Google AI Overviews are built on the same index as Google Search. If a crawler cannot reach your page, or reaches it and finds an empty shell, nothing downstream can save you. This is the foundation the other four pillars stand on.
The short answer
AI-Powered SEO is the unglamorous half: can crawlers reach the site, does the server send real HTML, do the pages load, and does the internal link structure tell an engine what you are authoritative about. It is where we start when a firm is invisible for a reason that has nothing to do with content.
Every one of these looks fine when you visit your own site. All four are checked against what a crawler actually receives, not what your browser renders.
We read your robots.txt per user agent and then probe the site live as each crawler, because a firewall can challenge a bot your robots.txt welcomes. Search crawlers and training crawlers are separate decisions, and firms block the wrong one more often than you would expect.
We compare your server HTML against your rendered page. If your headings, copy and structured data only appear after JavaScript runs, the engines that do not document rendering are reading a near-empty document. This is a build question, not something more schema can fix.
Core Web Vitals, image delivery, render-blocking scripts and layout shift. Not for the score itself, but because a slow site gets crawled less thoroughly and refreshed less often, which shows up as stale content in the index the AI layer reads.
We map every page to the question it should answer, find the pages competing with each other for the same one, and fix the orphans nothing links to. This is how an engine works out which of your pages is the authority on a topic.
Measured against what crawlers receive, not what your browser shows you, with the evidence attached.
Where this sits on the ladder
Includes the crawler access gates and render-mode check, so the foundation questions get answered first.
A written 4-page memo naming what is wrong and what to fix first, in the order that actually matters.
When the foundation is the problem. A site that ships server-readable HTML, schema and answer structure from day one.
The technical checks are identical everywhere. What differs is which pages carry the weight and which platform constraints you are working inside.
Practice-area pages and attorney bios are the pages that have to be crawlable and distinct. Firms running several near-identical practice pages for different cities usually have them competing with each other rather than covering more ground.
Condition and service pages plus physician bios. Many practices run a patient portal on a subdomain that swallows link equity and confuses the entity picture, which is worth resolving before anything else.
Service and team pages. Gated content is the recurring issue: the substantive material sits behind a form where no crawler can read it, so the site looks thin to an engine while the firm has plenty to say.
Neighborhood and listing pages, usually on an IDX feed that generates thousands of near-duplicate URLs. Canonical handling here is the single highest-value technical fix in this field, and almost nobody has done it.
On a platform you cannot fully control, Wix, Squarespace, a proprietary IDX, we will tell you which fixes are available to you and which are not, before you pay for anything.
This page is about the engagement. These go deeper on how the mechanism works, and they are free to read.
Context
How the classic index and the answer layer on top of it actually relate.
ReadBuild
The foundation decisions that are cheap at build time and expensive later.
ReadMigration
The preservation checklist, if a rebuild is on the table.
ReadAI SEO is one of five. A full AI Visibility engagement covers all of them, because an engine checks more than one before it names a firm.
The audit probes your site as each search crawler, checks whether your server sends real HTML, and reports the access gates before it scores anything else. If the foundation is the problem, you will know inside a day.
No charge and no obligation. If a call is not the right next step, we will say so and send something useful instead.
or call 305-209-8453 · se habla español
Yes, and more directly than people assume. Google AI Overviews and AI Mode draw on the same index that powers Google Search, so a page the index holds poorly cannot appear in an AI answer either. The classic work has not stopped mattering. What has changed is the goal: you are no longer optimizing purely for a click, you are making sure the page is reachable, readable and clearly about one thing, so the layer sitting on top of the index can use it.
It is one of the most common cases we see. A site built as a single-page app renders in the browser, which means the HTML sent to a crawler can be close to empty: a root div, a bundle of scripts, and little readable text. Google can process JavaScript when it is not blocked, but OpenAI, Perplexity and Anthropic do not document rendering behaviour for their crawlers, so the safe assumption is that the server HTML is what they read. A polished React site can be fully visible to people and nearly blank to three engines out of four.
Check per user agent rather than in general, because they are governed separately. The search crawlers that decide whether you can appear at all are Googlebot, Bingbot for Copilot, OAI-SearchBot for ChatGPT search, PerplexityBot and Claude-SearchBot. Training crawlers like GPTBot and Google-Extended are a different decision and blocking them does not remove you from search results. The trap is a CDN or firewall quietly challenging a crawler your robots.txt allows, which only a live probe reveals.
It is a legitimate choice, and it is worth understanding what it does and does not do. Blocking GPTBot opts you out of training. It does not remove you from ChatGPT search results, which is governed by OAI-SearchBot. Firms sometimes block both by accident and then wonder why they vanished. We will lay out the trade-off and implement whichever way you decide, but the two decisions should be made separately.
Indirectly but really. Crawlers have finite budgets, and a slow or unstable site gets crawled less thoroughly and re-crawled less often. That shows up as stale or missing content in the index the AI layer reads from. We are not chasing a perfect score for its own sake. We are making sure the pages that matter load fast enough to be fetched reliably.
A page nothing else on the site links to. It can be excellent and still be invisible, because engines build their picture of what you are authoritative about by following links between your pages. A service page with no inbound internal links reads as unimportant regardless of its content. It is one of the cheapest things to fix and one of the most commonly missed.
We work with what is there wherever that is honest. Schema in a Week and the AI Visibility Intensive are both built to improve a site you already own without touching the design. When the foundation is the actual problem, a client-rendered app with no server HTML, we say so rather than layering fixes on something that cannot carry them, and the conversation becomes a Website Build instead.