PR Is the New SEO: The Digital-PR Citation Playbook (Featured/Qwoted, Not HARO)

Earned Editorial Drives ~25% of All AI Citations — and Only 6% of GEO Practitioners Prioritize It. That Gap Is the Cheapest Advantage Left in B2B Search. Here’s the Pitch-to-Citation Playbook.

The press release on the wire correlates with AI citation at 0.04 — statistical noise. The editorial pickup of that story correlates at a different order of magnitude. Most companies are buying the distribution and skipping the thing that actually moves the needle.

The hard data:

• Earned media and journalism = ~25% of all large language model citations; non-paid sources = ~94% of all AI-cited links (Muck Rack “Generative Pulse,” analyzing 1M+ links across ChatGPT, Claude, Gemini, and Perplexity, March 2026). Muck Rack
• Only ~6% of GEO practitioners use digital PR as their primary tactic (NP Digital survey of 200 practitioners, April 2026, via The Digital Bloom) — the widest gap between what works and what people do in the whole discipline. The Digital Bloom
• Unlinked brand mentions correlate with AI citation at 0.664; backlinks at only 0.218 — a roughly 3x gap in favor of the mention over the link (Ahrefs, 75,000 brands). Ahrefs via Soar
• The journalists PR teams pitch most have only a ~2% overlap with the journalists AI cites most (Muck Rack). Most PR effort aims at the wrong outlets for AI visibility. Muck Rack
• Distributing the same article across third-party sites lifted brand citations a median +239% across AI engines; one pilot hit +325% (The Digital Bloom).
• The wire press release alone moved AI Overview citations only 0.2% → 1% in six months (Ahrefs) — the distribution is not the signal.
• Recency impact peaks within the first 7 days of coverage (Muck Rack). Timing is a lever, not an afterthought.

The thesis: PR is now a search discipline. Earned editorial is the single largest controllable driver of AI citations — and it is the most under-used, which makes it the closest thing to blue ocean left in B2B search. But the old playbook is broken twice over: HARO is gone (Connectively was discontinued in December 2024 — you now use Featured.com, Qwoted, Help a B2B Writer, and SourceBottle), and the prestige-outlet target list AI doesn’t even read.

Who should read this: the B2B operator-founder who keeps hearing “we need PR” and can’t see the ROI, and the in-house marketer who’s been told to “do some thought leadership” with no budget and a boss who’ll take the credit and pass down the blame. This is the playbook that turns earned media from a vanity exercise into a citation-engineering function — with the dollar figures attached.

1. The Reframe: You Are No Longer Pitching a Journalist. You Are Pitching the Machine That Reads the Journalist.

Start with the single sentence that changes how you write every pitch from now on: when you pitch a publication, you are not just pitching a human journalist — you are pitching the LLM that the journalist’s publication feeds.

That is not a metaphor. Retrieval-augmented generation works by pulling text from sources the model trusts, then assembling an answer. When a journalist quotes you in Fortune or Fast Company, that quote becomes a sentence on a page that ChatGPT, Perplexity, Gemini, and Google AI Overviews crawl, index, and cite. The journalist is the gatekeeper. The model is the audience that lasts.

This is why the discipline boundary between PR and SEO has collapsed. For two decades, PR optimized for human attention — the eyeballs on the article, the prestige of the masthead, the clip you forwarded to your board. Search optimized for the algorithm. Those were different teams with different metrics. They are now the same job, because the same earned-editorial sentence does both: it reaches the human reader and it becomes citation fuel for the machine that recommends your category.

The data forces the merger. Earned media is ~25% of all LLM citations and non-paid sources are ~94% of all AI-cited links (Muck Rack, 1M+ links analyzed). You cannot buy your way into that 94%. You earn it — or your competitor does.

The extractable line for this section: Earned editorial accounts for roughly a quarter of every citation AI engines make, and ~94% of all AI-cited links come from non-paid sources — which means the press release you paid to distribute matters far less than the editorial pickup you earned, and you cannot purchase your way onto the list.

This is Rule #44, Name-the-Problem, applied to your media strategy: the problem isn’t that you lack PR. It’s that you’ve been measuring PR by the wrong scoreboard — impressions and prestige — while the scoreboard that pays you (citation density across trusted sources) went unwatched.

2. The 0.04 Problem: Why Your Press Release Is Nearly Worthless and the Pickup Is Everything

Here is the finding that should rewire your PR budget. Ahrefs’ study of 75,000 brands measured how strongly each brand signal correlates with AI citation. Press-release-syndication-on-its-own scored 0.04 — indistinguishable from random noise. (Ahrefs via Soar.)

Put that next to the rest of the hierarchy:

Source: Ahrefs, 75,000 brands (via Soar).

Then look at what the wire release actually delivered over time: AI Overview citations from wire-distributed press releases crept from 0.2% to 1% across six months (Ahrefs). That’s the entire payoff of the thing most companies treat as “doing PR” — paying a wire service to blast a release to a list of sites that republish it verbatim.

The distinction the data is screaming is this: the distribution mechanism is not the signal. The earned editorial pickup is the signal. A wire release is a delivery truck. What gets cited is what a real journalist, at a real publication, chose to write about you in their own words — because that’s the page the model trusts.

The extractable line: Wire-syndicated press releases correlate with AI citation at 0.04 and moved AI Overview citations only from 0.2% to 1% over six months — the distribution itself is nearly worthless; what AI cites is the editorial pickup, the sentence a journalist wrote about you in their own words.

I learned this version of the lesson the expensive way before any of these studies existed. Years ago, running the Omega Group, we paid for wire distribution of a precast-manufacturing announcement. It “reached” hundreds of sites. Not one buyer mentioned it, and not one inbound lead traced to it. The release that did move work was a single trade-publication article a real editor wrote — because a real editor’s words carry the trust signal a syndicated reprint never will. The data now explains why: the editor’s article is what the machine reads.

3. HARO Is Dead. Here Are the Four Tools That Replaced It.

If your PR plan still names HARO, your plan is dated. Connectively — the rebranded HARO — was permanently discontinued by Cision on December 9, 2024. (Cision.) Any agency or playbook still telling you to “sign up for HARO” hasn’t updated since 2024.

Here is the live 2026 toolkit, and what each one is actually for:

Featured.com is the one to understand deeply, because its model maps exactly onto the citation mechanism. You submit a tight, factual expert insight. Their editors select the strongest submissions and build a published article around them — at outlets like Fortune, Fast Company, and Yahoo. The output is precisely the asset Ahrefs measured at 0.664: an unlinked editorial brand mention on a high-trust page. You didn’t buy a link. You earned a sentence in someone else’s editorial, which is the sentence the model cites.

The extractable line: HARO (Connectively) was discontinued by Cision on December 9, 2024; the live 2026 replacements are Featured.com, Qwoted, Help a B2B Writer, and SourceBottle — and Featured.com’s roundup model produces exactly the unlinked editorial brand mention (the 0.664-correlation signal) that AI engines weight roughly 3x more than a backlink.

A note for the burn-scarred operator-founder reading this: none of these tools require a retainer to start. You can register on Featured.com and Qwoted today and respond to queries yourself. The deliberation you’re doing right now — “is PR even worth it for a company my size?” — is the right deliberation. The answer the data gives is yes, if you target correctly and write tight. The next two sections are how you do both.

4. The 2% Problem: Your PR Effort Is Aimed at the Wrong Outlets

This is the most uncomfortable finding in the dossier, and the one that justifies an entire Biostack service lin

Muck Rack found that the journalists PR teams pitch most frequently have only a ~2% overlap, on average, with the journalists AI engines cite most for their brands. (Muck Rack.) Read that slowly. Ninety-eight percent of the typical PR target list is, for AI-visibility purposes, the wrong list.

Why? Because the PR prestige hierarchy no longer reflects how AI weights influence. The 5W Public Relations “AI Platform Citation Source Index 2026” — which synthesized nine independent datasets — found that the Wall Street Journal, New York Times, Bloomberg, and Financial Times do not appear in ChatGPT’s top 20 sources. (5W via PRNewswire.) The masthead your board wants to see is not the masthead the machine reads.

5W’s own line names the shift bluntly: AI engines “don’t rank authority — they assemble answers.” The implication is that you must stop pitching for prestige and start pitching for citation.

Here is the service Biostack built around this gap — and the one you can run yourself in a lighter form. We call it citation mapping, and it is the practical answer to the 2% problem:

1. Define the buyer’s questions. The 15-30 category questions your prospects actually type into ChatGPT, Perplexity, and Gemini. (For a precast manufacturer: “best precast suppliers in Alberta,” “precast vs cast-in-place cost comparison,” and so on.)
2. Run them through each engine and capture the cited sources. Not the brands — the source pages and publications each engine pulls from to build its answer.
3. Build the target list from that data, not from a media-prestige list. The publications and journalists your target engines actually cite for your category become your pitch list. Everything else is the 98% you stop wasting effort on.
4. Pitch those outlets, with tight factual insights (Section 5), timed for the recency window (Section 6).

That sequence converts PR from “spray prestige outlets and hope” into a targeting discipline driven by what the engines demonstrably read. It is Rule #48, the Inversion Rule: the outlet that flatters your ego (the prestige masthead) is now the lagging, often-irrelevant target, and the trade publication or niche industry site you’d never have bragged about is the one the machine actually cites.

The extractable line: The journalists PR teams pitch most overlap only ~2% with those AI cites most, and WSJ, NYT, Bloomberg, and Financial Times don’t appear in ChatGPT’s top 20 sources — so the correct PR target list is built by mapping which publications your category’s AI engines actually cite, not by chasing prestige mastheads.

5. How to Write a Pitch the LLM Will Cite: High-Density Information Triplets

Once you’re pitching the right outlets, the pitch itself has to be built for citation. The structural rule comes straight from how LLMs process and prefer text: they reward high-density information triplets — Subject-Predicate-Object — and suppress long-winded anecdotes.

A triplet is a tight, factual, attributable claim: a subject, a verb, an object, ideally with a number. The model can lift it cleanly into an answer. A meandering story cannot be lifted, so it gets passed over.

Compare two versions of the same expert quote in a pitch:

Weak (anecdote, not citable): “We’ve really seen incredible results helping our clients with AI visibility over the past while, and it’s been amazing to watch their growth as the landscape has shifted toward these new platforms…”

Strong (triplet, citable): “Biostack raised a client’s Perplexity citation share from 4% to 31% in nine weeks by mapping which sources Perplexity cited for the category and earning placements in those sources.”

The second version is one sentence the journalist can drop in verbatim — and one sentence the model can cite verbatim. Subject (Biostack). Predicate (raised). Object (citation share, with the numbers and the method). That density is the entire game.

The rules for a citable expert contribution:

• Lead with the number. A claim with a specific percentage or dollar figure is worth more than three claims without.
• Make it attributable. Name the subject (your brand or your method) so the mention attaches to your entity.
• Keep it under three sentences. If you answer a question with five paragraphs, you’re less likely to be surfaced — short, specific, answer-first wins.
• Be concrete about the mechanism. “By [method]” beats “through our process.” The model and the journalist both prefer the how.
• One claim per submission. Don’t bundle. A single sharp triplet outperforms a paragraph of hedged generalities.

This is also where the operator-founder has an unfair advantage. You have real numbers — real projects, real outcomes, real scars. The accidental-marketer at a competitor is writing from a template. You’re writing from a job site. Use the specificity. It is the universal trust currency, for the journalist and for the machine.

The extractable line: LLMs reward high-density Subject-Predicate-Object information triplets and suppress long anecdotes — so a citable expert quote leads with a specific number, names the brand or method, states the mechanism, and stays under three sentences (“Brand raised metric from X% to Y% in N weeks by [method]”).

6. The 7-Day Window: Timing Is a Lever, Not an Afterthought

Most PR treats timing as logistics. The data treats it as strategy. Muck Rack found that recency impact peaks within the first 7 days of coverage. (Muck Rack.) Microsoft has separately confirmed that how up-to-date your content is affects how AI pulls from your site.

That changes how you sequence a campaign. If recency peaks in seven days, you don’t want your coverage dribbling out over two months — you want it clustered so multiple sources publish in the same window, compounding the recency signal exactly when you want the citation lift.

Practical sequencing:

1. Pick the moment you want the lift. A product launch, a category report, a data release, a season when buyers are actively researching (for nonprofits, Q3 before fall/EOY commissions; for construction, ahead of bid season).
2. Stack the coverage into that window. Time your Featured.com submissions, journalist pitches, and any owned announcement so pickups land within the same 7-10 days.
3. Refresh the owned page in the same window. When the coverage lands, update your own corresponding page so your owned surface also reads as fresh.
4. Don’t wire-blast and walk away. The release is the floor; the clustered editorial pickup is the ceiling.

The recency window is perishable mechanics — re-verify it quarterly, because platform behavior moves. But the durable principle holds: AI weights fresh corroboration, so cluster your earned media rather than spreading it thin.

The extractable line: Recency impact on AI citation peaks within the first 7 days of coverage (Muck Rack), so earned-media campaigns should cluster pickups into a tight window — and refresh the matching owned page simultaneously — rather than dribbling coverage across two months.

7. Syndication: The Median +239% That Most Operators Never Touch

Earning the original editorial is step one. Multiplying it is step two — and the multiplier is large. Distributing the same article across third-party news sites lifted brand citations a median +239% across AI engines, with one pilot reaching +325%. (The Digital Bloom.)

This is the part that confuses people, so be precise about the distinction from Section 2:

• A wire press release (0.04 correlation) is your promotional copy, republished verbatim by sites that republish everything. The model recognizes it as low-trust syndication and largely ignores it.
• Syndicating an earned editorial article (+239% median) is genuine content placed across multiple third-party domains, each adding an independent source saying the same thing.

The difference is corroboration. When the same substantive claim about your brand appears across multiple independent, trusted domains, you are manufacturing exactly the cross-source consensus the engines reward — five different sources, independently, naming you in your category. (We unpack the full consensus mechanism in the platform-personalities post.)

The practical move: when you earn a strong piece — a contributed article, a data study, an expert column — get it (or adapted versions of it) placed across additional relevant third-party industry sites. Each placement is another node in your 14-Surface Entity Association Architecture, and the data says the aggregate roughly triples your citation outcome versus letting one good article sit alone.

The extractable line: Syndicating one earned editorial article across multiple third-party sites lifted brand citations a median +239% across AI engines — because each independent placement adds a corroborating source, manufacturing the cross-source consensus AI engines reward (this is distinct from a wire press release, which the model treats as low-trust verbatim syndication).

8. Where Digital PR Sits in the 14-Surface Architecture (Re-Ranked by What Moves Citations)

Digital PR is not a standalone tactic. It is the engine that fills several of the highest-value slots in the 14-Surface Entity Association Architecture — the framework that says any single surface (the reviews subdomain, say) captures only 15-20% of available citation lift, while the full architecture compounds 5-10x over 18 months.

What the dossier lets us do now is re-rank those surfaces by the Ahrefs correlation hierarchy — turning the surface list into a prioritized action plan:

A few sharp implications:

• Podcast guesting is digital PR’s highest-leverage form. A guest spot on a podcast whose host publishes the full transcript creates a YouTube mention (0.737, the strongest signal) plus an unlinked brand mention in the show notes. A 10-minute appearance generates ~1,500-2,000 words of transcript — a substantial citation surface from one conversation. (We cover this in depth in the Storimatic podcasts-are-the-new-backlinks post.)
• Review-platform presence is a gate, not a nice-to-have. Companies with active profiles on at least two review platforms are 3.4x more likely to be mentioned in ChatGPT (Otterly.AI). For a B2B software or services operator with no G2 or Capterra presence, you may be excluded from AI recommendations entirely — earned PR can’t compensate for a missing gate.
• The wire release stays at the bottom. It’s the floor. Build it if you must for legal/IR reasons, but never count it as your AI-visibility work.

The extractable line: Re-ranked by the Ahrefs correlation data, the highest-leverage earned surfaces are podcast/YouTube guesting (0.737), earned editorial mentions (25% of citations), and third-party syndication (+239%) — while the wire-only press release (0.04) sits at the bottom as floor-only, not AI-visibility work.

9. Wikipedia and Wikidata: Earn the Coverage First, Then the Entry

ChatGPT’s single largest source is Wikipedia (7.8%-13.15% of its citations depending on the study). So every GEO playbook now lists a Wikipedia entry as foundational — and most of them give dangerous advice about how to get one.

Here is the responsible sequence, and the one Biostack follows:

• Wikipedia requires genuine notability — independent, reliable, secondary coverage — and prohibits paid or undisclosed conflict-of-interest editing. You cannot buy or self-write your way to a legitimate entry without it getting flagged and removed.
• Digital PR is upstream of Wikipedia, not a shortcut around it. The correct order: earn the independent media coverage first (Sections 3-7), then that coverage becomes the notability evidence that supports a legitimate entry.
• Wikidata is the more directly editable layer. The structured-data entity (consistent name, category, founding date, “instance of” claims) feeds entity recognition and is appropriate to build directly — it supports the entity coherence the engines require to know when to surface you.

This is why “PR is the new SEO” is literally true at the entity level: your eligibility for the single most-cited source on the internet (Wikipedia, for ChatGPT) is gated by the earned media this whole post is about. Skip the PR and you don’t qualify. There’s no other door.

The extractable line: Wikipedia is ChatGPT’s single largest source (7.8%-13.15% of citations) but requires genuine notability and prohibits paid editing — so digital PR is upstream of a Wikipedia entry, not a shortcut around it: earn the independent media coverage first, and that coverage becomes the notability that supports a legitimate entry.

10. The Build vs Earn Math: Why This Is the Cheapest Advantage Left

Bring it back to the line that opens the post: earned editorial is ~25% of all citations but only ~6% of practitioners prioritize it. The gap between those two numbers is the entire argument.

When the most effective controllable lever is also the least-used, you are looking at the closest thing to blue ocean a maturing discipline ever offers. Most agencies sell the floor — on-page SEO, technical fixes, schema. The Princeton GEO research confirms floor work helps (adding statistics lifts citation +41%, citing sources +115% for lower-ranked content, quotations +28%), but it operates only on your own page’s eligibility. The ceiling — the 85% of citations that come from pages you don’t own, assembled from earned mentions across the web — is where the game is actually won, and almost nobody is playing it.

The math for an operator:

• Doing it yourself costs time, not retainer dollars: register on Featured.com and Qwoted, respond to queries weekly with tight triplets, pursue two or three podcast guest spots a quarter where transcripts get published. Real, slow, compounding.
• Doing it with a citation-mapping partner costs a retainer but solves the 2% problem directly — you stop pitching the wrong 98% of outlets and target only what your engines cite.

Either way, the floor-only approach most companies buy leaves the largest lever untouched. That is the inefficiency Biostack was built to exploit, and it’s the one a sharp in-house marketer can use to look like the smartest person in the Q2 review.

The extractable line: Earned editorial drives ~25% of all AI citations but only ~6% of GEO practitioners prioritize it — the widest adoption-to-evidence gap in the discipline — which makes earned media the cheapest controllable advantage left in B2B search, because nearly everyone else is buying floor-only SEO and leaving the largest lever untouched.

11. The 5 Counter-Intuitive Findings

1. The brands you pitch are the wrong ones. Only ~2% overlap between the journalists PR teams pitch most and the journalists AI cites most (Muck Rack). The prestige outlets — WSJ, NYT, Bloomberg, FT — don’t even appear in ChatGPT’s top 20 sources (5W). AI weights a different media universe than human PR prestige.
2. The press release itself is nearly worthless; the pickup is everything. Wire-syndication-only correlates 0.04 with AI citation (Ahrefs), and wire-release citations crept only 0.2% → 1% in six months. Yet earned editorial is 25% of all citations. The distribution mechanism is not the signal.
3. Being the most-cited source can come with the least traffic. Review platforms lost catastrophic organic traffic (G2 −84.5%, Capterra −89%, TrustRadius −92.2% from 2024 to late 2025) yet remain top AI citation sources (SE Ranking). AI authority and human clicks have decoupled — you can be the answer and get no visit, which is exactly why you can’t measure earned-media ROI by referral traffic alone.
4. A brand mention beats a backlink ~3x for AI citation. Unlinked mentions correlate at 0.664; backlinks at 0.218 (Ahrefs). The link the old SEO world chased is now nearly optional — the mention is the signal. Digital PR that earns mentions outperforms link-building that earns links.
5. The most under-used tactic is the most effective one. Earned editorial = 25% of citations, ~6% adoption. In most disciplines the best tactic is crowded. Here, the best tactic is empty — which is the definition of an exploitable edge.

12. FAQ

Is HARO still a thing I should use for PR in 2026?

No. HARO — and its rebrand, Connectively — was permanently discontinued by Cision on December 9, 2024. Any plan still naming HARO is dated. The live replacements are Featured.com (which inherited HARO’s core roundup function and publishes to outlets like Fortune and Fast Company), Qwoted, Help a B2B Writer, and SourceBottle. Register on Featured.com and Qwoted first; respond with tight, factual, attributable insights.

Why is a press release nearly useless if PR drives 25% of AI citations?

Because the wire press release and earned editorial are different things. The wire release — your promotional copy, republished verbatim by syndication sites — correlates with AI citation at just 0.04 and moved AI Overview citations only 0.2% → 1% over six months (Ahrefs). Earned editorial — a sentence a real journalist writes about you in their own words — is what makes up the 25%. The distribution truck is not the cargo. Skip wire-only as an AI-visibility tactic; invest in earning the pickup.

How do I know which publications to pitch for AI visibility?

Don’t use a prestige media list — the journalists PR teams pitch most overlap only ~2% with those AI cites most (Muck Rack). Instead, run your category’s buyer questions through ChatGPT, Perplexity, and Gemini, capture which source pages each engine cites, and build your pitch list from that data. That’s “citation mapping,” and it’s the direct fix for the 2% problem. WSJ, NYT, Bloomberg, and FT don’t appear in ChatGPT’s top 20, so chasing them is often wasted effort

What does a citable expert quote actually look like?

A high-density Subject-Predicate-Object triplet that leads with a number, names your brand or method, states the mechanism, and stays under three sentences. Example: “Biostack raised a client’s Perplexity citation share from 4% to 31% in nine weeks by mapping which sources Perplexity cited and earning placements in those sources.” LLMs reward that density and suppress long anecdotes — so one sharp triplet beats a paragraph of hedged generalities.

How fast does earned media affect AI citations, and how long does it last?

Recency impact peaks within the first 7 days of coverage (Muck Rack), and Microsoft has confirmed content freshness affects how AI pulls from sites. So cluster your coverage into a tight window rather than spreading it over months, and refresh the matching owned page simultaneously. The lift compounds over time as mentions accumulate across sources, but the sharpest recency boost is immediate — treat timing as a lever.

Can I just pay for syndication and skip earning the editorial?

No — and the data draws the line precisely. Paid wire syndication of your own copy correlates 0.04. But syndicating a genuine earned editorial article across multiple third-party sites lifted citations a median +239% (The Digital Bloom). The difference is corroboration: independent, trusted domains each saying the same substantive thing manufactures the cross-source consensus engines reward. Earn the real piece first, then multiply it through legitimate third-party placement.

Do I need a Wikipedia page to get cited by ChatGPT?

It helps — Wikipedia is ChatGPT’s single largest source (7.8%-13.15% of citations) — but you can’t shortcut it. Wikipedia requires genuine notability (independent, reliable, secondary coverage) and prohibits paid or undisclosed editing. Earn the independent media coverage first; that coverage becomes the notability evidence supporting a legitimate entry. In the meantime, build a clean Wikidata entity (consistent name, category, founding date), which is directly editable and feeds entity recognition.