5 Questions to Ask Any Marketing Agency About Their AI Search Capabilities

This is the fastest signal of genuine capability. Any agency selling AI search optimization should have built an llms.txt file for their own site first. If they haven’t practiced it on themselves, they’re either selling theory or outsourcing the work to someone else and marking it up.

An llms.txt file is a structured plain-text file that gives AI tools like ChatGPT, Perplexity, and Gemini a clear, organized summary of a website and the business behind it. It’s the foundation of serious AI search optimization work — and it’s something you can verify in about five seconds.

“Yes — here’s the URL.” They should be able to show you the file immediately. LGX’s is live at growwithlgx.com/llms.txt. You can look at it right now and see what a well-structured file looks like.

“We’re building that out.” “We do the equivalent of that.” “That’s a newer standard — we focus on the underlying principles.” These all mean no.

AI search optimization is measurable — not in the impressionistic way that “brand awareness” campaigns are measurable, but concretely. Before the work, your business either appears or doesn’t appear when relevant questions are asked of major AI tools. After the work, that should change.

An agency with real experience will have run discovery audits — systematic tests of how a business appears across ChatGPT, Perplexity, Gemini, and Claude using the questions real buyers ask. They’ll have documented the baseline, done the optimization work, and re-run the audit to show what changed. If they can’t show you a before-and-after — even an anonymized one from their own business or a past client — they haven’t actually closed the loop on the work.

A documented example — even anonymized — showing a specific business’s AI search presence before and after optimization. Ideally with screenshots from actual AI tools showing what changed.

“We track impressions and search rankings.” That’s Google optimization. “We monitor AI mentions of your brand.” That’s brand monitoring. The specific question is: did your AI search presence change, and can you prove it?

AEO — Answer Engine Optimization — is the practice of structuring content specifically so that AI tools can cite it in response to user questions. This is one of the most important and most misunderstood components of AI search optimization. AI tools don’t recommend businesses because they have good logos or compelling mission statements. They recommend businesses because those businesses have content that directly and clearly answers the questions users are asking.

FAQ architecture is the most direct way to build that content — but only if it’s done correctly. Correct means: questions are written the way actual users ask them, not the way a marketing team would phrase them. Answers are specific, declarative, and accurate. The content is formatted with FAQPage schema markup so AI tools can identify it as Q&A content. And the questions map to real AI query patterns, not just generic topics brainstormed in a meeting. Most agencies claiming AEO capability are writing FAQ content the same way they’d write blog posts — optimized for keywords, written to persuade, structured for humans. That’s not what AI tools cite.

A specific process: researching the actual questions buyers ask AI tools in a given category, writing direct Q&A answers, implementing FAQPage schema, and testing whether the content gets cited. They should be able to describe what “AI-friendly” content looks like structurally — not just say they write good content.

“We write high-quality content that’s optimized for user intent.” That’s traditional content marketing. The question is specifically about structure — how the content is formatted to be machine-readable and citable by AI systems.

Schema markup is structured code added to website pages that tells AI tools and search engines precisely what type of content they’re looking at. For AI search optimization, certain schema types are especially important: FAQPage, LocalBusiness, Service, HowTo, and Article. Without schema markup, an AI tool processing your content has to infer what your business is and whether your FAQ answers are actually answers to the questions you’re claiming to answer. Schema removes that inference — it explicitly declares what everything is.

Agencies doing real AI search optimization will have a systematic approach: auditing what’s in place, identifying what’s missing, implementing the right types for each page, and validating that the markup is correctly structured using Google’s Rich Results Test or equivalent tools.

A specific list of schema types they implement for businesses like yours, an explanation of why those types matter for AI citation, and a description of how they validate the markup is working — not just that it exists.

“We use structured data.” That’s technically accurate but says nothing. The question is what types, on what pages, and how they’re verified for AI citation specifically.

This one separates agencies that do a one-time setup from the ones who actually stay in the category. AI search is not a set-it-and-forget-it optimization. AI tools are updated constantly. How ChatGPT processes local business content today is different from how it will process it in six months. New AI tools emerge. Search behavior shifts. Competitors don’t stop moving.

Serious AI search optimization involves ongoing work: updating llms.txt files as services or locations change, adding new FAQ content as new customer questions emerge, running periodic AI discovery audits to catch gaps, and monitoring whether your business is being cited accurately. An agency that sells you a one-time AI optimization package with no ongoing maintenance plan has given you a good head start — but not a durable advantage.

A description of what ongoing AI optimization maintenance looks like — update cadences, re-audit schedules, content expansion plans, and how they stay current with changes in how major AI tools operate.

“Once we build it, you’re set.” “The optimization is evergreen.” Neither of these is true. The underlying content is relatively stable, but the AI ecosystem it’s designed to influence is not.

An agency that can answer all five questions specifically — with their own work, their own files, and real examples — has done the work. An agency that deflects, generalizes, or pivots to related topics it’s more comfortable discussing has not.

You can ask LGX all five right now. Our llms.txt is published at growwithlgx.com/llms.txt. We have before-and-after documentation from our own site. We have a defined FAQ and AEO content process. We implement FAQPage, LocalBusiness, Service, and Article schema across client sites. And AI optimization maintenance — including llms.txt updates and quarterly AI discovery audits — is built into our ongoing retainer packages.

We’re not the only agency doing this work with integrity. But we can answer all five questions, and we’re happy to walk through them in a 30-minute discovery call.

The fastest way to start is a free AI discovery audit. We query the major AI platforms with the questions your customers ask, show you what’s coming up right now, and tell you what it would take to change it. All pricing and service options are at growwithlgx.com/ai-search-optimization.