation, with a Confidence B grade and all six engines citing it.

Healthcare services publishes three buyer segments in this release: best tools, how buyers choose and top lists. Fierce Healthcare holds no top-100 slot in any of the three. Its profile shows two other healthcare observations, both still collecting: 3 cited runs of 101 on "is it worth it" over 6 dates, and 2 of 102 on problem-first over 6 dates. Neither has enough evidence to carry a rank.

The rest of the healthcare news list behaves the same way. Healthcare Dive ranks 7th of 1,264 on news, cited in 43 of 613 runs (7.01%). MedCity News ranks 14th, cited in 33 of 613 (5.38%). Neither appears in a published buyer top 100. Twenty of the 25 publications in the healthcare news answer are news-only.

That is the entire healthcare trade pitch list, working exactly as intended, in a question shape that no buyer asks before signing.

Cybersecurity and fintech, same pattern

SecurityWeek ranks 7th of 1,340 on cybersecurity news, cited in 44 of 623 runs (7.06%). Across the whole release its only buyer-question observation anywhere is in a different category: rank 225 of 272 on AI security problem-first research, cited in 1 of 106 runs (0.94%). CRN ranks 14th of 1,340 on the same news segment, cited in 30 of 623 runs (4.82%), with one buyer observation — rank 120 of 195 in deep tech, 1 of 130 runs (0.77%). Infosecurity Magazine ranks 13th on news and holds no buyer slot.

TechCrunch is the one every founder asks for. It ranks 64th of 1,340 on cybersecurity news, cited in 14 of 623 runs (2.25%), and it is cited in seven different news segments. Its three buyer-question appearances are rank 240 of 272, rank 162 of 170 and rank 73 of 136. It is a news source in this dataset, comprehensively, and a buying source almost nowhere.

Fintech is the mildest of the four and still splits. Fintech Magazine ranks 12th of 1,531 on fintech news, cited in 32 of 616 runs (5.19%), and ranks 155th of 224 on best tools, cited in 1 of 108 runs (0.93%). Finextra ranks 59th on news, cited in 14 of 616 (2.27%), and sits at rank 149 of 200 on top lists and rank 108 of 136 on comparisons, one cited run each.

Why the two lists are different lists

A news question and a buying question are retrieved against different evidence.

News rewards recency, an event, and a publication whose business is covering that event fast. The engines have 53 or 54 days of observation on these segments and a deep field — 1,264 to 1,531 distinct cited domains in each — and the outlets that surface are the ones filing daily.

A buying question rewards something else entirely: a page that sets out options, criteria, prices or a head-to-head. TechTarget's position is not an accident of prestige. It ranks 3rd on how-buyers-choose and 3rd on comparisons in enterprise software because that is structurally what it publishes. Across the whole release it is cited in 240 of 15,883 runs (1.51%) on 80 of 126 days, at rank 6 of 1,224 classified editorial publications, with Confidence B and six engines.

How to read the editorial-class rank. The editorial-class rank above (TechTarget at rank 6 of 1,224) is read off the Machine Relations Index's source_role field, which in the current release combines source-type evidence with AuthorityTech's placement catalog; 96 of the 1,231 class members are admitted through the catalog, and the Index is moving editorial classification to source-type evidence alone. The class currently includes six platform domains alongside edited newsrooms: medium.com (class rank 1), amazon.com (7), prnewswire.com (8), apple.com (16), wordpress.com (239) and blogspot.com (330). At the same time substack.com, beehiiv.com and dev.to — the same hosted open-publishing product as Medium — are filed as community platforms. Those six hold 1,358 of the editorial class's 13,525 cited runs, 10.04%, and they include the class's top slot. The class denominator also moves between releases (published figures carry 1,025 to 1,230) because it counts the domains classified in each release. What is unaffected: citation rates, cited runs, days cited, engine breadth and confidence are direct per-domain observations over a fixed run set, and are the primary measure; read the rate first. See the MRI methodology and update log.

Coverage of your funding round is not the evidence a buying answer is assembled from. It never was. The difference is that a search result used to let a buyer click through to whatever else you had; an answer does not.

What this does not say

It does not say these publications are never cited on buying questions. It says they hold no slot in a published top 100. For the eight publications named above the full domain profile was read — it lists every segment where the domain has been observed at any rank — and the profiles agree with the leaderboards. For the other 77, absence means absence from the top 100 of that segment.

The two lists are not equally mature. News segments here carry 596 to 623 observed runs over 53 or 54 dates. Buyer segments carry 75 to 136 runs over 7 dates. News is the deeper measurement. A publication absent from a buyer list may be genuinely absent or merely early, and the release marks the difference: a segment publishes only after 10 observed runs across 7 distinct dates, and anything below that reads as collecting rather than zero.

The source-role class is the release's own classifier. The editorial-publication class in these segments includes properties operated by vendors — nordlayer.com, oneidentity.com and securityscorecard.com all carry it. The class is reported here as measured, not re-cut by hand. Every publication named in the body is unambiguous trade or business press, so the split is not an artifact of where the classifier draws its line.

No causal claim. The index records which domains engines cited, not why, and a placement is never shown here to have caused a citation. Segment-level confidence reads Unavailable; grades in this release are domain-level.

What to do with it

1. Split your media list in two. One list for the news answer, one for the buying answer. They share about three names in ten.
2. Decide which one you are buying before the next pitch. If the goal is awareness on a launch, the news list is the right list. If the goal is being in the answer when someone asks how to choose, it is the wrong list and no amount of it will do the job.
3. Read your own category's shapes, not the average. The five categories here split at 19 news-only of 30, 12 of 15, 7 of 15, 20 of 25 and 14 of 19. Your category's leaderboards are public and rebuild daily.
4. Pitch the publication that holds the shape. For enterprise software buying questions that is TechTarget before anything on the news list. For cybersecurity, the per-shape breakdown is in the cybersecurity pitch list.
5. Report placements scoped to their shape. "Cited in the news answer for our category" is a true and useful result. Let it be that, rather than letting it get reported as buyer visibility it does not carry.

The sales handoff

Our coverage in [publication] holds a top-ten position in what AI engines cite for industry news in our category. It is not currently a source those engines cite when a buyer asks how to choose — that slot is held by [publication]. Both are real; they answer different questions.

A citation is evidence of presence in one question shape and nothing else. Saying which shape is what keeps an earned-media result from being handed a job it was never built to do.

Why this is a Machine Relations problem

Machine Relations is the practice of managing the relationship between your brand and the machines that answer questions about it. A tier-one media list is a media artifact, built for a world where coverage and consideration lived on the same page. They no longer do. The engines keep two lists, and only one of them is in front of a buyer at the moment of the decision.

Sources and method

Every figure above was re-read on 2026-09-21 from the live public leaderboards of Machine Relations Index release mri_score_v2.0+2026-09-21+d73e239e7c5b, window 2026-05-10 to 2026-09-21. Ranks and totals are read from the release's own per-segment standing rather than re-derived, because the release uses competition ranking and a re-sort disagrees with it on ties.

Segments read, all 200: the news segment and all six buyer shapes for cybersecurity, enterprise software and fintech, the news segment plus the three published buyer shapes for healthcare services, and the news segment plus the newly published how-buyers-choose shape for HR & Talent. Worked examples: cybersecurity news, enterprise software how-buyers-choose, healthcare services best tools, HR & Talent news and HR & Talent how-buyers-choose.

Domain profiles read for every publication named in the body: saastr.com, techtarget.com, fiercehealthcare.com, securityweek.com, crn.com, techcrunch.com, fintechmagazine.com, finextra.com, unleash.ai, hrtechfeed.com, peoplemanagingpeople.com, selectsoftwarereviews.com, hrexecutive.com, techtarget.com and techradar.com. The full release is machine-readable at machine-relations-index.json.

Publications cited in the body, read at their own sites on 2026-09-19: SaaStr, SaaS Mag, TechTarget, ITPro, Fierce Healthcare, Healthcare Dive, MedCity News, SecurityWeek, CRN, Infosecurity Magazine, TechCrunch, Fintech Magazine, Finextra.

FAQ

Does this mean industry press is worthless for AI visibility? No. It means it is worth something specific. In these five categories the trade press dominates the news answer, which is the answer a buyer gets when they ask what is happening in a market. It is not the answer they get when they ask what to buy, and 72 of 104 publications appear in one and not the other.

Why does TechTarget beat publications with far more readers? Because the buying question retrieves a different kind of page. TechTarget ranks 3rd of 172 on enterprise software how-buyers-choose, cited in 18 of 99 observed runs, and 1,234th of 1,375 on the same category's news segment. Audience size is not what is being measured; what the page does for the question is.

How many categories does this cover? Five — the ones where the release publishes both a news segment and at least one buyer segment. Two more categories publish a news segment with no published buyer segment yet, so they cannot be compared.

How often do these lists change? The index rebuilds daily and each segment states its release id, window and run counts. News segments move slowly because they carry 53 or 54 dates of evidence; buyer segments carry 7 and will move faster as they accrue.

What should I ask my agency? Which question shape each placement is aimed at, and which publication currently holds that shape in our category. If the answer is a tier list, the list was built for the news answer.