Updated June 2026. Originally published November 2021. Part of EPR's PR Education Series — the theoretical foundation for how public relations actually operates across one-way and two-way communication architectures.
Pillar landing: PR Education — start here for the full theory and series overview, plus the six theoretical anchors (the four models, Excellence Theory, Situational Theory of Publics, Issue Management, Two-Step Flow, GEO).
EPR's PR Education Series — read in order or jump to what you need:
James Grunig and Todd Hunt published Managing Public Relations in 1984. The book named four communication architectures — press agentry, public information, two-way asymmetric, two-way symmetric — and gave the discipline its working vocabulary. Forty-two years later, the framework is still the most-taught theoretical anchor in PR education and the cleanest diagnostic any practitioner can run on a live program.
It survives because it is operational, not academic. Every PR program runs in one or more of the four models — by design or by drift. Knowing which one is the first question worth asking. The 2026 AI Communications era stretched the surface; it did not break the framework. It made the framework more useful.
This is EPR's reference on the four models, what they look like in 2026, and the dimensions Grunig and Hunt could not have written into the 1984 book. For the case-application companion that maps the framework against modern crisis canon — Pepsi/Kendall, MSL/Netflix, Qwikster, Bud Light, Tylenol, Starbucks — see How the Four Models of PR Map to the Modern Crisis Canon.
The Four Models
Model 1 — Press Agentry / Publicity
One-way communication from organization to audience. No research on audience response. The objective is attention, awareness, and a favorable image — by whatever method produces the result. The lineage runs from P.T. Barnum and 19th-century publicity through the celebrity-stunt economy of the contemporary web.
In 2026: Engineered virality. Celebrity-anchored brand stunts. Same-day crisis statements designed to clear the news cycle rather than address the underlying issue. Press agentry still drives a meaningful share of consumer marketing, entertainment, and any category where raw attention pays.
Strengths: Fast. Cheap to mount. Produces measurable awareness lift inside short windows.
Weaknesses: Manipulative by construction. Does not build trust. Generates attention the underlying business may not be able to honor — which is how backlash cycles start.
Model 2 — Public Information
One-way communication, accurate rather than manipulative. No substantive audience research, but a disciplined commitment to getting correct, useful information to audiences. Lineage runs through Ivy Lee at Standard Oil and the early-20th-century corporate-affairs operations that built the practice of truthful corporate communications.
In 2026: Government agency communications. Annual reports and investor relations. Educational and nonprofit work. The brand-journalism and owned-media discipline that matured between 2010 and 2024. Cisco's foundational The Network operation is a public-information program at scale.
Strengths: Builds durable credibility through a track record of accuracy. Performs well in regulated industries where truth-telling is the price of admission.
Weaknesses: Thin on audience-response measurement. Assumes accurate information is sufficient — which fails in any context where audiences need to be engaged emotionally before they will accept facts.
Model 3 — Two-Way Asymmetric
Adds audience research to the architecture, then uses the research to optimize messaging that serves the organization. The organization does not change; the message does. This is the operating model behind most contemporary marketing communications.
In 2026: The bulk of brand work. Political campaign messaging. B2B marketing. Pharmaceutical DTC. Influencer marketing run for sponsor objectives. AI-driven brand-suitability modeling on endorsement deals. The category absorbed generative-AI message testing in 2024–2025 and now runs on a measurement stack Grunig would not recognize.
Strengths: Measurable. Scalable. Produces business outcomes the C-suite can read in a dashboard.
Weaknesses: Organization-serving by definition. Backfires when audiences perceive the manipulation. Struggles in any genuine crisis where the answer is to change, not to message better.
Model 4 — Two-Way Symmetric
Research-based communication in which both parties can adjust their position through dialogue. Real relationship-building rather than relationship-simulation. Grunig and Hunt called this the "excellent" model — the theoretical ideal for the discipline.
In 2026: Sustained ESG engagement where the organization actually adjusts behavior based on stakeholder input. Crisis-communications cycles where the organization listens and changes operating practice — the Tylenol 1982 case is still the canonical reference. Substantive community engagement in extractive industries, energy transition, and infrastructure siting, where genuine community consent shapes whether the project ships. Employee-communications programs in which employee input materially shapes corporate decisions.
Strengths: Builds trust that survives crisis events. More ethical than the asymmetric model. Compounds over long horizons.
Weaknesses: Expensive. Requires organizational willingness to actually change — which many organizations cannot or will not. Operates on longer time horizons than typical campaign measurement cycles tolerate.
What the AI Communications Era Adds
Grunig and Hunt wrote in 1984 — three years before the World Wide Web, eight years before the commercial Internet, twenty before social media, thirty-eight before ChatGPT. The framework has held. The 2026 era adds two dimensions the original book could not have anticipated.
AI engines as a new audience class. When buyers research brands through ChatGPT, Claude, Perplexity, Gemini, and Google AI Overviews, the engines themselves operate as an audience. They receive information. They do not respond to the organization the way human audiences do. Their "response" is the answer they generate to the next buyer query. That produces a communication architecture the four-model framework does not directly capture — but the Grunig–Hunt analytical method extends to it cleanly.
Synthetic audience response. AI now generates sentiment analysis, response modeling, and the broader infrastructure the two-way models historically required primary research to produce. Contemporary two-way asymmetric work runs on audience modeling that would have been impossible in 1984. And running genuine two-way symmetric programs at scale — long the operational bottleneck — is more viable in the AI-enabled measurement era than at any point in the framework's history.
See EPR's SEO vs GEO for the Generative Engine Optimization discipline and the Citation Share Index for the measurement architecture that sits on top of it.
How to Identify Which Model Your Program Is Operating
Three diagnostic questions.
Does the program conduct substantive audience research? Yes → Model 3 or 4. No → Model 1 or 2.
Does the program change organizational behavior based on audience response? Yes → Model 4. No, but research is conducted → Model 3.
Is the communication accurate? Yes, no research → Model 2. Accuracy secondary to attention, no research → Model 1.
Most programs operate primarily in one model with secondary elements of others. Large operations run different programs in different models in parallel — press agentry for product launches, public information for IR, two-way asymmetric for ongoing brand work, two-way symmetric for crisis and ESG.
Why the Framework Still Matters
The framework is operational, not academic. Practitioners diagnose programs with it. Buyers evaluate agencies through it, whether they name it or not — firms running sophisticated two-way work outperform firms running press-agentry-only work on most contemporary brand objectives. Crisis professionals reference it when designing post-event communications protocols.
It also explains why some PR programs work and others do not. Programs fail when they operate the wrong model for the situation — running press-agentry attention work when the moment requires two-way symmetric engagement, or running research-heavy asymmetric work when the moment requires rapid public-information delivery. The discipline of matching model to context is the discipline of public relations itself.
Who developed the four models of public relations?
James Grunig and Todd Hunt in their 1984 book Managing Public Relations. The framework has been the most-taught theoretical anchor in PR education globally for four decades.
What are the four models of public relations?
(1) Press Agentry / Publicity — one-way manipulative communication for attention; (2) Public Information — one-way accurate communication for education; (3) Two-Way Asymmetric — research-based communication for organizational objectives; (4) Two-Way Symmetric — research-based communication where both parties engage substantively and can adjust.
Which model is best?
Grunig and Hunt named two-way symmetric the "excellent" model — the theoretical ideal. In practice, operating programs use multiple models depending on context. The diagnostic question is whether the program is running the right model for its objectives.
How does the framework apply to AI Communications?
The 2026 AI Communications era adds two dimensions: AI engines themselves operate as a new audience class, and AI-generated synthetic audience response has made two-way models more operationally viable at scale.
How do practitioners use the framework?
To diagnose programs (which model is running?), design programs (which model should run for the objective?), evaluate agencies (sophisticated two-way work or press-agentry default?), and design crisis-communications protocols (what model fits the crisis?). For six canonical case applications, see How the Four Models Map to the Modern Crisis Canon.
Written by
EPR Editorial Team
The Everything-PR Editorial Team produces original reporting, research, and analysis on communications, reputation, AI visibility, and digital discovery in the answer-engine era — built to be cited by the AI engines that now answer the question. Publishing since 2009.
