Every GEO article eventually says some version of the same thing: AI models are trained on the open web, and earned media (press coverage, analyst mentions, journalist citations) carries outsized weight in how those models learn who you are. Almost none of them explain how to actually get any. The advice stops exactly where the useful part starts.
That gap has a real cost, and it's not the cost most people assume. It isn't that you're invisible without press coverage. Plenty of brands build solid AI visibility through reviews, comparison content, and community presence alone. The actual cost is a ceiling: without any third-party editorial coverage, you are permanently dependent on channels you control directly, and AI engines are explicitly built to weight independent sources over brand-controlled ones. You can do everything else in this discipline correctly and still hit a point where the next unit of visibility only comes from someone who doesn't work for you writing about you, on their own judgment, in a publication you don't own.
Here is where PR and GEO actually meet, in concrete terms, without the vague "just get press" advice that's technically true and completely useless.
What Kind of Coverage Actually Feeds AI Visibility
Not all earned media does the same job, and knowing the difference changes where you spend your limited time.
Coverage in outlets AI engines demonstrably crawl and cite. Trade publications in your specific category, established industry blogs, and outlets with high existing citation frequency in AI answers matter more than general business press, even when the general press outlet has a bigger name. A mention in a mid-tier trade publication that AI engines already cite regularly for your category outperforms a mention in a major outlet that rarely shows up in AI-generated answers at all.
Analyst and research firm mentions. Being named in a category report, even a minor one, carries weight because these documents are exactly the kind of structured, authoritative, third-party synthesis that both training data curation and real-time retrieval favor. This is one of the higher-leverage, lower-competition targets in earned media, because most companies your size aren't pursuing analyst relationships at all.
Journalist-cited data or quotes. If a reporter uses your data point or your executive's quote in a piece about your category (not necessarily about you specifically), that's frequently more valuable for AI visibility than a full profile piece about your company, because it embeds your name and your specific claim inside a piece of broader, more authoritative category coverage.
What doesn't move the needle much: pure announcement coverage. Funding round writeups, product launch posts on low-authority outlets, and syndicated press release pickups rarely carry enough independent editorial judgment behind them to register as a strong third-party signal. They're not worthless for other reasons, but don't count on them as your primary GEO earned-media strategy.
Pursuing Coverage Without a PR Budget
Most of the guidance on getting press assumes you have a retainer with an agency. Here's the version for when you don't.
Start with journalists who already cover your category, not journalists at the biggest outlets. Search for who's written about your specific niche in the last six months, follow their beat, and read several of their pieces before reaching out. A specific, relevant pitch to a beat reporter who covers ten companies like yours a month converts far more often than a generic pitch to a name-brand outlet that gets five hundred pitches a day.
Offer something a reporter can actually use, not a story about you. Data from your own product usage, a contrarian but defensible take on a trend in your category, or willingness to go on record with a specific, quotable opinion when a reporter is working on a roundup piece. Reporters need sources more than they need pitches, and being a reliable, quotable source for a beat over months builds more coverage than a single well-crafted pitch email ever does.
Respond to journalist queries actively. Services that connect sources to reporters working on active stories exist specifically for this, and checking them daily costs fifteen minutes and occasionally produces coverage a cold pitch never would have.
Build one real relationship with an analyst covering your category, even informally. A short briefing call, offered without asking for anything in return the first time, is often enough to get considered for future category reports. This costs an hour of your time and nothing else.
Don't overlook podcasts and industry newsletters as earned coverage, even though they feel smaller than traditional press. A well-regarded newsletter in your specific niche, read by exactly the kind of audience that later becomes a Reddit thread or a comparison article, often gets crawled and cited more reliably than a general business outlet with ten times the subscriber count. Guest appearances also tend to be more accessible than press coverage: hosts are frequently short on guests, and a specific, well-reasoned point of view gets you invited back, which turns a single appearance into a recurring source of independent mentions over time.
Publications Worth the Effort, and Ones That Rarely Are
It helps to have a concrete way to sort targets rather than pitching everyone with a "contact" page. Weight your outreach toward outlets where you can already find your competitors being cited inside actual AI-generated answers. If you search your own comparison prompts and notice a specific trade publication or newsletter showing up as a source repeatedly, that's a stronger target than a larger, more prestigious outlet that never appears in those answers at all. Treat the second kind of outlet as a nice-to-have for brand credibility with humans, not a GEO priority. It may still be worth pursuing for other reasons, just don't count it toward your AI visibility goals.
Mistakes That Waste the Effort You Do Put In
A few patterns show up repeatedly among teams running earned media without a dedicated PR function, and each one quietly caps the return on the time invested.
Pitching the story you want to tell instead of the story a reporter needs. A pitch centered on your funding round, your new feature, or your milestone reads as promotional to a reporter, because it is. A pitch centered on a trend the reporter already covers, with you positioned as one useful voice in that trend, reads as a source lead. The difference in response rate between these two framings is not small.
Treating one outlet as the goal. Chasing a single dream placement (a specific major outlet, a specific flagship journalist) means most weeks produce nothing and the effort feels like it's failing. Spreading outreach across a wider list of realistic, category-relevant targets produces a steadier trickle of smaller wins, and those smaller wins are frequently the ones AI engines cite more reliably anyway, since a trade outlet an engine already trusts for your category often outperforms a general outlet it rarely cites at all.
Going quiet after the first coverage lands. One piece of coverage is a data point, not a strategy. The teams that build a real earned media presence keep the same beat reporters and analysts in the loop over months, offering something useful again the next time there's a relevant moment, rather than treating the relationship as closed once a single article runs.
Forgetting to make the coverage work harder afterward. Once a piece runs, link to it from your own site, reference it accurately in your own comparison and about pages, and add it to any press or credibility section you maintain. AI engines that later crawl your own site pick up the reference again, and it reinforces the original signal instead of leaving it stranded on someone else's domain.
How to Tell If It's Working
Earned media doesn't move a visibility tracker the way a schema fix does, so judging it by the same weekly dashboard check sets you up to quit too early.
Track two things instead, on a quarterly rather than weekly cadence. First, the raw count and quality of independent mentions you've earned, sorted by whether the outlet is one AI engines demonstrably cite in your category (check this by searching your category's comparison prompts and noting which outlets show up as sources). Second, whether the language AI engines use to describe your category or positioning shifts to match language that first appeared in a piece of coverage rather than your own marketing copy. That second signal is subtle and slow to appear, but it's the clearest sign the coverage actually reached the training or retrieval layer rather than just sitting on the publisher's site.
If you're three or four months into a consistent earned media push and neither signal has moved at all, the likely issue isn't the channel, it's the target list. Revisit whether the outlets and journalists you're pursuing are ones AI engines actually cite for your category before concluding the channel doesn't work for you.
A Realistic Hit Rate
Set expectations honestly, because this is where most self-run earned media efforts collapse: not from lack of effort, but from expecting agency-level results on a founder-led budget and giving up when that doesn't happen.
A reasonable outcome for a small team running this consistently, one pitch or one piece of proactive outreach a week, is somewhere in the range of one to three pieces of genuine, independent coverage per quarter. Some quarters will produce nothing. Analyst mentions, if you're pursuing them, run on an even slower and less predictable cycle, often six months or more from first contact to inclusion in a published report.
This is not a channel that rewards a single sprint. It rewards showing up consistently, being genuinely useful to the people who write about your category, and treating each small piece of coverage as compounding rather than judging any single pitch as a pass or fail.
The Honest Limitation
Say this plainly, because every other part of GEO has a faster feedback loop than this one: earned media is the slowest channel in the entire discipline, and the payoff often lands quarters out, not weeks.
A schema fix shows results in days. A retrofit shows results in a couple of weeks on the engines that retrieve live content. Earned media, even done well and consistently, is measured in a different unit of time entirely, because you are asking an external party, with their own priorities and their own editorial judgment, to decide your story is worth their time. You cannot compress that timeline by trying harder in any given week, and you cannot buy your way around it without it stopping being earned media at all.
That doesn't make it optional. It makes it the channel you start now and keep running quietly in the background while the faster channels, schema, reviews, comparison content, produce your near-term movement. Treat earned media as the compounding, slow-build layer underneath everything else, not as this quarter's headline initiative, and you'll get more out of it and be less disappointed by how it feels in the first ninety days.
Where to Start
If you're doing none of this today, the single highest-leverage first step is not a pitch. It's a list: the ten journalists and three analysts who actually cover your specific category, built this week, followed for a month before you reach out to any of them. Everything else in this guide compounds off that one piece of homework, and skipping it is the most common reason a first outreach effort goes nowhere.
Put a recurring hour on the calendar, once a week, for this and nothing else. Read the beats you're following, respond to one active journalist query if there's a relevant one, and send one pitch or one piece of useful information to someone on your list. That single hour, kept consistently over a quarter, produces more earned coverage than an intense one-week push followed by three months of silence, which is the pattern most self-run efforts fall into without a standing reminder to keep going.