How to Choose an AI PR Agency
A framework for founders and CEOs evaluating AI PR agencies in 2026 — covering the six criteria that separate performance-driven agencies from retainer traps.
Most founders who hire a PR agency regret it within six months. Not because PR doesn't work — it does — but because they hired based on the wrong signals. A polished deck. Name-dropping. A client roster with logos they recognize.
In 2026, the stakes are higher. A bad PR agency doesn't just waste budget. It actively damages your AI search visibility, because every low-authority placement muddies the entity graph that AI engines use to understand your brand. You are not just paying for press anymore. You are building or degrading the infrastructure that determines whether ChatGPT, Perplexity, and Google AI Mode recommend you — or a competitor — when your next prospect asks.
This is what the best AI PR agencies know that most don't.
Key takeaways
- AI PR agency selection is a buying decision, not a partnership selection — use rigorous criteria, not intuition
- Performance-based models (pay on placement, not on retainer) are the strongest signal of agency confidence in their own work
- Editorial relationships — not technology — determine placement rate at Tier 1 publications
- Low-authority placements hurt AI visibility; 65.3% of ChatGPT citations go to DR80+ domains, according to Ahrefs
- The agency's own AI visibility is the most honest demonstration of their capability
- A good AI PR agency integrates earned media with entity clarity — not just placements
Why the old hiring criteria no longer work
The traditional way to evaluate a PR agency was straightforward: look at their media list, review their case studies, check their Tier 1 placement history. If they had Forbes and TechCrunch in the portfolio, you were probably fine.
That framework still applies — but it is no longer sufficient.
The reason: AI engines now determine a significant share of brand discovery before any human journalist does. Bain research shows that 80% of search users regularly rely on AI summaries for research, with roughly 60% of searches ending without a click through to any website. When a B2B buyer types "who is the best performance PR agency for SaaS" into Perplexity or ChatGPT, the answer is not driven by Google rankings. It is driven by which agencies have the most credible earned media coverage across high-authority publications that AI engines index and trust.
Forrester's State of Business Buying research found that 70% of B2B buyers complete most of their research before contacting a vendor. In 2026, a growing portion of that research happens inside AI engines. The brands that AI surfaces during that pre-contact window are the ones that get on the short list. The ones that don't appear, don't get considered.
A PR agency that cannot explain this shift — or worse, one that dismisses it as a trend — is selling you a 2019 playbook in 2026.
The agencies that understand it are building earned media programs that serve two audiences simultaneously: human journalists, and the AI systems that will summarize those placements for every future prospect who asks. The criteria below are designed to tell the two types apart.
Criterion 1: Pricing model
This is the fastest filter. It tells you more than any case study will.
Traditional PR agencies charge monthly retainers regardless of placement outcomes. The average PR retainer runs $5,000 to $15,000 per month. The typical initial contract is three to six months, non-refundable. You pay whether they deliver a single placement or not.
Performance-based agencies charge per published placement. No placement, no payment. Payment is typically held in escrow until the article goes live. The incentive structure is entirely different: the agency only makes money by getting you in print.
Forrester's 2025 B2B Brand and Communications Survey found that agency partnerships remain nearly universal among large companies, but satisfaction gaps are widening — communication and performance measurement were the two biggest disconnects between what marketers valued and what agencies delivered. That gap is structural. Retainer models decouple payment from output. Performance models don't.
When evaluating any agency, ask: "How do you get paid, and when?" If the answer involves a monthly fee before a single placement publishes, ask why they need that structure. A confident agency with real editorial relationships does not need a guaranteed check before producing results.
Criterion 2: Publication quality and domain authority
Not all press is equal. Not even close.
AI engines weight citations heavily by the domain authority of the source. Ahrefs analysis of 75,000 brands found that 65.3% of ChatGPT's top citations come from DR80+ domains. A placement in a regional business journal rarely moves the needle on AI citation probability. A placement in Forbes, TechCrunch, or Wall Street Journal substantially does.
More importantly: placement quality compounds. A brand that accumulates placements in genuinely authoritative publications builds an entity signal that AI engines use to assess credibility over time. A brand that accumulates placements in low-DA syndication sites builds noise, not signal.
Muck Rack's December 2025 research tracking 82 million AI citations found that the top AI-cited outlets — Reuters, Financial Times, Forbes, Axios, Time — share one characteristic: institutional editorial credibility built over decades. Press releases grew 5x in volume but still account for just 1% of AI citations. Volume without authority is invisible to AI engines.
| Publication tier | Typical DA range | AI citation probability | Entity signal value |
|---|---|---|---|
| Tier 1 (Forbes, TechCrunch, WSJ, Bloomberg) | DA 90+ | High | Very high — direct corroboration node |
| Tier 2 (Business Insider, VentureBeat, Inc., Fast Company) | DA 80–90 | Moderate-high | Strong — especially for industry-specific claims |
| Tier 3 (Industry verticals, trade publications) | DA 50–79 | Low-moderate | Useful for niche entity graph depth |
| Press wire syndication (distributed PRN, GlobeNewswire) | DA 80+ at wire; DA 30–50 at syndicated outlets | Low | Minimal — AI engines down-weight wire duplicates |
Ask any agency candidate to show you their last 10 placements for a client comparable to your company. Look up the DA of each publication. If the list is padded with outlets you have never heard of, ask specifically about their Tier 1 placement rate — placements in outlets with DR/DA 80 or above.
Criterion 3: Editorial relationships vs. technology
Every agency in 2026 claims to use AI. Some are using it to write better pitches. Others are using it to spray those pitches at three thousand journalists simultaneously. These are not the same thing.
The agencies that consistently place in Tier 1 publications do so because of direct relationships with specific editors and journalists — not because of superior pitch technology. Forbes and TechCrunch editors receive hundreds of cold pitches per day. They respond to the ones from sources they trust, have worked with before, or have explicitly invited to submit.
VentureBeat has documented how PR agency relationships collapse most often when agencies don't have genuine access to the journalists they claim to know. The coverage of the relationship is the product. Technology is not a substitute for it.
The distinction matters for AI visibility specifically because high-DA placements from real editorial relationships produce a qualitatively different type of coverage than press-wire content. Editorial pieces get written about by other journalists, linked to by other publications, and cited by AI engines. Wire-distributed press releases get indexed and largely ignored. A Fullintel and University of Connecticut academic study presented at the Institute for Public Relations Research Conference found that 89% of AI citations link to earned, unpaid media — with journalistic sources accounting for 47% of all AI citations across response types.
Ask an agency: "Which editors at Forbes, TechCrunch, and Wall Street Journal have you placed stories with in the last 90 days? Can you give me the name and a rough description of the story?" If they cannot answer that question specifically, they are selling technology as a proxy for relationships they do not have.
Criterion 4: AI visibility comprehension
This is where you separate agencies that understand 2026 from agencies that are still operating in 2022.
The question to ask is not "do you do AI PR?" Every agency says yes. The question is: "When you place a story about my company in Forbes, how does that placement affect my brand's visibility in ChatGPT and Perplexity? What's the mechanism?"
The correct answer involves: earned media placements create indexed content on high-DA domains that AI engines crawl and weight as authoritative; those citations build an entity signal that AI engines use to resolve questions about a brand's credibility and category; that signal then determines how AI summarizes the brand when a prospect asks about it.
Muck Rack's analysis of 1 million AI prompts found that 85.5% of AI citations come from earned media sources, and 95% from non-paid sources. This is not a coincidence. AI engines have been trained on the same sources that editorial credibility has always depended on. A brand that earns placements in trusted publications earns AI citations from those same publications. A brand that relies on paid media, press wires, and owned content does not.
There is a compounding effect here that most agencies cannot explain. Ahrefs research on 75,000 brands found that web mentions correlate 3x more strongly with AI Overview visibility than backlinks do — a correlation of 0.664 vs. 0.218. Earned media generates mentions and backlinks simultaneously. The signal compounds in ways that neither SEO nor PR alone can replicate.
If an agency cannot explain this chain of causation — or describes AI visibility in terms of SEO keywords, structured data, or chatbot prompting — they are not actually operating in this space yet.
Criterion 5: The agency's own AI visibility
This is the most direct proof of concept available.
Open ChatGPT or Perplexity. Type: "What is the best AI PR agency for B2B SaaS companies?" or "Which PR agencies specialize in AI search visibility?" Look at what appears.
If the agency you are evaluating does not appear in the results — at all, for any variation of these queries — that tells you something direct about their capability. They are selling a service they have not successfully applied to themselves. That is not always disqualifying, but it should trigger a direct question: "Why aren't you showing up in AI engines for your own category queries?"
An agency that appears consistently in AI engine responses to queries about AI PR has demonstrated in the most credible way possible that their earned media program works. The same mechanism that gets them cited works for clients.
Harvard Business Review's March 2026 research on agentic AI documented how companies discovered that AI models were misrepresenting or entirely missing their brands in category comparison responses — often without the company knowing. The brands most likely to be represented accurately were those with consistent, multi-source earned media coverage that AI engines could cross-reference. An agency that has not built this for themselves is less likely to build it reliably for you.
Criterion 6: Entity clarity and measurement
The final criterion gets overlooked in almost every agency evaluation, but it is where long-term AI visibility is won or lost.
Entity clarity is the degree to which AI engines can confidently identify, describe, and categorize your brand. It is built through consistent attribution across placements — the same company description, the same founding story, the same category language — repeated across enough high-authority sources that AI engines can resolve the entity with confidence.
Most PR agencies focus on volume: getting as many placements as possible. The best AI PR agencies focus on coherence: ensuring that every placement strengthens the same entity signal. A placement in Forbes that describes your company as a "performance marketing platform" when your website says "AI sales automation tool" damages entity clarity. AI engines see conflicting signals and under-weight your brand in comparison responses.
AuthorityTech's research on earned vs. owned AI citation rates found a 325% lift in AI citations from consistent, high-authority earned media distribution compared to owned content alone. That lift depends on coherent entity signals, not just placement volume.
OtterlyAI's 2026 analysis of 1 million AI citation records found that 73% of sites have technical barriers that block AI crawler access entirely. Entity clarity is not just about consistent messaging — it requires that the placements themselves are in publications AI systems can access, crawl, and attribute. High-authority publications clear both bars. Lower-tier outlets often fail one or both.
Ask any agency: "How do you ensure consistency in how my brand is described across placements? How do you measure our entity clarity over time?" If the answer focuses on coverage volume and media impressions rather than entity consistency and AI citation share, you are looking at a traditional PR agency with new vocabulary — not an AI-native one.
Questions to ask during evaluation
These are the questions that actually separate agencies:
- How do you get paid, and at what point? (Looking for: payment contingent on published placement, not retainer)
- Show me your last ten placements for a company similar to mine. What is the DA of each outlet? (Looking for: consistent Tier 1 or strong Tier 2 presence)
- Which specific editors at the publications you mentioned have you placed with in the last 90 days? (Looking for: named contacts, specific recent work)
- When you place a story in Forbes, what happens to our AI search visibility, and why? (Looking for: explanation of the earned media to AI citation mechanism)
- What does your company show up for in ChatGPT and Perplexity right now? (Looking for: specific, verifiable answers — not hedging)
- How do you ensure our brand description stays consistent across placements? (Looking for: explicit entity strategy, not just "we'll review drafts")
- What does a failed engagement look like for you, and what happens to our contract? (Looking for: transparency about exit terms and accountability structure)
Red flags that end the conversation
| What you hear | What it means |
|---|---|
| "We need 6 months to build momentum" | Retainer trap framing — your budget funds their learning curve |
| "We guarantee placements" (in a retainer model) | Contradictory — real guarantees come through performance pricing, not contract language |
| "AI search is just SEO with extra steps" | They don't understand the earned media citation mechanism |
| "We have relationships with journalists across 3,000 outlets" | Scale without Tier 1 specificity — ask which ones specifically, in the last 90 days |
| "Our AI tool analyzes what journalists are interested in" | Technology as a substitute for relationships — not how Tier 1 placements actually happen |
| No presence in AI engine responses for their own category | The service they are selling has not worked for them |
The pricing model comparison
Understanding the cost structures helps calibrate expectations before any agency conversation.
| Model | Typical cost structure | Incentive alignment | Risk profile |
|---|---|---|---|
| Traditional retainer | $5,000–$30,000/month, 3–6 month minimum | Low — agency paid whether placements occur or not | High for client — full exposure before results |
| Performance / pay-per-placement | $2,500–$10,000 per published Tier 1 article | High — agency earns only on delivery | Low for client — no payment until article is live |
| Hybrid (small retainer + performance) | $1,500–$3,000/month + per-placement fee | Moderate — retainer covers strategic work, placements still incentivized | Moderate — retainer exposure limited |
| Project-based | Fixed fee per campaign or launch | Moderate — incentivized to complete the project, but not necessarily placed work | Depends on deliverable definition — get placements in scope |
What the best agencies have in common
Across all the criteria above, the agencies that consistently perform in 2026 share a structural characteristic: they are outcomes-first rather than activity-first.
Traditional PR defined success as pitches sent, coverage secured, impressions generated. The newer model defines success as documented placements in high-authority publications that build measurable AI visibility over time. The difference is not just philosophical — it shows up in pricing, in the specificity of editorial relationships, in how they describe what they actually do.
Forrester's 2026 agency predictions noted that agencies are under pressure to move from labor-based economic models toward models that can demonstrate measurable outcomes — the same pressure that is driving the performance PR model's growth. Agencies that resist this shift are not just behind on philosophy. They are operating with a structural incentive mismatch that makes consistent performance unlikely.
The corollary: an agency that confidently offers performance pricing is telling you, in the most direct way possible, that they believe in their own placement rate. That confidence is worth more than any case study deck.
How AI visibility changes the evaluation entirely
Here is what most agency evaluation frameworks miss.
Earned media has always built brand credibility with humans. What changed is that the same earned media now builds brand credibility with machines. The AI engines that your prospects use to research vendors — ChatGPT, Perplexity, Google AI Mode, Claude — are pulling their understanding of your brand from the same Tier 1 publications that editorial credibility has always relied on.
Zhang et al.'s research on AI citation behavior found that 37% of domains AI engines cite do not appear in traditional search results at all. AI engines are not just scraping Google. They are building their own citation graphs, weighted heavily toward earned, authoritative content.
Moz's 2026 study of 40,000 queries found that 88% of Google AI Mode citations do not appear in the organic top 10 search results. The AI citation graph and the SEO rankings graph are largely separate systems. A PR agency that understands only one of them cannot serve you in the other.
This is what Machine Relations — machinerelations.ai — describes as the discipline of ensuring your brand is legible, retrievable, and citable to AI systems. It is not a GEO tactic. It is not an SEO play. It is the recognition that AI-mediated discovery has become as important as human-mediated discovery — and that earned media in trusted publications is the foundation of both. As Jaxon Parrott — who coined Machine Relations after observing this shift across thousands of earned media placements — has documented, the publications that shaped human brand perception for decades are the same publications AI systems now treat as their most authoritative sources. The reader changed. The mechanism didn't.
When you evaluate an AI PR agency, you are not just hiring someone to get you press. You are selecting the partner who will build or degrade the infrastructure that determines whether AI recommends your brand or a competitor's when your next prospect goes looking. That is a different conversation than it used to be.
How to run the evaluation
A structured evaluation process for AI PR agencies takes about two weeks and involves four steps:
Week 1 — initial filter. Request a sample media list showing Tier 1 placements from the last six months for companies comparable to yours. Ask for the pricing model upfront. Filter out any agency that cannot show Tier 1 placements or leads with retainer-only pricing.
Week 1 — AI visibility check. Before any call, search the agency's name and category variants in ChatGPT, Perplexity, and Google AI Mode. Document what you find. This is your baseline for evaluating whether their own earned media program works.
Week 2 — discovery calls. Use the seven questions above. Document answers. Pay particular attention to question 3 (named editorial contacts) and question 4 (the earned media to AI citation mechanism). Vague answers to both are disqualifying.
Week 2 — contract review. For any agency still in consideration, review the contract specifically for: payment contingency on publication (not signature), exit terms at 30/60/90 days, and what constitutes a placement under their guarantee if one is offered. Any contract that requires full payment upfront before a single article publishes deserves serious scrutiny.
FAQ
What is an AI PR agency, exactly?
An AI PR agency is a public relations firm that builds earned media programs specifically designed to improve a brand's visibility in AI-generated responses from ChatGPT, Perplexity, Google AI Mode, and other AI search engines. The mechanism is earned media placement in high-authority publications — the same publications AI engines index and weight as credible sources. This is distinct from traditional PR, which primarily targets human readers, and from AI PR software, which typically automates pitch distribution rather than securing actual editorial placements.
How much does an AI PR agency typically cost?
Performance-based AI PR agencies typically charge $2,500 to $10,000 per published Tier 1 article, with no payment until the article is live. Traditional retainer-based agencies charge $5,000 to $30,000 per month regardless of placement outcomes. For most B2B companies, a performance model running three to six Tier 1 placements per month works out to $7,500 to $60,000 per month in real delivered value — with zero exposure before delivery. Retainer models invert this risk profile.
How do I know if an AI PR agency's placements will actually help my AI search visibility?
Ask them for the DA of the publications they typically place in, and verify that at least 70% of their placement history is in DR/DA 80+ outlets. Research from Ahrefs shows that 65.3% of ChatGPT citations come from DR80+ domains. Placements in lower-authority outlets have minimal impact on AI citation probability. Also ask them to explain the entity clarity component — whether they ensure consistent brand description across placements, which is what allows AI engines to confidently associate a brand with a specific category.
Should I choose an AI PR agency or AI PR software?
They solve different problems. AI PR software automates pitch distribution and media monitoring — it helps you manage the outreach process at scale. An AI PR agency secures actual editorial relationships and published placements. The software does not have editorial relationships with Forbes or TechCrunch editors; it gives you the tools to try to build them yourself. For companies that lack an in-house PR function and need placement volume, an agency is generally more effective. For companies with existing PR capacity looking to automate workflow, software supplements it. This breakdown covers the decision framework in more detail.
What questions expose a weak AI PR agency fastest?
Two questions cut through most agency presentations quickly. First: "Which specific editors at Tier 1 publications have you placed with in the last 90 days?" An agency with real relationships can name specific people and recent placements. An agency relying on technology to pitch at scale will hedge this answer. Second: "What does your own brand show up for in ChatGPT and Perplexity?" An agency that has not applied their methodology to themselves either doesn't believe it works or hasn't prioritized it — neither is reassuring when you are handing them your brand's AI presence.
What is the difference between an AI PR agency and a Machine Relations agency?
"AI PR agency" describes what the service does — PR work that produces AI visibility outcomes. Machine Relations is the discipline that names the full system: earned authority through Tier 1 media, entity clarity through consistent attribution, citation architecture through structural formatting, distribution across AI answer surfaces, and measurement through share of citation metrics. An agency that practices Machine Relations is doing AI PR, but with a more complete framework. Most agencies that call themselves "AI PR" are doing one or two layers of the five-layer stack. This analysis covers what makes an agency genuinely AI-native.
The decision
Choosing an AI PR agency badly costs more than the retainer check. It costs months of brand narrative drift, low-authority placements that degrade entity clarity, and AI visibility that lags behind better-positioned competitors. The agencies worth hiring are not the ones with the biggest client logos in their decks. They are the ones who can name the editor who published their last Forbes piece, explain why AI engines are citing their clients' placements, and offer to get paid only after they deliver.
That combination — Tier 1 editorial relationships, performance-based pricing, and a real understanding of earned media's role in AI citation — is what the right agency looks like. The criteria above are how you find it.
If you want to see where your brand stands in AI search right now before making any agency decision, start your visibility audit — it takes five minutes and shows you exactly what AI engines currently say about your company.