Earned Media ROI: Measure It Per Outlet, Not Per Clip

Earned media ROI is the return on third-party editorial coverage against the full cost of earning it: the pipeline it influenced, the branded search it lifted and, now, how often AI answers cite the outlets it ran in. Measure it per outlet, not per clip. On the October 20, 2026 Machine Relations Index release, AI engines cited forbes.com in 309 of 14,928 B2B buyer-question runs and bloomberg.com in 4.

That spread is the part of earned media's return you can see before you spend a dollar. Pipeline and branded search are read after a placement runs. The AI-answer term is a property of the outlet, measured in your category's buyer questions, and on that release it differed by a factor of more than fifty between Forbes and the Wall Street Journal, two outlets most target lists put in the same tier.

Key points

• Earned media ROI = (attributable return − total program cost) ÷ total program cost. The hard part is the return, not the arithmetic.
• Earned media value (EMV) is not ROI. EMV prices exposure as if it were advertising. It measures output, and the AMEC Integrated Evaluation Framework exists to move measurement past outputs to outcomes.
• Three terms make up the return for a B2B company: pipeline influence, branded search demand and AI-answer reach. The first two are read in your CRM, Google Analytics and Search Console. The third is read per outlet.
• AI-answer reach is concentrated. In the 14 B2B categories of the October 20, 2026 Index release, techradar.com was cited in 378 of 14,928 runs, wsj.com in 6, and nytimes.com in none.
• The citation rate is an outlet's prior, not a placement's promise. The Index records which domains engines cite, not why, so no placement can be credited with a citation.

What is earned media ROI?

Earned media ROI is the business return created by coverage a third party chose to publish, divided by what it cost to earn that coverage. The cost side is simple: agency or securement fees, internal hours, research and any assets you produced for the story. The return side is where most programs go soft.

The common answer is earned media value: take the audience of each article, price it at an advertising rate, and add it up. It produces a large dollar figure and answers a different question. EMV says what the exposure would have cost as an ad. It says nothing about whether a buyer moved. AMEC's framework puts the standard plainly: "Activity without meaningful outcomes is irrelevant." The Barcelona Principles 4.0, AMEC's 2025 revision of the industry's measurement principles, call for measurement that is meaningful, transparent and outcome-driven.

So the honest return has three terms for a company that sells to other companies:

1. Pipeline influence: opportunities in which the coverage touched the account before or during the deal.
2. Branded search demand: more people searching for your company or product by name after the coverage ran.
3. AI-answer reach: how often AI engines cite the outlet when buyers in your category ask the questions that start a purchase.

The first two belong to your company and are read after the fact. The third belongs to the outlet, and it can be read before you buy.

Why measure earned media ROI per outlet, not per clip?

Because the outlet sets the ceiling on every term of the return, and the outlet is the decision you actually make.

A clip report treats every placement as a unit: ten placements, ten wins. A finance team cannot fund "placements." It funds a plan that names outlets, and each outlet carries its own audience, its own search footprint and its own standing with AI engines. Two features of identical length and quality can return very different amounts because one ran where your buyers and their research tools look, and one did not.

Per-outlet measurement changes three things:

• The budget question becomes concrete. "Is earned media worth it?" turns into "Is a feature in this outlet worth this fee for this category?"
• The comparison becomes fair. A trade title your buyers read every week can beat a famous masthead on every term that matters to you.
• The AI term enters the plan before the spend, instead of being discovered in a quarterly report.

That third shift is new. A public, per-category measure of how often a given outlet shows up inside AI answers now exists, and it is free to read.

What the Machine Relations Index shows about AI-answer reach by outlet

The Machine Relations Index (MRI) counts which sources AI answer engines cite when buyers research a category: best-of questions, how-to-choose questions, is-it-worth-it questions, comparisons and problem-first research.

It is published by Machine Relations Research, the research organization behind Machine Relations, founded by AuthorityTech's founder. AuthorityTech practices the discipline and applies the Index; the Index asks the market's buyer questions, not questions chosen for AuthorityTech's clients.

Release mri_score_v2.1+2026-10-20+bf82d9d98978 covers 28,201 answer runs from six engines (ChatGPT, Claude, Gemini, Google AI Mode, Google AI Overviews and Perplexity) between May 10 and October 20, 2026. A run is one question asked of one engine on one day, and a domain counts once per run however many of its pages the answer cites. The 14 categories whose buyer is a company account for 14,928 of those runs.

In those 14,928 B2B runs, AI-answer reach differed by outlet nearly a hundredfold. The table lists the three most-cited classified editorial publications, the two most-cited wire services, and eight widely known business and technology outlets we chose for contrast.

Outlet	B2B runs citing it (of 14,928)	Share of B2B runs	All runs citing it (of 28,201)	Engines that cited it, whole release (of 6)
TechRadar	378	2.53%	792	6
Forbes	309	2.07%	1,076	6
TechTarget	259	1.73%	367	6
PR Newswire	124	0.83%	232	6
Axios	79	0.53%	165	6
Business Wire	79	0.53%	135	6
TechCrunch	66	0.44%	172	5
CNBC	25	0.17%	268	6
Business Insider	21	0.14%	171	5
Fortune	7	0.05%	122	5
The Wall Street Journal	6	0.04%	120	5
Bloomberg	4	0.03%	15	3
The New York Times	0	0.00%	294	3

Read it as a buyer would. TechRadar appeared in 378 B2B runs and Bloomberg in 4. Forbes appeared in 309 and the Wall Street Journal in 6. CNBC and the New York Times are cited often across the whole release, mostly outside the B2B categories, and rarely or never when the buyer is a company. Our analysis of paywalled publications breaks that pattern down by category and engine.

The per-outlet view cuts across tiers in both directions. prnewswire.com appeared in 124 B2B runs, more than wsj.com, bloomberg.com and fortune.com together. Yet as a class, the 36 domains the Index treats as wire and press-release distribution were cited in 357 domain-runs across the B2B categories, against 5,675 for the 1,039 classified editorial publications. Both are floors: 60,803 of the 95,630 domain-run citations in those categories (63.6%) sit with domains the Index has not yet classified. A class average hides the outlet, so the outlet is the unit.

Concentration runs through the editorial class too. 544 of the 1,039 classified editorial publications were cited at least once in the B2B runs; the other 495 were not cited in any of them. Most outlets on a long media list carried no measurable AI-answer reach for a company buyer in this window.

Your category decides the list

The B2B total is an average across 14 categories, and buyers do not shop in averages. Take Enterprise Software, 1,642 runs on the same release. The most-cited domain there was erpresearch.com, a market and company database, in 193 runs. The best-placed editorial publications were TechTarget (71 runs, #8 in the category), TechRadar (68, #10), Forbes (66, #11) and SaaStr (48, #20). Other categories have other leaders.

That is why the AI-answer term of earned media ROI is read in your category, not from a global ranking. The publications index shows each outlet's live citation standing from the latest Index release, and the buyer-category pages, such as Enterprise Software buyers, show which outlets lead each question type. Check the outlet before it goes on the plan, not after the placement runs.

How to measure earned media ROI, step by step

1. Fix the outcome and the questions before the first pitch

Pick one primary commercial outcome: qualified pipeline in a named segment, branded demand, or presence in the AI answers for a named set of buyer questions. Then write down 20 to 50 questions your buyers actually ask when they start looking. These become your AI-answer test set and your outlet filter.

2. Score each candidate outlet on three terms

Term	What it tells you	Where to read it	When
Buyer audience	whether the people who sign read it	your sales team, win-loss notes, the outlet's own audience data	before you buy
AI-answer reach	how often engines cite the outlet in your category's buyer questions	the publications index and the MRI category pages	before you buy
Search footprint	whether its pages rank and are linked for your topics	your SEO tools	before you buy

An outlet that scores on all three earns the budget first. An outlet that scores only on prestige may still be right for an investor, recruiting or reputation job; fund it for that job and measure it on that job's terms.

3. Track pipeline influence honestly

Earned media rarely wins the last click. Google Analytics documents one reason its reports can undercount it: its attribution models exclude direct visits from credit unless the whole path to a key event was direct, and a reader who sees your company in an article often arrives later by typing your name. Three instruments cover the gap:

• Influence tagging in the CRM: mark opportunities whose contacts engaged with the coverage or whose accounts were in the outlet's audience during the deal.
• Self-reported attribution: a free-text "where did you hear about us" field on demo and contact forms.
• Sales notes: prospects naming the article or the outlet on calls.

Report influenced pipeline as influenced, not sourced. Finance will trust a modest honest number over a large modeled one.

4. Track branded search demand

In the Search Console Performance report, filter queries containing your brand and product names and compare impressions and clicks in equal windows before and after each placement. A placement in an outlet your buyers read should show up here before it shows up in pipeline.

5. Track AI-answer presence on a fixed set of questions

Run your question set across the major engines on a schedule. For each answer, record whether it names your company and which domains it cites. Over a quarter this tells you whether the outlets you funded are the ones engines draw on for your questions.

Keep the claim straight: the Index and your own runs show where engines looked, not what a placement caused. A rising presence after a placement is evidence worth reporting, and it is not proof the placement produced it.

What the AI-answer term cannot tell you

An outlet's citation rate is the best available prior for AI-answer reach, and it has limits worth stating to a finance team:

• It is domain-level. The forbes.com count covers every Forbes section, staff journalism and contributor content alike. The Index does not record which page or section was cited.
• It is a window, not a forecast. The figures describe May 10 to October 20, 2026. Engines change their sources; the publications pages update with each release.
• It is uneven by engine. The release marks Gemini coverage as degraded, and Google AI Mode and AI Overviews ran fewer questions than the other engines.
• It is not cost efficiency or brand benefit. A citation share says where engines look. It does not say what a placement is worth, what it costs, or whether it helps a particular brand.

None of this weakens the case for measuring per outlet. It sharpens it: the term you can see before you buy is the one most plans never look at.

How to fund earned media with this

Make it a forcing choice per outlet. Either the outlet earns its place on the AI-answer term for your category's questions, or it earns its place on a named human job: investors, recruits, a partner, a regulator. If it earns neither, it is a clip, and clips are not a return.

The AI Search Arena study, a 2025 arXiv preprint analyzing more than 366,000 citations from OpenAI, Perplexity and Google models, found that news citations "concentrate heavily among a small number of outlets." The Index shows the same concentration inside B2B buyer questions, outlet by outlet. A plan built on that concentration funds fewer, better-chosen outlets and measures each one.

AuthorityTech secures earned placements in the publications engines cite. It has no rate card: each placement carries its own securement fee per publication, held in escrow and released only when the placement is published. For the operating playbook that turns placements into AI citations, read PR for AI search; for the research behind earned media and AI answers, read what the research actually shows.

Frequently asked questions

What is earned media ROI?

Earned media ROI is the return on third-party editorial coverage divided by the full cost of earning it. For a B2B company the return has three measurable terms: influenced pipeline, branded search demand, and AI-answer reach, which is how often AI engines cite the outlets the coverage ran in.

How do you calculate earned media ROI?

Use (attributable return − total program cost) ÷ total program cost. Count fees, internal time and assets as cost. Count influenced pipeline at its expected value as return, report branded search and AI-answer presence alongside it, and label each term as influenced rather than sourced.

Is earned media value the same as earned media ROI?

No. Earned media value prices the exposure of an article as if it were advertising. It measures output, not outcome. AMEC's Integrated Evaluation Framework and the Barcelona Principles 4.0 both call for measurement tied to outcomes.

Which outlets give the best AI-answer return for a B2B company?

It depends on the category. Across the 14 B2B categories of the October 20, 2026 Machine Relations Index release, techradar.com was cited in 378, forbes.com in 309 and techtarget.com in 259 of 14,928 answer runs, while wsj.com was cited in 6 and bloomberg.com in 4. Check your own category on the publications index before you fund an outlet.

Does an earned media placement cause AI citations?

Citation counts cannot show that. The Machine Relations Index records which domains AI engines cite, not why they cite them. An outlet's citation rate is a prior for where engines look in a category, not a promise for any single placement.

How does AuthorityTech charge for earned media?

AuthorityTech has no rate card. Each placement carries its own securement fee per publication, held in escrow and released only when the placement is published.

Method note. All Index figures are from the public artifact for release mri_score_v2.1+2026-10-20+bf82d9d98978 (window May 10 to October 20, 2026, 157 days; 28,201 answer runs after the release's own exclusion of 948 runs from 81 engine-days that returned no eligible citation). The 14 B2B categories are AI Infrastructure, AI Security & Privacy, AI Visibility & GEO, Cybersecurity, Deep Tech & Hardware, Enterprise Software, Fintech, HR & Talent, Industrial, Logistics & Freight, Martech & Advertising, Professional Services, Sales, GTM & CRM, and VC & Private Equity, a grouping this page makes and the Index does not publish. B2B run counts are summed from each domain's per-segment rows; class totals sum those rows by the Index's source role. A zero means no B2B run cited the domain in this window. Engine breadth is the whole-release count of engines that cited the domain at least once. Re-query the release on the Index pages for forbes.com, techradar.com, nytimes.com and Enterprise Software, in the public JSON artifact and in the release manifest.

Sources

• Machine Relations Index, release mri_score_v2.1+2026-10-20+bf82d9d98978
• Machine Relations Index public JSON artifact and release manifest
• AMEC Integrated Evaluation Framework
• AMEC Barcelona Principles 4.0
• Google Analytics Help: attribution
• Search Console Help: Performance report
• News Source Citing Patterns in AI Search Systems, arXiv:2507.05301
• AuthorityTech publications index