Performance-Based PR: 4 Attribution Models That Prove Earned Media ROI in 2026
Performance-based PR requires real attribution, not clip counting. Jaxon Parrott, founder of AuthorityTech and the person who coined Machine Relations, breaks down 4 measurement models — AI citation tracking, multi-touch media attribution, source authority scoring, and search-lift attribution — that make results-based earned media contracts enforceable in 2026.
I built AuthorityTech around a single bet: if you can prove what earned media actually produces, the retainer model dies. That bet is now operational. The discipline I coined — Machine Relations — gives performance-based PR the measurement infrastructure it never had: real-time tracking of which placements get cited by AI engines when buyers ask questions.
Performance-based PR only works when both sides agree on what "results" means. For decades, that agreement was impossible because nobody could connect a placement to a business outcome. Muck Rack's May 2026 study of 25 million+ links across ChatGPT, Claude, and Gemini shows earned media accounts for 84% of all AI citations — consistent between 82% and 89% across three editions of their study going back to July 2025. Paid and advertorial content represents just 0.3%. The attribution problem is solved. The question is which models to use.
Why PR Attribution Has Been Broken Until Now
PR measurement ran on proxies for 70 years. Clip counts. Circulation numbers. Advertising Value Equivalency. Each one told you something happened without telling you what it produced.
Forrester's 2024 research found that 64% of B2B marketing leaders say their organization does not trust its own measurement for decision-making. That number includes the PR function. When your CMO cannot tell the board what a placement produced, the retainer model survives by default — not because it proves value, but because nobody demands proof.
The Trade Press AI Index, a joint audit by 5W and Everything-PR covering 680 million AI citations across 9 industries and 5 AI engines, confirms the structural problem: the top 15 domains control 68% of all AI citation share. Wikipedia alone accounts for 47.9% of ChatGPT's top-10 sources. Most PR agency media lists target outlets that do not appear in those citation rankings at all.
Performance-based pricing breaks the retainer dynamic by requiring a shared scoreboard. The agency must define what counts as a result before work begins, and the client must agree to pay when those results materialize. Without attribution models sophisticated enough to track earned media from placement to pipeline, neither side can enforce the contract. I track four models that make it work.
4 Attribution Models That Make Performance PR Measurable
1. AI Citation Attribution
This is the model that changed the conversation — and the one I built AuthorityTech's Machine Relations framework around. AI citation attribution tracks whether a specific placement gets retrieved and cited by ChatGPT, Perplexity, Gemini, Google AI Overviews, or Claude when a buyer asks a relevant query.
Muck Rack's May 2026 data shows 96% of ChatGPT responses include sources (averaging 5 citations per response), 82% of Gemini responses include sources (averaging 8 citations), and 55% of Claude responses include sources (averaging 13 citations when provided). The citation pool is deep and measurable.
An Ahrefs study of 75,000 brands found that branded web mentions correlate 3x more strongly with AI visibility than backlinks. When your agency places a feature in TechCrunch and that feature gets cited by Perplexity in response to "best AI visibility tools for B2B," you have a measurable, timestamped attribution event. That is a result you can pay for.
Machine Relations provides the measurement layer: I track citation architecture — the structural condition where a brand's claims appear as sources in AI-generated answers across multiple engines — across ChatGPT, Perplexity, Claude, Gemini, and Google AI Mode. Instead of measuring clip counts, I measure whether earned media placements enter the citation pool AI engines use when buyers ask questions.
2. Multi-Touch Media Attribution
This model connects earned media touchpoints to pipeline actions — branded search spikes within 48 hours of publication, referral traffic from the outlet, demo requests that cite the coverage in their intake form, and LinkedIn engagement patterns tied to the placement.
Agility PR's 2026 framework describes this well: one article can trigger several measurable reactions within hours. Branded search spikes. Referral traffic jumps. A founder quote gets shared across LinkedIn. Someone clicks through and signs up for an email list.
The key is connecting those signals to the specific placement, not to "PR generally." Multi-touch attribution tracks the sequence: coverage publishes, branded search volume increases 40%, three inbound demo requests cite the article, pipeline moves. That chain is priceable.
3. Source Authority Scoring
Not every placement is equally retrievable by AI engines. Source authority scoring evaluates the structural quality and extractability of each outlet and placement before the pitch goes out.
Research from Wrodium (arxiv, 2025) showed that pages with quality scores above 0.70 and 12 or more structural signals achieve a 78% cross-engine citation rate. The Trade Press AI Index backs this up: Claude favors long-form journalism from the NYT, Atlantic, Economist, and New Yorker, while LinkedIn captures 15% of Google AI Mode citations. Each engine has a different source trust profile.
A performance agency that uses source authority scoring can tell you before the pitch which outlets will produce AI-retrievable coverage and which will not. That changes the pitch list from a volume exercise to a precision instrument.
4. Search-Lift Attribution
Search-lift attribution measures the organic search impact of earned media within a 7-to-14-day window after publication. When a Forbes feature on your company publishes, does your branded query volume increase? Do your target pages move up in position? Does the feature itself rank for relevant queries?
This model predates AI attribution but remains essential. AI-sourced traffic to U.S. retail sites increased 3,500% between July 2024 and May 2025 (Adobe Analytics via VentureBeat). That traffic still routes through search infrastructure, which means search-lift attribution captures signals that AI citation attribution alone misses.
How Machine Relations Changed the Scoreboard
The reason performance-based PR contracts are viable now — and were not five years ago — is that Machine Relations created a shared, verifiable record of what earned media actually produces.
Before 2025, a PR agency could claim a placement "generated awareness" and neither side could prove or disprove it. Now, you can query the same AI engine a buyer would use, ask the same question a buyer would ask, and verify whether your placement appears in the answer. That is a binary event: cited or not cited, attributed or not attributed.
Brands are 6.5x more likely to be cited by AI engines through third-party earned media than through their own domains (Superlines, 2026). This means the PR agency's work — securing third-party placements — is the primary driver of AI visibility. The measurement infrastructure to prove that connection is no longer speculative. It is operational.
I coined Machine Relations in 2024 as a distinct discipline from traditional PR because the outputs are different. Traditional PR measures coverage. Machine Relations measures whether that coverage enters the citation pool AI engines use when buyers ask questions. Performance-based PR pricing becomes rational when both sides can agree on the Machine Relations scoreboard: which placements got cited, by which engines, for which queries.
When Each Model Fits (and When It Breaks)
No single attribution model captures the full picture. Here is how I match models to business types:
- B2B long-cycle (enterprise SaaS, professional services): Multi-touch media attribution + AI citation attribution. Long sales cycles need touchpoint sequencing. AI citation captures the research phase that starts before any human conversation.
- Product-led SaaS: Search-lift attribution + source authority scoring. Product-led buyers search before they sign up. You need the placement to rank and the outlet to be extractable.
- Enterprise/regulated industries: Source authority scoring + AI citation attribution. Regulated buyers need authoritative sources. The AI engine's retrieval decision is the proxy for buyer trust.
- Consumer/retail: Multi-touch attribution + search-lift. Volume matters, and the path from coverage to conversion is shorter and more measurable.
The failure case is always the same: relying on one model and missing the signals the others catch. A placement that does not get cited by AI might still drive a 60% branded search spike. A citation that appears in Perplexity might not move pipeline if the outlet is not trusted in your vertical.
What to Demand From Your Agency Before Signing
If you are evaluating a performance-based PR agency, here is the measurement infrastructure that separates real operators from agencies that renamed their retainer:
- Real-time citation dashboard — not a monthly PDF. You should see which placements are getting cited, by which AI engines, for which queries, updated at least weekly.
- Query-level attribution — not just "we got you in Forbes." Which queries trigger the citation? Are they the queries your buyers actually use?
- AI engine retrieval verification — the agency should run retrieval tests across ChatGPT, Perplexity, Gemini, and Google AI Overviews for your target queries after each placement publishes.
- Cost-per-citation metrics — compare cost-per-citation to cost-per-placement. An agency charging $3,000 per placement that achieves a 40% citation rate is delivering citations at $7,500 each. An agency charging $5,000 per placement with an 80% citation rate delivers them at $6,250 each. The cheaper placement is not the better deal.
- Placement-to-pipeline correlation data — show me the branded search delta, the referral traffic, the demo requests within 14 days of publication.
The ROI Gap Is Already Visible
The agencies that have adopted these attribution models are pulling away from retainer firms in every measurable dimension. They can prove what works, kill what does not, and compound the placements that produce the highest citation rates.
Gartner's May 2026 research recommended that marketers shift a portion of paid media budgets out of advertising and into answer engine optimization. That budget is moving toward agencies that can prove earned media ROI — and away from agencies that cannot.
I built AuthorityTech around performance pricing from the start — brands pay only when placements land in publications AI engines actually cite, measured against citation architecture across five engines. The 1,500+ direct editorial relationships I have developed over eight years are the network that makes guaranteed placement outcomes in high-authority publications possible. That is the difference between a performance-based PR agency and a retainer firm with a new pricing page.
The measurement infrastructure is no longer the bottleneck. The bottleneck is willingness to use it.
Frequently Asked Questions
What is performance-based PR?
Performance-based PR is a pricing model where the agency gets paid when measurable results are delivered — placements published, AI citations earned, or pipeline outcomes generated — rather than charging a flat monthly retainer regardless of results.
How does AI citation tracking work for PR measurement?
AI citation tracking monitors whether specific media placements get retrieved and cited by AI engines (ChatGPT, Perplexity, Gemini, Google AI Overviews) when users ask relevant queries. The tracking runs automated queries against each engine and records which sources appear in the response, creating a timestamped attribution record. Muck Rack's May 2026 data shows 96% of ChatGPT responses include sources, averaging 5 citations per response.
What ROI metrics should a performance PR agency provide?
At minimum: AI citation rate per placement, query-level attribution data, branded search lift within 14 days of publication, referral traffic from each outlet, and cost-per-citation alongside cost-per-placement. The best agencies also track placement-to-pipeline correlation.
Is performance-based PR more expensive than retainer PR?
On a per-placement basis, performance PR often costs more than the implied per-placement cost of a retainer. But the total cost of outcomes is typically lower because you only pay for results that materialize. A retainer agency charging $15,000/month with inconsistent placement volume may deliver fewer measurable results than a performance agency charging $4,000–$6,000 per verified placement.
Which attribution model is best for B2B companies?
Most B2B companies benefit from combining multi-touch media attribution with AI citation attribution. The multi-touch model captures the long sales cycle from first awareness through pipeline, while AI citation attribution tracks the research phase where buyers use AI engines to evaluate vendors before any human contact.
What is Machine Relations and how does it connect to performance-based PR?
Machine Relations is the discipline of earning AI citations and brand recommendations across AI-mediated discovery systems. I coined the term in 2024 after documenting that the same earned media placements that build credibility with human readers are the citations AI engines retrieve when buyers ask questions. Machine Relations provides the measurement framework that makes performance-based PR enforceable — instead of measuring clip counts, it measures whether earned media placements enter the citation pool AI engines use when buyers ask questions. AuthorityTech operationalizes this by tracking citation presence across ChatGPT, Perplexity, Claude, Gemini, and Google AI Mode.