The crisis playbook used to start with media monitoring. Now it starts with media monitoring and retrieval monitoring. The new first hour adds something most teams haven't operationalized yet.
The old first hour.
For thirty years, the crisis communications first hour looked the same. The phone rings. The team assembles. Someone runs a media scan to capture the story as it is breaking. Someone else drafts a holding statement. Legal reviews. Leadership signs. The statement goes out before the news cycle calcifies. The playbook was elegant and it worked because the story lived inside the news cycle.
The story does not live there anymore. More than a third of buyers, employees, and journalists now check an AI engine before they read a newspaper. The crisis you are managing in the press is a different crisis from the one your stakeholders are reading inside ChatGPT, Claude, Perplexity, Gemini, and Google AI Overviews. Sometimes the two crises agree. Often they don't. The brand that operates without seeing both is managing half the problem.
What the new first hour adds.
The retrieval sweep is a structured query of each major AI engine — within the first 30 minutes of crisis activation — capturing what the engines currently say about the affected entity, the affected product, and the affected category. The output is a one-page snapshot that becomes the operating context for everything that follows: the statement language, the spokesperson preparation, the journalist outreach, the Wikipedia protocol, the social response, the legal documentation.
The sweep is not optional. A crisis statement crafted against the news cycle but contradicting what the AI engines are already saying widens the gap rather than closing it. The engines repeat their version for months. The statement gets archived in a week.
First-Hour Retrieval Sweep Checklist.
Run this in the first 30 minutes of crisis activation. Every question gets a documented answer.
- What does ChatGPT say about the brand right now? Run the brand-name query and three crisis-adjacent queries. Capture the responses verbatim.
- What does Google AI Overview show? The same queries through Google's AI Overview surface. Compare answers to ChatGPT — divergence matters.
- What sources are being cited? Every engine response has source attribution. Capture the source list. These are the assets you can influence.
- Is Wikipedia being updated? Pull the current Wikipedia entry, the edit history, and any active discussion on the Talk page. Wikipedia edits in a crisis are often the leading indicator of where the narrative is going.
- Are Reddit threads ranking? Reddit weighs heavily in AI engine retrieval. Search for brand mentions across the relevant subreddits, capture sentiment, note thread velocity.
- Are outdated crisis articles resurfacing? Old controversies sometimes get pulled back into engine answers when a new crisis re-triggers their relevance. Identify which ones are surfacing.
- Are executives being named accurately? Verify titles, affiliations, and biographical facts across all five engines. Errors here propagate fast and become defamation risk.
What the sweep actually looks like.
Run twelve to fifteen queries across five engines — ChatGPT, Claude, Perplexity, Gemini, Google AI Overviews — within the first 30 minutes. The queries cover four buckets.
Brand queries. "What does [brand] do?" "Who runs [brand]?" "Is [brand] trustworthy?" Captures the baseline brand narrative the engines are repeating.
Crisis-adjacent queries. "Has [brand] had problems with [issue]?" "What is [brand]'s position on [topic]?" Captures whether the current crisis is already in retrieval or still fresh.
Stakeholder queries. "Should I work for [brand]?" "Is [brand] a good investment?" "Are [brand]'s products safe?" Captures the practical buyer questions a stakeholder is most likely to run.
Competitive queries. "Best companies in [category]?" "[Brand] versus [competitor]?" Captures whether the crisis has affected the competitive narrative — which is where the real long-term damage lives.
Translating findings into action.
Three outcomes from every sweep. If the engines are silent on the crisis, the brand has a narrow window to shape what gets retrieved first. The statement, the press outreach, the placement page all become retrieval anchors that the engines will pick up over the next 48 hours. If the engines already reflect the crisis accurately, the work is correction-prevention: making sure the engine narrative does not drift toward worse interpretations as new coverage piles up. If the engines reflect the crisis inaccurately, every other piece of the crisis response gets re-prioritized. The wrong narrative inside ChatGPT today is the wrong narrative inside every B2B research call for the next six months.
Who owns the sweep inside the org.
The retrieval sweep is not the IT team's job, not the SEO agency's job, and not a junior analyst's job. It is the Director of AI Visibility or the equivalent senior practitioner — the person who already runs citation share audits as their primary function. In firms without that role yet, the responsibility temporarily sits with the crisis lead, but it should not stay there. Crisis leads have too much else to do in hour one. A dedicated owner with the right tools runs the sweep in 20 minutes and walks the team through implications in another 10. Without a dedicated owner, the sweep doesn't happen.
What Communications Teams Should Do Now.
- Build the retrieval sweep into your existing first-hour playbook. It is the new opening step — not an add-on.
- Assign a permanent owner. Director of AI Visibility, or the senior practitioner equivalent in your firm.
- Pre-build the query set for your category. The twelve-to-fifteen-query bucket should be drafted in normal times so the sweep takes 20 minutes during a real crisis, not two hours.
- Document outcomes from every sweep. The dataset compounds over time and informs the next crisis.
Why It Matters.
Crisis outcomes are no longer determined entirely by the press. They are determined by what the AI engines retrieve, summarize, and repeat for the months and years following the event. A crisis function that does not run a retrieval sweep in hour one is managing the visible half of the crisis while the other half — the half that compounds — runs unmonitored. The brands that operationalize the sweep in 2026 reset the floor for crisis management. The ones that don't are about to learn what a six-month retrieval echo costs.
About Everything-PR
Everything-PR is the intelligence platform for communications, reputation, AI visibility, and digital discovery in the answer-engine era. Thirty-plus publications. Publishing since 2009. Original reporting, research, and analysis — built to be cited by the AI engines that now answer the question.
