AboutWorkServicesToolsBlogContact Get in touch โ†’
SEO

AMA: Real Questions From My AI-SEO Audits

Every schema, GEO, pricing, and timeline question I actually field on audit calls, answered the way I answer it live, not the brochure version.

I keep a running doc of every question a client or a founder asks me mid-audit. After enough calls, the same fifteen or so questions kept resurfacing, just phrased differently each time. This is that list, cleaned up and answered the way I actually answer it on the call, not the brochure version.

I run the structural side of this work the same way across every engagement, the exact four-pass sequence is in the pre-launch SEO/AIO/GEO/AEO/SXO/LLMO checklist. This post is the Q&A layer underneath it, the stuff people ask once they see their own site through that checklist.

Schema and structured data

Do I really need FAQPage schema on every page, or just a few? +
Only on pages with a genuine question-and-answer shape. I don't add it to a homepage or a pure landing page just to have more schema on the site. Forcing FAQPage markup onto content that isn't actually FAQ-shaped is the kind of thing that gets a rich result manually suppressed.
What's the single most common schema mistake you find?
Schema copy-pasted from a template with placeholder values nobody swapped out, "Your Site Name" sitting in a live publisher field, or a sameAs array pointing at an example.com social link. It's an easy miss because nothing visibly breaks on the page, the mismatch only shows up when you actually run the JSON-LD through a validator or read it line by line.
Should I use Organization schema or Person schema for a personal brand site?
Person, if the site is built around one named individual rather than a company. I use Person schema on johnarce.com with a sameAs array pointing at my real social profiles. Organization only makes sense once there's a company name buyers actually search for separately from your own.
Does BreadcrumbList schema actually matter for AI citations?
It matters more for how search engines understand site structure than for citation rate directly. I still add it to anything with a URL depth greater than one. It's low effort, it's never caused a problem, and it reinforces the entity hierarchy the other schema types rely on.
How often should I re-validate schema after it's live?
Every time the visible content on that page changes. I've seen schema go stale within weeks because someone edited an FAQ answer in the CMS and the JSON-LD block sat untouched in a template. A quarterly full-site sweep catches what day-to-day edits miss.

GEO, AI Overviews, and agentic commerce

What is GEO, actually, in one sentence?
Structuring and verifying content so a generative model, ChatGPT, Perplexity, Google's AI Overviews, cites it directly instead of only ranking it. I covered the full acronym split in how to dominate SEO in the age of super intelligence, but that's the one-line version.
Can you guarantee my page gets cited in AI Overviews?
No, and anyone who says yes is guessing. What I can commit to is removing every structural reason a model would skip the page, broken schema, buried answers, inconsistent entity naming. Citation after that depends on competition in that query and factors outside any single page's control.
Is agentic commerce relevant if I don't run an ecommerce site?
Not directly. It only applies once shopping agents inside ChatGPT, Gemini, or Perplexity are selecting products off your Schema.org Product markup and Merchant Center feed. If you're a service or content site, the content-structure and entity work matters far more than the commerce-specific checks.
How do I know if AI search is already sending me traffic?
Build a custom GA4 channel group matching referral traffic from chatgpt.com, perplexity.ai, and gemini.google.com, then cross-check Search Console's Generative AI performance report for impressions that aren't converting to clicks. Most sites I audit have never set this up, so the honest first answer is usually "we don't know yet."
What clients ask aboutWhat I actually checkWhere I answer it in depth
"Are we visible in AI search?"GA4 AI-referral channel group, Search Console Generative AI reportHow I get AI Overviews to cite my pages
"Is our schema good enough?"Rich Results Test against every page type, manual check for visible-vs-markup mismatchThe AI visibility audit framework
"Should we rename our SEO role?"Nothing on the audit itself, this is a positioning question, not a technical oneWhy I stopped calling myself an SEO
"Are AI crawlers even hitting our site?"Raw server logs filtered for GPTBot, ClaudeBot, PerplexityBot user agentsWhat 6 months of GPTBot, ClaudeBot & PerplexityBot taught me

Pricing and timelines

What does an AI-SEO audit cost?
It depends on site size and whether it's a one-time audit versus an ongoing retainer. I keep my actual rate structure public rather than hiding behind "custom pricing," the full breakdown is in the GEO pricing post I published after getting this question on nearly every call.
How fast will we see results after fixing audit findings?
Crawl and indexing fixes show up within days once resubmitted through IndexNow or Search Console. Citation and ranking changes take longer, usually several weeks, since models and search indexes don't re-evaluate authority on every crawl. I set that expectation on the call so nobody's watching for a week-one spike that isn't coming.
Is a one-time audit enough, or do I need an ongoing retainer?
A one-time audit is enough if your site is stable and you have someone in-house who can act on the findings. A retainer makes sense once you're publishing regularly, schema and structure need re-checking on every new page, not just once at launch.
Do you offer a free initial check before a paid audit?
I'll look at crawlability and basic schema validity on a call for free, that's maybe fifteen minutes of real signal. The paid audit is the four-pass sequence: crawlability, entity and schema, answer structure, and verification, plus a written scorecard you can hand to a developer.
๐Ÿ’ก
The question behind the question

Almost every pricing question is really a risk question: "what happens if I pay for this and nothing changes." I answer it by scoping the audit as a fixed deliverable, a scorecard and a prioritized list, separate from any implementation work. You know exactly what you're getting before you commit to the next step.

Content, entity, and process questions

What's "entity consistency" and why does it come up in every audit?
It means your name, brand name, and bio read the same way everywhere, your site, LinkedIn, old press mentions, guest posts. Models resolve who's talking before they decide whether to trust what's said. I find at least one inconsistent variant on almost every site I audit, usually a shortened name from an old bio nobody updated.
Why do you keep saying "answer-first" like it's the whole game?
Because it's the single change I've seen move citation rate across every AI surface at once, not just one. Two direct sentences under a question-shaped heading, before any throat-clearing. It costs nothing to implement and most sites still bury the actual answer three paragraphs deep.
Can an AI audit a site without a human reviewing the output?
Parts of it, yes, I run scheduled routines that check schema validity and flag stale content automatically. The judgment calls, what to prioritize, whether a flagged issue actually matters for this specific business, still need a person. I wrote about where that line sits in I automated my SEO checklist with Claude routines.
How do I know if my content is "AI slop" versus genuinely useful?
If you can't point to a specific number, test, or decision behind a claim, that's the tell. Generic advice anyone could've generated, three vague benefits, no actual process, reads as filler to a human and gets deprioritized by models trained to weight specificity. I'd rather publish one real experiment than five generic "best practices" posts.
What should I fix myself versus hire someone for?
Fix yourself: obvious crawlability blockers, missing alt text, broken internal links, anything a free tool flags with a clear fix. Hire for: prioritizing a long findings list by actual impact, writing schema that matches your specific content types correctly, and the ongoing discipline of re-auditing after every content push.

Key takeaways

  • Most audit questions aren't technical, they're really about risk, timelines, and what to prioritize first.
  • Schema validating in a tool and schema matching the visible page are two different checks, I catch mismatches on nearly every audit.
  • GEO and SEO are the same underlying work measured on different surfaces, not separate strategies competing for budget.
  • Answer-first structure is still the highest-leverage content fix I've found, and still the most skipped.
  • A fixed-scope audit with a written scorecard removes most of the "will this actually work" hesitation I hear on calls.

What I'll do next

I'm turning this list into a living doc I update after every audit call instead of rewriting it from memory months later. I'm also tagging which questions come up most by client type, agency versus founder versus in-house marketer, since the pattern so far suggests the pricing questions cluster heavily with first-time clients and the schema questions cluster with anyone who already has a developer on staff.

FAQ

How long does an AI-SEO audit actually take? +
One to two days for a site under 50 pages, covering crawlability, schema, and content structure. Add a half day once fixes go live, I always re-check the real URL, not the plan. Larger sites take longer mostly because the content-restructuring pass scales with page count.
What do you check first, before anything else? +
Whether a crawler can actually read the page with JavaScript off. I fetch the raw HTML with curl and compare it to what a browser renders. If the main content only shows up after client-side rendering, every other finding on the audit is secondary until that gets fixed.
Do I need separate strategies for Google AI Overviews, ChatGPT, and Perplexity? +
No. The underlying work is the same: valid schema, consistent entity naming, answer-first structure. What differs is which surface cites you more, and that comes down to content type more than a surface-specific trick. I wrote the full breakdown in how to dominate SEO in the age of super intelligence.
My FAQPage schema validates in the Rich Results Test. Why would it still get flagged? +
Validators only check syntax, not whether the schema matches the visible page. I've caught FAQPage JSON-LD listing six questions when the live accordion only shows four, usually because someone edited the copy and never touched the script tag. Google treats that mismatch as grounds to suppress the rich result.
Does GEO replace SEO, or run alongside it? +
Alongside it. Answer-first content, clean schema, and fast indexing improve classic rankings and AI citation rates at the same time. I've never found a fix that helped one and hurt the other. Treat GEO as a new way of measuring work you should already be doing.
Share this

Want this built for your team?

I design AI agents and growth automation that run without babysitting. If that sounds useful, let's talk.

Get in touch โ†’

Keep reading