Get Your SEO and AEO Baseline

Search engines solved that problem by doing three jobs, in this order. This is the model the rest of the guide hangs on.

Discover. Pages are created faster than any human could submit them. A search engine needs a system that goes out and finds pages on its own.

Categorize. Once it has found the pages, it has to work out what each one is about. When there are 3,000 home and garden pages, it knows they belong together. This is what turned the internet into a searchable database.

Qualify. Those 3,000 pages cannot all be good. Which one gets shown first? This is the hard job, because quality is subjective and the search engine has to measure it mechanically.

One reframe before moving on. Google is not the internet. Google is a copy of the internet that it lets you query. It copies pages so it can categorize and rank them. That explains almost everything that follows.

Google was not the early leader. AltaVista, Lycos, Excite and Yahoo were all reasonably good at discovering and categorizing. Where they stumbled was qualifying.

Google's insight was to treat the web like a body of research papers. In science, you judge a paper's credibility partly by how often other papers cite it. The web has its own citation: the backlink, the blue link in a page that sends you somewhere else.

A backlink is a meaningful act. If I link from my content to yours, I am telling my own visitors to leave my site and go to yours. That costs me something, so it had better be worth it. Google made that act of judgment the backbone of its ranking system.

Once search engines got popular, every marketer asked the same question: how do I get my website in there? That question is the birth of SEO.

SEO is reverse-engineering how a search engine works and then making your site more favorable than the competition's. It maps onto the three jobs. You want to be favorable when the engine discovers, when it categorizes and when it qualifies.

Early on, all three could be gamed. Sites were built in messy ways that search engines could not read, so simply getting indexed was an advantage. Categorization was gamed by stuffing a phrase onto a page a hundred times, sometimes in text the same color as the background. Qualification was gamed with networks of sites linking to each other.

Now add AI assistants: ChatGPT, Perplexity, Gemini, Claude, Copilot and Google's AI Overviews at the top of search results.

The behavior underneath has not changed. A person comes to a system, asks a question and expects an answer. The interface is different. The intent is the same. The same person often uses Google in the morning and an AI assistant in the afternoon.

An AI assistant answers in one of two ways:

• From what it learned in training. The model has read a large portion of the web and answers from memory.

• From a live web search. When the model is not confident, or the question needs current information, it runs a search, reads the results and writes an answer from them. In that moment it behaves like a user on a search engine.

You do not need to know every tool. You need to know what kinds exist and what each is for.

Most reports bury you in numbers. These are the ones that matter, in the order they matter.

Two numbers you will be shown that deserve less attention:

• Site health scores. A single grade for your whole website. Google does not grade your website. It matches individual pages to individual searches.

• Domain authority scores. A third-party estimate of a whole site's strength. Search engines rank pages, not sites. One strong page on an average site can outrank a weak page on a famous one.

A strategy in this channel is an answer to one question: how do your customers search for what you sell, and which of those searches should you win?

Most businesses cannot answer it. They describe themselves in their own language: branded product names, internal jargon, the terms used in sales meetings. Customers search in plain, general terms. A business that only uses its own vocabulary is invisible to the people using everyday words.

Building the answer takes four steps.

Learn the business first. Before anyone opens a tool, they should talk to the people who hear customers every day, especially sales. What do you sell, to whom, against which competitors, in which locations? Which customers do you want and which do you not want? Keyword research done without this is guesswork.

You cannot tell whether work is paying off unless you know where you started. Before changing anything, record your starting position in both channels.

For search, the keywords from your landscape go into a rank tracker, organized by intent group. It records your position for each keyword, and your main competitors' positions too.

For AI search, each intent group is represented by a small set of prompts that ask for the same thing in conversational language. More than one, because phrasing varies and answers vary. A single prompt drifting in or out of an answer should not swing your numbers. An AI visibility tracker runs those prompts and records whether your brand appears in the answers, and whether competitors do.

Now you have one map of customer needs with two scoreboards against it. For any customer need, you can see where you stand in Google and where you stand in AI answers. Every later effort is measured against this snapshot.

There are three tactics, and they follow the three jobs of a search engine. The order matters, because each one is wasted until the one before it is handled.

Indexation means your pages have been found and stored by the search engine. This is what most people mean by technical SEO.

It matters far less than it used to. Google has become very good at finding pages, and it wants more pages in its index, not fewer. This is one of the few areas where the search engine is working with you.

So the check is short: are the pages you care about indexed? It is close to a yes or no question. If yes, move on. If a page is indexed and not ranking, the cause is relevance or authority, not site health.

Reporting is the best protection you have against poor work. It answers two questions: did they do what they said, and is it working?

Track efforts, not just numbers. Every piece of work should be logged with a date and a scope: "updated titles and headings on these ten pages in March," "built links to this page in Q2." Call each one an initiative. Without that log, a rankings chart shows movement with no cause attached. With it, you can say what was done, what happened and what to do next.

Read the metrics in sequence. Results arrive in a fixed order:

• Rankings move first. A page might go from nowhere to position 30. That is progress, but nobody sees page three. The goal is page one, then the top three, where most clicks happen.

• Clicks and impressions follow. Once you reach the middle of page one, Search Console starts showing real traffic.

• Conversions come last. Visitors start doing what you wanted.