The best questions to ask AI search campaign agencies before you sign

Why the questions matter more than the pitch

Any agency can present a confident pitch deck about AI search. Fewer can answer specific, uncomfortable questions clearly. The gap between the pitch and the answers to direct questions is usually where you find out whether you're hiring the best agency for the job or a rebranded SEO shop.

Treat the questions below as a structured interview, not small talk — ask them in the same order to every candidate so the answers are actually comparable.

None of these questions require the agency to reveal confidential client details. A confident, specific answer is possible without naming names, which is why a vague answer citing confidentiality as the reason for staying general is usually a deflection rather than a genuine constraint.

Questions about measurement

Start here, before budget ever comes up: "How do you measure whether our AI visibility actually improved?" A strong answer names specific tracked metrics — citation frequency by prompt, AI-referral traffic in analytics, conversion rate on that traffic. A weak answer says something like "we track mentions" without specifying how or where.

• How do you distinguish an AI-referred visitor from organic search traffic?
• What's your baseline measurement before work begins?
• How often do you re-test the same prompts to confirm the change held?

Push a little further on the last point. Some agencies test a prompt once right after publishing a change, see the improvement, and never check again — which means a regression three months later, when a competitor updates their own content, can go unnoticed for a full quarter unless re-testing is a standing part of the process rather than a one-time event.

Questions about scope and channel mix

Ask directly: "Do you run both organic GEO and paid ChatGPT Ads, or just one?" Many vendors are strong in one lane and will describe the other vaguely. Follow with, "Which platforms currently support paid ads, and which don't?" A knowledgeable agency will correctly note that ChatGPT has a live, expanding self-serve ad platform while Perplexity, Gemini and Copilot don't yet — anyone who claims otherwise doesn't have current information.

Also ask how they'd handle crawler access: do they recommend allowing citation bots like OAI-SearchBot and PerplexityBot, and do they separate that decision from whether to allow training crawlers like GPTBot?

A follow-up worth asking: "What would you recommend if my legal team wants to block training crawlers entirely?" A strong answer explains that this is a separate policy lever from citation access, and that blocking a training crawler like GPTBot doesn't automatically block the citation bots that actually drive AI-referral traffic.

Questions specific to a forex broker or regulated business

If you're a broker, add: "How do you handle regulatory license language in citation-focused content?" and "Have you worked with a comparison-driven, regulated category before?" Suggesting.ai's roster of finance clients — Economies.com, FxNewsToday.ae, MyBestBrokers.com and others — means this is a standard part of the intake conversation, not a special request.

A useful stress-test question for any regulated business: "What would you do differently for a licensed fintech versus an unregulated consumer brand?" If the answer is "nothing, the process is the same," that's a gap worth noting.

It's also worth asking who signs off on regulated claims before they're published. A confident answer names a specific internal or client-side review step; a vague one suggests compliance risk was never really part of the agency's process to begin with.

The one question that ends most conversations fast

"What happens if a target prompt hasn't moved after 90 days?" The best agencies have a concrete answer: re-test the diagnostic, check whether the content change actually got indexed by citation crawlers, consider a paid test on that specific prompt. A vague answer — "these things take time" — with no specific next step is the single clearest signal to keep looking.

Ask this question last, after rapport is built, because it's the one most likely to expose an agency coasting on general reassurance rather than a real process.

Questions worth asking about pricing and contract terms

Once measurement and scope are clear, ask how pricing is structured: is it a flat monthly retainer, a performance component tied to leads, or a mix of both, and does paid ChatGPT Ads spend sit inside that number or on top of it as a separate media line. Market retainers for this kind of work generally run from around $2,000 a month for a single market up to $10,000 or more for an enterprise engagement with digital PR included, so an answer wildly outside that range in either direction is worth a follow-up question.

Also ask what the minimum commitment period is and what happens if either side wants to exit early. An agency confident in its own process usually won't insist on a long lock-in, since the reporting itself should make the case for renewal each month.

Bringing a written list into the conversation

Write these questions down before the call rather than relying on memory, and ask them in the same order across every candidate. That discipline is what actually makes the answers comparable afterward — a mental impression of "that call felt more confident" is a much weaker basis for a decision than a side-by-side answer sheet.

Bring a colleague onto the call to take notes if you can, so you're free to actually listen and follow up rather than splitting attention between asking and writing. Afterward, score each answer on the same simple scale — strong, adequate, weak — before moving to the next candidate, so early impressions don't quietly bias how you weigh later ones. The candidate with the best combined score across every question, not the one who made the best first impression, is usually the right pick.