Congressional staffers, agency analysts, regulators, and journalists covering policy now run their first-pass research through AI engines. Hearing prep. White paper drafting. Constituent response. Briefing memos. The AI synthesizes across think tanks, trade associations, academic literature, agency reports, and tier-1 press — and names two or three sources.
Generative Engine Optimization (GEO) is the discipline that gets a coalition, a company, or a movement cited in that synthesis. Whoever owns the citation shapes the policy frame.
Why public affairs is uniquely exposed
Policy research is high-stakes, low-time-budget. Staffers and analysts triage research under deadline pressure. AI engines compress a week of reading into a synthesized briefing. The cited sources shape the resulting policy work.
The think-tank citation graph is concentrated and contested. Brookings, Heritage, AEI, RAND, CSIS, Pew, Cato, ITIF, EPI, CAP. Whoever gets cited alongside these — as an industry voice, a coalition, or a primary-source data provider — enters the policy conversation. Whoever doesn't is invisible.
Regulators are using AI tools internally. Agencies are adopting AI for review work — comment period analysis, rulemaking research, enforcement-priority setting. The sources retrieved by those tools are influencing agency posture.
What policy buyers ask AI engines
A representative prompt universe across federal and state public affairs:
Issue briefing — "what is the policy debate around [issue]," "main arguments for and against [bill]"
Stakeholder mapping — "key advocates for [position]," "coalitions opposing [regulation]"
Economic and impact — "economic impact of [proposed rule]," "jobs at stake in [industry]"
Comparative — "how does [state/country] regulate [issue]," "[bill] vs [bill]"
Hearing prep — "questions to ask [witness] at [hearing]"
Each of these is a winnable retrieval slot for a coalition, association, or company that has built the right citation infrastructure.
What works in public affairs GEO
Original research and primary data. Studies, economic impact analyses, polling, and white papers — published in retrievable, well-structured form — get cited heavily. AI engines prefer primary-source data over aggregated commentary. Trade associations and coalitions sitting on proprietary industry data are mispricing the asset if they're not publishing it for AI retrieval.
Named expert positioning. Policy directors, chief economists, general counsel, former officials. AI engines retrieve named experts at meaningful frequency on policy prompts. Coalitions with cultivated named-expert citation across Politico, Axios, The Hill, Bloomberg Government, NYT, Washington Post, Wall Street Journal own the expert layer of the synthesis.
Coalition and signatory documentation. AI engines retrieve coalition lists and signatory blocks when they're published in clean, structured form. "Who supports [position]" is a high-frequency prompt that often returns whoever published the cleanest list.
Tier-1 policy press citation. Politico, Axios, Punchbowl, The Hill, Bloomberg Government, Roll Call, Inside Health Policy, Inside Defense. Plus the mainstream tier-1 — NYT, WaPo, WSJ, Reuters. Each named placement is a retrieval anchor at the policy layer.
Regulatory comment letters in retrievable form. Filed comments published openly with named authors, organizations, and structured framing get cited in regulatory rulemaking research. AI engines now retrieve them.
The five-layer GEO stack for public affairs
1. Entity foundation — organization, leadership, member companies, coalitions as clean entities with full schema
2. Owned canonical content — issue hubs, original research, position papers, member resources — schema-marked, primary-source
3. Earned-media citation infrastructure — tier-1 policy press, original research distribution, named-expert visibility
4. Measurement — Citation Share across issue-specific prompt universes via Curium.io
5. Continuous optimization — legislative calendar, hearing cycles, rulemaking windows, election cycles
Why this is bigger than traditional lobbying budgets recognize
Most public affairs operations are still resourced as lobbying + government relations + a press secretary. The citation layer is treated as adjacent to comms. But the synthesized answer that's now sitting between an analyst and the policy work is doing more frame-setting per dollar than a quarter of in-person Hill meetings.
The coalitions that recognize this in 2026 are repricing their budgets toward citation infrastructure. The coalitions that don't are funding lobbying programs whose effectiveness is shrinking because the upstream framing is being shaped without them in the room.
What to do this quarter
1. Map your policy prompt universe — 200 to 500 prompts your target staffers, analysts, and journalists actually run.
2. Baseline Citation Share across issue prompts and tier-1 outlets. 5W runs this audit with policy-specific framing.
3. Lock named-expert infrastructure — policy directors and chief economists positioned as cited experts.
4. Audit primary research output — what proprietary data are you sitting on that isn't published in retrievable form.
5. Build the regulatory and comment letter pipeline with GEO discipline — open publication, named authorship, structured framing.
The policy battles of 2027 will be framed by the citation infrastructure built in 2026.
