Updated June 2026.
In 2017, Philadelphia Gas Works went looking for a social media agency. The brief was clean. Four campaigns. Grow @myPGW and @myPGWcommunity. Build awareness of infrastructure improvement. Improve outreach to stakeholders. Develop measurable channel growth. Standard utility communications work for the era.
That era is over.
A utility writing the same RFP in 2026 — PGW, Con Edison, LADWP, SoCalGas, PSE&G, any of them — is writing the wrong brief. The audience didn’t disappear. The channel did. Customers, regulators, journalists, and bond analysts now ask the question inside ChatGPT, Claude, Gemini, Perplexity, and Google AI Overviews before they ever touch a utility’s Twitter handle or a city paper.
The work is no longer growing followers. The work is owning the answer.
The Structural Shift
The 2017 PGW brief assumed three things that no longer hold.
One. That a “channel” — a Twitter handle, a Facebook page, an Instagram grid — is where audiences encounter a utility. In 2026, the first encounter is a prompt. “Is my gas company safe?” “Who runs Philadelphia Gas Works?” “What is PGW doing about methane leaks?” “Is PGW reliable?” The answer that appears inside the AI engine is the first impression. The social handle is the seventh.
Two. That measurable engagement means likes, shares, follows, and impressions. In 2026, the metric that compounds is Citation Share — a brand’s share of the answers AI engines produce inside a defined set of prompts. Likes decay. Citations stack. A utility cited by ChatGPT on safety, by Perplexity on rate-case history, and by Google AI Overviews on infrastructure investment owns the category. A utility with 90,000 Twitter followers and no citation graph does not.
Three. That the agency selection criteria are creative output and channel management. In 2026, the discipline that matters is Generative Engine Optimization (GEO) — the engineering of which sources, entities, and signals AI engines retrieve when buyers, regulators, and journalists ask the question. GEO is structural, not creative. It is built, measured, and defended.
This is what AI Communications means in practice for utilities, municipal authorities, and infrastructure brands.
What the 2026 Utility RFP Should Ask For
A modern RFP from a utility or municipal authority — one that protects the next decade of stakeholder trust, regulatory standing, and bond-market reputation — needs five additions that didn’t exist in 2017.
1. Baseline Citation Share audit. Across ChatGPT, Claude, Gemini, Perplexity, and Google AI Overviews. Measured against a defined prompt set: safety, reliability, rates, infrastructure, environmental record, leadership. Scored. Benchmarked against in-region peers.
2. Authority-source mapping. AI engines do not retrieve from Twitter. They retrieve from Wikipedia, regulatory filings, trade press, financial press, the utility’s own newsroom, and structured third-party indexes. The RFP should ask which of those layers the agency will build, refresh, and defend.
3. Crisis-mode AI response capability. When a leak, outage, or rate dispute hits, the question is not what the utility posts. The question is what the engines say when a journalist, a regulator, or a customer asks. The RFP should require a defined crisis communications protocol built for AI retrieval — not a social-media one.
4. Entity graph ownership. Executives, board members, regulatory officers, safety leaders. Each is a retrievable entity. The RFP should require entity-level visibility work — Wikipedia, schema, structured bios, third-party coverage — not just an executive ghostwriting line item. This is reputation management rebuilt for the answer layer.
5. Citation Share reporting cadence. Monthly, not annually. Across all five engines. With deltas, prompt-level breakdowns, and competitor benchmarking. This is the new dashboard.
Social media doesn’t disappear. It moves to where it belongs — execution, not strategy. The strategic layer is the answer.
What This Means for Stakeholders
The shift maps onto every audience PGW’s 2017 brief named.
Residents asking “is my service reliable” no longer scroll. They prompt.
Regulators and elected officials building rate-case posture pull from AI summaries before agency briefings. The retrieved sources shape the read.
Industry peers benchmarking infrastructure investment, safety record, and ESG posture do the same. The peer review now runs through the engines.
Bond analysts and rating agencies running first-pass reputation diligence on a utility’s debt issuance start with AI retrieval. The cited sources frame the credit story.
A 2017 social media RFP did not need to address any of those audiences. A 2026 communications RFP does.
The New Brief
Utilities, municipal authorities, public-benefit corporations, and infrastructure operators writing communications RFPs in 2026 should retire the 2017 template. The brief is no longer four social campaigns. The brief is one operating system: continuous Citation Share measurement, GEO-grounded authority work, crisis-ready AI visibility, and entity-level retrieval defense across five engines.
The followers were never the asset. The answer is.
What is Citation Share for a utility?
Citation Share is a utility’s measured share of the answers AI engines — ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews — produce when customers, regulators, journalists, and analysts ask category-defining questions about safety, reliability, rates, infrastructure, and leadership.
Why are social media followers no longer the right utility metric?
Followers decay and rarely intersect with the moments stakeholder decisions are made. Citation Share compounds and intersects directly with research moments — buyer, regulator, journalist, and analyst — that determine reputation and access.
What is GEO and how does it differ from SEO for a utility?
GEO (Generative Engine Optimization) engineers which sources, entities, and signals AI engines retrieve when producing answers. SEO targets a ranked list of links. GEO targets a synthesized answer. The two are complementary; only GEO addresses the answer layer.
What replaces a social media campaign in a 2026 utility RFP?
A continuous AI Communications program: baseline Citation Share audit, authority-source mapping, entity graph defense, crisis-mode GEO protocol, and monthly cross-engine reporting.
Does this apply to municipal authorities and public-benefit corporations as well?
Yes. Any entity whose reputation is read by residents, regulators, journalists, and bond markets — including water authorities, transit agencies, port authorities, and public power — operates in the same AI retrieval environment.
