Do I need an expert for AI search visibility, or can I do it in-house?

You can run most AI visibility work in-house. The reason is where the evidence sits. The levers that hold up under scrutiny are access levers: per-bot robots.txt permission, firewall and CDN reachability, server-side rendering, index eligibility and snippet controls. Each one removes a block. Each one is mechanically checkable, and the engine operators document how they work.

The other stages are mostly out of anyone's hands. What a model knows without web access is frozen at its training cutoff. Which retrieved pages an engine cites is the model's call. A July 2026 critical survey of 45 GEO studies found that no reviewed technique shows a stable, cross-platform causal effect on discoverability.

So an expert can't buy you a better place inside the answer. What a good one can do is find and fix blocks faster than you would. The practical question is whether you have a developer with some time, and whether you know what to look for. What an AI SEO expert actually knows covers the skill set in more depth.

1. robots.txt, bot by bot. OpenAI's crawler docs say GPTBot governs training use, while OAI-SearchBot controls whether you appear in ChatGPT search answers, and each setting is independent. Perplexity uses PerplexityBot. Claude search uses Claude-SearchBot. The common mistake is a blanket AI block that removes search eligibility along with training. A site that wants ChatGPT search but no training use can write:

User-agent: OAI-SearchBot
Allow: /

User-agent: GPTBot
Disallow: /

2. The firewall and CDN. robots.txt can allow a crawler while your firewall returns a 403. OpenAI's help center says eligibility needs both: allow OAI-SearchBot and confirm your host or CDN allows its published IP addresses. From 15 September 2026, Cloudflare blocks Training and Agent crawlers by default on ad-displaying pages for new domains and existing free-tier customers. A site can lose access without anyone editing anything. The AI Crawler Checker requests a page as each AI crawler and shows which ones get turned away.

3. Rendering. Vercel found that ChatGPT and Claude crawlers fetch JavaScript files but don't execute them. Load your key pages with JavaScript disabled. If pricing, product names or headings vanish, those crawlers can't read them. Googlebot renders JavaScript, so this gap mostly affects non-Google engines.

4. Snippet and index controls. Search your templates for noindex, nosnippet, max-snippet:0 and data-nosnippet. These keep content out of AI answers, and they often arrive by accident from a template. Google says a page must be indexed and snippet-eligible to appear as a supporting link in AI Overviews or AI Mode.

5. Index status, engine by engine. Check Google Search Console and Bing Webmaster Tools separately. One being healthy tells you nothing about the other.

6. Fact consistency. List your core facts: prices, plan names, founding year, location, integrations. Check that every page states them the same way. A specific fact stated plainly on a fetchable page gives an engine a sentence it can quote.

Bring in help when a finding needs skills or access your team lacks. Typical cases:

• The rendering fix means rebuilding. Moving a client-side app to server-side rendering or prerendering is engineering work. Someone who has done it on your framework can save you a lot of trial and error.
• Your bot management is complex. Multiple CDNs, custom WAF rules and verified-bot settings are easy to break. OpenAI warns against allowlisting by IP alone, so someone has to validate user agents and verified-bot programmes properly.
• Engines repeat wrong facts from other sites. When engines cite other sites instead of yours, being named accurately on those sites is associated with getting into the answer. PR and review-site work can be worth outsourcing. Nobody can promise the effect.
• You need ongoing, per-engine reporting. Engines behave differently, so one blended score can hide a problem on a single platform.

If you already pay an SEO agency, first ask whether your current agency should already handle AI search. Smaller teams should read whether a small business needs an AI search service at all.

Push back when a pitch rests on claims that primary sources contradict:

• "Add schema to get cited." Google says structured data isn't required for generative AI search. Ahrefs tracked 1,885 pages that added schema against matched controls and found AI Overview citations fell 4.6%. Existing markup does no harm. It just isn't a lever for AI answers.
• "Publish an llms.txt." Google says Search ignores these files, and no other engine documents reading them.
• "We guarantee citations" or "first position in AI." Nobody controls what models say, and Google exposes AI impressions only, with no position data to check a rank against. See whether to trust an agency that guarantees AI citations.
• "We'll raise your E-E-A-T score" or "build domain authority." Google says E-E-A-T isn't a specific ranking factor. Domain Authority is a Moz metric that no engine computes.
• "Up to 40% more visibility." That figure comes from a lab engine with sources already loaded into context.
• "Allow GPTBot so ChatGPT finds you." GPTBot governs training only.

A credible expert talks about access, rendering and facts, and says plainly what they can't control. Take these vetting questions into the first call.

1. Run the robots.txt, firewall and rendering checks on your homepage, pricing page and top product pages.
2. Fix what's blocked. Most fixes are configuration changes, not content projects.
3. Write down your core facts and correct every page that contradicts them.
4. Take a snapshot of how AI answers describe you. The free checker on this site shows whether engines name you for up to three buyer questions.
5. Only then decide on outside help, scoped to the gap you found.

A scoped brief such as "server-render our pricing pages" is easier to price and easier to verify than "do GEO for us". If you do shop around, read how AI SEO agencies work and how to vet one first.

Can a marketer run these checks without a developer? Partly. A marketer can read robots.txt, review Search Console and Bing Webmaster Tools, and audit facts across pages. Rendering checks, firewall rules and template-level snippet tags usually need a developer, because the fixes live in code or CDN settings. The most efficient setup is a marketer who owns the checklist and a developer who spends a few focused hours on the technical items.

Should I block GPTBot? That is a training decision, not a visibility one. OpenAI's documentation says GPTBot governs whether content may be used to train its models, while OAI-SearchBot controls ChatGPT search, and the two settings are independent. You can block GPTBot and still allow OAI-SearchBot. Nothing in the evidence shows that allowing GPTBot improves how often ChatGPT search shows you.

Is an AI visibility tool a substitute for an expert? A tool replaces the measuring, not the fixing. It can show which engines name you, what they get wrong, and which crawlers are blocked. Someone still has to change robots.txt, adjust the firewall, or rebuild a page so its content is in the HTML. If your team can make those changes, a tool plus this checklist covers most needs.

What should a fair AI SEO engagement deliver? Named findings tied to checks: which crawler gets a 403, which page hides pricing behind JavaScript, which fact differs between your pricing and about pages. Then specific fixes and a way to confirm each one shipped. It should not promise citations, rankings or a score, because no one controls what the models say.