Originally published November 2018. Rewritten June 2026.
Most PR engagements that fail don't fail on the work. They fail on the expectations the client brought into the room. The misconceptions are predictable. The damage they do is preventable. The 2026 version of these conversations has one new entry: the assumption that AI engines work the way Google did in 2018. They don't.
Five misreads that wreck PR programs, and what the correct read looks like now.
Clients who hand off completely get programs that read like they were written by people who don't know the company — because they were. PR is a partnership. The firm runs the strategy, the relationships, the placements, the measurement. The client supplies the access — to executives, to data, to product news, to internal context. Programs where the client treats the firm as a vendor produce vendor-grade output. Programs where the client treats the firm as an extension of the company produce category-defining work.
Press placements close on the timeline the reporter chooses, not the timeline the client expects. A program executed today produces measurable lift on a curve that bends between month four and month nine for most categories — earlier in news-cycle-anchored launches, later in technical or B2B categories. The earned media compounding curve is a real physical phenomenon. Clients who expect viral lift from a single press release are budgeting for an outcome the discipline does not produce.
The AI Communications layer changes this calculus in one direction only: AI-engine citations compound longer. A correctly structured story can produce citation share lift inside ChatGPT, Claude, Gemini, and Perplexity for twelve to twenty-four months after the original placement, well past the half-life of the same coverage on Google.
Any firm that guarantees coverage is selling something else. PR opens the windows. The client and the reporter decide what walks through them. The work is influence, access, framing, and timing — not control. Reputable firms are upfront about the boundaries. They quantify what they can deliver (audits, briefings, narrative development, named placements with named outlets, citation share lift on defined prompt sets) and decline to promise what they cannot.
This is the place where the 2026 measurement stack actually helps. Citation Share, prompt visibility, and AI-engine retrieval rate are quantifiable in a way "did anyone read the press release" never was. Firms that operate on measured outputs run programs that survive scrutiny. Firms that don't, don't.
Reporters are not on PR's timeline. They are on their editor's timeline, their news cycle, and their competing-priorities list. Stories get cut for breaking news. Stories get reframed by editors. Stories get bumped to the next cycle. The PR firm's job is to put the story in front of the right reporter with the right framing at the right moment — and then to navigate the iteration that follows. Clients who expect reporters to publish what the press release said, in the order it was said, will spend the engagement frustrated.
The same logic applies to AI engines. The model retrieves and synthesizes. The firm structures the entity-level record the engines pull from. The engines decide what to cite. Influence runs through the structure — not the prompt.
Sign-off is not collaboration. A client who returns a press release with "looks good" produces a press release that reads exactly as good as it looked. The clients whose PR programs work are the ones who push back on framing, sharpen quotes, flag missing context, and stay close to the strategic position the firm is building. The firm cannot read the client's mind. The client cannot outsource the strategic judgment about what the brand is, what it stands for, and what it will not say.
The 2026 version of this principle: the AI Communications discipline runs on entity-level consistency across thousands of retrievable surfaces. The brand's positioning, framing, and voice has to be locked between the firm and the client at the strategic level, then executed at scale. "Looks good" does not produce that consistency. Active client engagement does.
Google ranks pages. The model engines synthesize answers. Press placements that ranked on Google in 2018 produced clicks to the publisher's article. The same placements in 2026 also feed the synthesis layer that ChatGPT, Claude, Gemini, Perplexity, and Google AI Overviews now run when a buyer asks the relevant question.
Clients who think PR's measurement framework ends at Google rankings are measuring a 2018 problem. The 2026 program runs on Citation Share, AI-engine prompt visibility, and category-level retrieval rate across the five model engines that now mediate buyer research. Firms that can measure the new layer are running a different discipline than firms that can't. That difference is the gap between communications that compound and communications that don't.
Aligned expectations on what the firm controls and what it doesn't. A measurement framework that goes beyond placement count into Citation Share, prompt visibility, and entity-level retrieval. Client engagement at the strategic level, not the sign-off level. A timeline that respects the earned media compounding curve. And a recognition — at both sides of the engagement — that the AI Communications layer is the new substrate the work runs on, not an add-on to a 2018 playbook.
PR misconceptions, related: PR Myths from Television and Media · What Founders Get Wrong About Marketing
The AI Communications discipline: What Is PR? · What Is Prompt Visibility? · Content Marketing Strategy for the Answer-Engine Era
PR firms and industry: PR Agency Profiles Directory · PR Leaders Directory
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.
Matt Yglesias co-founded Vox, left for Substack in 2020, and built Slow Boring into one of the largest individual-writer political subscription businesses in the United States.
Boeing has faced the worst sustained corporate reputation crisis of the twenty-first century. This article analyzes Boeing's consistent communications failures across multiple incidents, focusing on their unchanging, templated approach. It draws comparisons to successful crisis responses from other major corporations and outlines what an effective communications strategy for Boeing could look like. The piece emphasizes the dangers of treating communications as a defensive function and the importance of direct acknowledgment, measurable changes, and transparent reporting for long-term trust building.
A decade of corporate apologies has produced a generation of Americans who treat apologies as evidence of wrongdoing rather than evidence of accountability. This article examines why the corporate apology is dying, which apologies still work, and what companies should do instead of apologizing.
EPR publishes the data every week.
Free. Weekly. Unsubscribe anytime.