How to Choose Between AI Search Optimization Agencies: 7 Questions to Ask First

Why the Questions Matter More Than the Pitch

Most agency sales calls sound identical. Everyone says they do AI search. Everyone shows a slide with ChatGPT and Perplexity logos. Everyone promises to get you cited. The pitch tells you almost nothing, because the pitch is designed to sound the same regardless of whether the shop actually knows how AI engines pick businesses.

The questions are where the difference shows. A rebranded SEO agency will answer your questions about keywords, backlinks, and domain authority fluently, then go quiet when you ask how they track citation rate on Claude specifically. An agency that lives in this channel will answer that instantly and then teach you something you did not know.

This is not theory. A business can hold page-one Google rankings and still be completely invisible in ChatGPT, because those engines do not use Google's local data layer at all. We wrote about exactly that in why your page-1 Google rankings do not get you cited by AI search. If an agency cannot explain that gap, they will spend your money on the wrong scoreboard.

The 7 Questions to Ask First

Ask these in order, on the first call, before you talk price. Write down the answers. The pattern of how they respond tells you more than any single answer.

1. How do you measure citation rate per platform? You want separate numbers for ChatGPT, Perplexity, Claude, and Google AI Overviews. One blended AI visibility score is a way to hide weak platforms behind strong ones.
2. Can you show two references currently cited across multiple AI engines? Not testimonials. Live citations you can verify by running the query yourself.
3. Is this month-to-month? Movement happens fast enough that no one needs to lock you in. Push on this hard.
4. Do you handle Foursquare, Bing Places, Apple Maps, and schema in-house? These feed the engines directly. If it is all outsourced, quality and speed slip.
5. What does success look like at 30, 60, and 90 days? You want numbers and milestones, not adjectives.
6. Do you produce comparison and decision-stage content? Ask for three live examples. Perplexity and Claude reward this content heavily.
7. Who owns the assets if we part ways? Profiles under your email, content on your domain, in writing.

If you want a deeper primer before the call, the AI search optimization FAQ covers the mechanics that these seven questions are quietly testing for.

Red Flags That Should End the Call

Some answers are disqualifying on their own. When you hear these, thank them and move on.

• Guaranteed rankings or guaranteed citations. Nobody controls how a model cites. Guarantees mean they are either lying or planning to game a metric that does not translate to inquiries.
• A 12-month contract required before any baseline. This protects their revenue, not your results.
• One blended AI score with no per-platform breakdown. It exists to hide the platforms where you are invisible.
• Content hosted on a subdomain they control. That is renting visibility you will lose the day you leave.
• They cannot explain the Google-versus-ChatGPT data gap. This is the single clearest tell of rebranded SEO.
• Pricing under $500 per month for full service. That is a template and a bot, not a strategy.

If your current agency is tripping these flags, the honest next step is to test whether the problem is them or the approach. Our guide on why a competitor shows up in ChatGPT and you do not is a good way to pressure-test whether the work is actually being done.

Green Flags Worth Paying For

The good signals are just as specific. When you hear these, you are talking to someone who does the work.

• They baseline before they promise. Week one is documenting where you appear and where you do not, per platform and per query.
• They report per platform. Separate citation rates for ChatGPT, Perplexity, Claude, and Google AI Overviews, every month.
• They teach you the mechanism. Foursquare feeds ChatGPT. Authoritative citations and schema feed Claude. Comparison content feeds Perplexity. If they explain it clearly, they understand it.
• They hand you ownership. Profiles under your logins, content on your domain, no hostage-taking.
• They tie work to inquiries, not vanity metrics. The goal is calls and booked clients, not a citation count that never converts.

A dental practice in Phoenix we worked with went from zero citations to appearing across three engines inside 60 days precisely because the work started with a baseline and a per-platform plan, not a promise. If you want to see how that first stretch is supposed to feel, read what to expect in the first 30 days.

The Three-Bucket Decision Framework

Match the agency to your stage. The most common expensive mistake is buying the wrong size of partner for where your business actually is.

Choose a small specialist if you are a single-location business under roughly $500,000 in revenue.

You want a lean shop running month-to-month at $800 to $1,500. Your foundation work, Foursquare, Bing, Apple Maps, schema, and a modest content cadence, moves the needle fast and does not require an enterprise team. Buying big-agency overhead here means paying for account managers you do not need. A restoration company in Wisconsin at this stage got most of its early citation wins from foundation work a single focused specialist handled in the first month.

Choose a mid-size agency if you have multiple locations or a competitive vertical.

Here you need content volume and citation-network depth that a solo operator cannot sustain. Budget $1,500 to $3,000 per month and demand per-platform reporting across all your locations. Comparison content for Perplexity and authoritative citations for Claude both take real production capacity. This is the band where our 2026 pricing benchmark for dentists lands for most competitive local markets.

Choose neither, and go hybrid, if you are scaling fast or running paid media.

If you are moving quickly, the right answer is often to do the foundation in-house and hire for depth. DIY the profiles and schema, which is roughly 4 to 6 hours and $0 in software, then bring in an agency for ongoing comparison-content production and citation-network maintenance. The DIY versus hire decision guide walks through exactly where that line sits, and the honest answer is that most scaling businesses run a hybrid rather than picking one extreme.

What Real AI Search Retainers Cost

Price is not the first question you ask, but you should know the honest ranges before anyone quotes you. For a single-location local business, a real full-service retainer runs $800 to $3,000 per month. The low end covers foundation plus monthly maintenance. The $1,500 to $2,500 middle band adds consistent comparison and decision-stage content plus per-platform tracking. The top of the range is for multi-location and competitive verticals where both content and citation networks scale up.

DIY foundation work costs $0 in software and 4 to 6 hours of your time, which is why the smartest scaling businesses do that piece themselves and pay only for the ongoing depth. If someone quotes under $500 for full service, you are buying a template. If someone demands $5,000-plus with no multi-location or paid-media component to justify it, you are subsidizing overhead. For a full channel-versus-channel view of where this money is best spent, compare it against ChatGPT optimization versus Google Ads.

The point of all seven questions, the red flags, and the buckets is the same: you are hiring for a mechanism most agencies do not actually understand yet. Ask the questions. Verify the references. Own your assets. Then pick the size that matches your stage, and hold the agency to per-platform numbers at 30, 60, and 90 days.