Why a Wire Stays Off My Citation Plan

I would buy a wire release for the audiences it has always served, and I would keep it off the line I call AI visibility. Wires are cited. They are cited far less often than the editorial publications a pitch deck says we can skip. On one denominator, 16,039 monitored answer runs, forbes.com is cited in 665 and prnewswire.com in 182. Forbes is cited 3.65 times as often as PR Newswire. The wire class holds none of the top-ten slots in any published segment.

The domain-level record I would open beside this decision is PR Newswire AI Citation Authority | MR Research. That page is the deeper dataset. This essay is how I would use it as a founder: when a wire is the right invoice, when it is a distraction from citation, and what would put it back on the citation plan.

The rates, on one window

Figures are from Machine Relations Index release mri_score_v2.0+2026-09-20+47973f373a20, window May 10 to September 20, 2026, 127 days, 919 monitored prompts, 16,039 answer runs, 22,320 cited domains, six engines. A domain counts once per answer run in which it is cited. Rates publish only past an evidence floor of 10 observed runs across 7 distinct run dates.

Publication	Citation rate	Cited runs	Days cited	Engines	Confidence	Standing
medium.com	5.04%	808 of 16,039	109 of 127	6	A	rank 1 of 1,230 classified editorial publications
forbes.com	4.15%	665 of 16,039	106 of 127	6	A	rank 2 of 1,230
prnewswire.com	1.13%	182 of 16,039	75 of 127	6	B	rank 8 of 1,230
techcrunch.com	1.02%	164 of 16,039	44 of 127	5, Gemini absent	B	rank 9 of 1,230
businesswire.com	0.61%	98 of 16,039	51 of 127	6	C	rank 74 of 22,320 observed domains
globenewswire.com	0.46%	74 of 16,039	37 of 127	at least 5	C	rank 110 of 22,320

I would read the editorial-class rank with the field it comes from. The index's source_role in this release combines source-type evidence with AuthorityTech's placement catalog. Ninety-six of the 1,231 class members are admitted through the catalog, and editorial classification is moving to source-type evidence alone. The class includes six platform domains beside edited newsrooms: medium.com at class rank 1, amazon.com at 7, prnewswire.com at 8, apple.com at 16, wordpress.com at 239, and blogspot.com at 330. Substack, beehiiv, and dev.to, the same kind of hosted publishing as Medium, are filed as community platforms. Those six hold 1,358 of the editorial class's 13,525 cited runs, 10.04 percent, and they include the top slot. The class denominator moves between releases. Published figures carry 1,025 to 1,230, because each release counts the domains it classified. Citation rate, cited runs, days cited, engine breadth, and confidence on a fixed run set do not depend on that denominator. I would read the rate first.

Engine counts need the September 26, 2026 correction. Three of the six monitored engines had collection gaps in this window, not two. Google AI Overviews recorded no citations from August 15 to September 24, 2026. Google AI Mode recorded none from September 14 to September 24. Gemini recorded none on 22 intermittent days, May 11 to 15 and then scattered days from August 11 to September 16. ChatGPT, Claude, and Perplexity collected throughout. One Claude date was flagged with exactly one answer run, a scan artifact rather than a gap. An engine count can only fall because of a gap. It cannot rise. Where one of the three gapped engines is among those absent, I would state the count as a floor, "at least N," because this release cannot tell an engine that passed the domain over from an engine that had no chance to see it. A count is exact only when every absent engine is ChatGPT, Claude, or Perplexity, and then I would name the absent engine. TechCrunch's five, with Gemini absent, is that kind of exact count. GlobeNewswire's "at least 5" is a floor. Cited runs, days cited, confidence, and standing are computed on the full run set. Citation rates are not: a blind run stays in the denominator and adds nothing to the numerator, so every rate on a window that includes these gaps is a floor and understates.

The wire and press-release distribution class is 9 domains holding 232 cited runs out of 112,518 across the nine source classes, 0.21 percent of classified citation. The index files prnewswire.com, the largest wire domain in it, as an editorial publication, so the nine-domain class excludes it. Counted with the wires, the class is ten domains holding 414 of those 112,518 cited runs, 0.37 percent. PR Newswire's best standing in any published segment is number 15 of 448. Across all 88 published segments on the September 19, 2026 release, wire distribution held 0 of 880 top-ten slots. The finding I would act on survives the inclusion of PR Newswire: this class is not where top-ten citation slots are won.

Syndication-heavy surfaces sit lower still. techbullion.com is cited in 12 of 16,039 runs, on 11 of 127 days, by at least 3 of the six engines, at a 0.07 percent rate, and it sits at Collecting rather than a confidence grade. I would not pay a syndication network to imitate a citation strategy when the measured member of that pattern has not cleared the floor.

The definitions I would use for a citation rate and a confidence grade are in the glossary. The reason I would rather earn a page in a cited publication than multiply URLs is the discipline on Machine Relations.

Inventory is not selection

A cumulative URL count rewards whoever publishes the most pages earliest. A wire release lands as one story on hundreds of syndicated URLs, so a counter that counts URLs puts wires on top before any engine has chosen a source. Rank the same market by the share of monitored answer runs in which a domain is cited at all, and the order changes. Forbes at 665 runs and PR Newswire at 182 are that second ranking. Medium, at 808, is ahead of both, which is a separate argument about what the editorial class currently contains. It is not an argument that a wire's URL count was the selection.

A trailing window should be able to rise and fall. A figure that only climbs under a "30-day" label is a cumulative count wearing a window's name. I would ask any vendor for the window, the denominator, and the evidence floor. A raw count with no denominator is inventory.

Structure makes a page eligible. It does not make the engine choose it.

Specific, sourced, structured writing extracts more cleanly than narrative. I would still write that way. What does not follow is that paying a wire to distribute the structured page buys the citation. Structure is a property of the page. Citation is a decision the engine makes. The index measures the decision.

The vendors are selling the other reading in their own words. GlobeNewswire, read on September 20, 2026, offers to help a brand "shape your presence in AI answers" and leads with a framework it calls "The blueprint for AI-visible press releases." PR Newswire carries a Create with AI product. Cision has announced the acquisition of a generative-AI monitoring product. Set that copy beside globenewswire.com cited in 74 of 16,039 runs, on 37 of 127 days, at confidence C. The positioning and the standing are different objects. Only the standing is checkable from the release. I would let a vendor sell distribution. I would not let the brochure redefine 74 cited runs as a citation strategy.

Pew's March 12, 2026 account of how Americans use and view AI is the demand-side reason I want an instrument rather than a counter. The interest is not a passing topic. A counter that climbs with every syndicated URL will keep looking like success while the selection numbers stay where this table puts them.

The decision

I would rank publication targets by citation rate on one denominator, using the domain pages in the index, each of which carries rate, runs, days, and confidence. I would treat a wire as a distribution cost for the audiences that need the release: investors on a list, a regulator, a partner who asked for a document, a newsroom that prefers to receive news that way. I would not book it as the AI-visibility line. I would ask any citation number offered in a proposal for its window, its denominator, and its floor of 10 runs across 7 dates. And I would control what the engines can fetch from us directly. The crawlers are documented: Google's common crawlers, OpenAI's bots, Perplexity's bots, and Anthropic's crawler documentation.

Scope: the September 20 release, the six domains in the table, and any budget I had been asked to defend as "citation" that is actually a wire invoice. PR Newswire at 1.13 percent and confidence B is a real, modest editorial-class rate, 182 runs, 75 days, six engines. It is also 3.65 times behind Forbes and absent from every top-ten slot as a member of the wire class once the class is counted the way the segment tables count it, 0 of 880. Business Wire at 0.61 percent and GlobeNewswire at 0.46 percent, both confidence C, are weaker still. TechBullion at Collecting does not clear the conversation.

What would change my mind: a later release, same floor, in which a wire domain holds a top-ten slot in a published segment that matches our buyer's question shape, with the rate computed on a window I can read. One campaign that spikes a cumulative URL count would not qualify. If GlobeNewswire's 74 of 16,039 moved to a rate and a confidence grade comparable to Forbes's 4.15 percent at confidence A, I would reopen the "shape your presence in AI answers" claim and pay for the distribution as citation. Until then I would pay for it as distribution, and I would put citation budget on domains that already win selection. The collection gaps mean I would treat current rates as floors. A floor that still leaves the wire class at 0.21 percent, or 0.37 percent with PR Newswire folded in, is not a close call.

FAQ

Do answer engines cite wire services at all?

Yes. PR Newswire is in 182 of 16,039 runs, Business Wire in 98, GlobeNewswire in 74. Forbes is in 665 and Medium in 808 on the same runs. Cited and competitive are different descriptions. The 3.65 times gap between Forbes and PR Newswire is the one I would use in a budget conversation.

Why is PR Newswire both a wire and an editorial rank?

Because this release files prnewswire.com as an editorial publication, rank 8 of 1,230, so the nine-domain wire class excludes it and totals 232 of 112,518 cited runs, 0.21 percent. Adding it back makes ten domains and 414 runs, 0.37 percent. Its best segment standing is 15 of 448, so the class still holds 0 of 880 top-ten slots across 88 segments on the September 19 release. I would mention both filings so a rank of 8 cannot be waved as a contradiction of the slot count.

What would I say to a vendor quoting a citation total?

I would ask for the window, the denominator, and the evidence floor. The floor on this index is 10 observed runs across 7 distinct dates, which is why TechBullion at 12 runs and Collecting status is not a grade. I would also ask whether the number can fall. A 30-day label on a figure that only climbs is a cumulative inventory count. Selection, on this release, is share of 16,039 answer runs.

Does a well-structured release change the table?

It changes the page. Engines extract specific, sourced writing more cleanly, and I would write releases that way for the humans and the archives that need them. Citation in this index is the decision to use the domain in an answer. GlobeNewswire's own AI-answer language, read September 20, 2026, sits next to 74 cited runs and confidence C. I would not let the blueprint substitute for the rate.

How should I read engine counts after the collection gaps?

ChatGPT, Claude, and Perplexity ran throughout, with one Claude date reduced to a single run by a scan artifact. Google AI Overviews, Google AI Mode, and Gemini had dated gaps, so an absence that includes one of them is an "at least" count. Rates that span the gaps are floors. I would not promote a precise absence story for an engine that was not collected, and I would not ignore a rate that remains low even as a floor.