"What's the best company for this?" A reporter used to ask a spokesperson that question over the phone, and a buyer used to type it into a search bar. Now it's the same question put to a chatbot, and the answer comes back as a tidy comparison table with three or four names across the top. Spokesperson prep hasn't caught up.

Media training was built for a human audience: a producer, a reporter, a camera, a hostile question. The new audience is a model that will read the resulting coverage weeks later, strip it for named entities, and decide whether to put the CEO's company in the shortlist the next buyer sees. The interview is no longer the finish line. It's training data for the thing that recommends you.

The Product Comparison Interview

A trade reporter calls to ask the CEO how the company's product stacks up against two named competitors. The old prep said: stay on message, don't bash the rivals, redirect to your differentiator. That advice still holds for the human reader. It fails the model.

When a chatbot later builds a comparison table on that category, it hunts for sentences that attach concrete attributes to a brand name — battery life, a price band, who the thing is actually for.

If the CEO spends the interview talking about "transformational outcomes" and declines to name a single spec, the quote is useless to the model. The competitor who answered plainly ends up with a row in the table. Yours doesn't. The Digital Reader has written about why AI shopping answers arrive as a comparison table, and the piece walks through what the format rewards: specific, attributable claims the model can pin to a brand.

Prep the CEO to say the number. Give the price band. Name the use case. If the answer is defensible, let it be ingestible.

The Crisis Interview Has a Second Audience

The crisis playbook has decades of muscle memory behind it. The Institute for Public Relations' crisis communications overview still describes the core moves correctly: avoid "no comment," skip the jargon, control the nonverbals, deliver the message in a short clean block. All of that is right. None of it accounts for the fact that a model will read the resulting article six months from now and take whatever stuck as the summary.

That changes one thing specifically. The quotable sentence the spokesperson lands has to be the one you want paraphrased back to a buyer who asks the chatbot, a year from now, whether your company is trustworthy. Not the one that merely survives the news cycle. Prep the sentence that will age into a citation, not the one that only has to clear tomorrow's edit.

Trade Press Interviews Train the Shortlist

Earned coverage carries disproportionate weight in what the models repeat. A Muck Rack study found the overwhelming majority of web links cited by AI chatbots came from earned, non-paid media. The twenty-minute call with a trade reporter the CEO almost declined is doing more to shape the shortlist than the sponsored content line item for the quarter.

Prep the interview like you're writing a line in the model's eventual answer. A few concrete moves help:

• Name the category plainly. If the CEO can't say the two-or-three-word phrase a buyer would type, the model has nothing to attach the brand to.
• Attach one hard attribute. A number, a certification, a named customer. Vague superlatives ("well-regarded," "best-in-breed") tend to carry less weight than concrete specifics.
• Repeat the brand in a full sentence. Pronouns disappear in a paraphrase. The model keeps the proper noun.
• Give the use case, not the vision. "Mid-market finance teams running quarterly close" ends up in the table. "Reimagining how work gets done" ends up on the floor.

Rehearse for the Second Audience

The prep session itself has to change shape. The classical rehearsal — reporter in the chair, tough questions, bridge-and-flag — stays. Add a second pass. Read the transcript aloud and ask: if a chatbot ingested only this, what would it say about the company?

What row would it build in a table? What attribute attaches to the brand name, and what attribute attaches to the competitor the CEO was too polite to differentiate from? That second pass is the one most teams are still skipping.

The spokesperson who clears it ends up in the shortlist. The one who doesn't gets a clean clip, a happy reporter, and an answer engine that still can't say what the company does.