What off-site signals matter for AEO?

Google's guide to AI search says its generative AI features can surface what people say about a product or service elsewhere on the web, and it names blogs, videos and forums. So a page someone else published about you can end up inside an answer about you.

The same passage advises against chasing inauthentic mentions. The reason it gives is that the AI features rely on Google's ranking systems and its spam systems.

That guidance covers Google Search, and it sets no order of importance among kinds of pages. For other assistants we'd start from the same assumption, that a useful page is more likely to be cited than an empty one. Then we'd check that assumption against the sources each assistant shows in its answers.

A review matters when it holds detail a buyer couldn't get from your own sales page: what was tested, what it cost, where the product fell short. Google's guide makes a similar point about a site's own content, where it sets a first-hand review above a summary of what others have said. We apply the same test to pages about you. A row of one-line ratings gives a buyer less to go on than a single account that explains what the customer needed and what happened.

Local businesses have documented guidance from Google. Its local ranking help page names three main factors for local results: relevance, distance and prominence. A profile with full, correct details is more likely to appear, it says, and a larger number of reviews and good ratings can help a business's position in those results. That page describes Google Maps and local Search results. It sets no review count to aim for, and it isn't a rule for every assistant. Google's AI guide does say a Business Profile can help a business appear in AI responses as well as in other Search results.

All of this assumes the reviews are genuine. Ask customers for a review, and let them write what they think in their own words.

A listing earns its place through its facts: your name, your category, what you sell, what it costs, where you are. Those facts change, and a listing stays as it was until someone updates it.

Suppose a software directory still says your product connects to a tool you dropped last year, and shows a starting price you've since changed. A buyer asks an assistant whether you support that tool. The directory page says yes and your site says no. The assistant has two answers in front of it, and the buyer can get the old one.

Nothing was wrong with the listing when it was written. The fix is maintenance. Keep a list of the profiles you control, and update them when the product, the price or the address changes. We cover the wider set of profiles and feeds in what AEO covers beyond website content.

An article you contribute to an industry publication is your own explanation, published where your buyers already read. It isn't an independent review, and it doesn't need to pose as one. It earns attention the way any article does, by answering a question well.

That means it should say something a generalist couldn't. Google's guide distinguishes content built on common knowledge from content that carries an expert's own insight, and a contributed article should sit in the second group. It should also carry your name and company, so the reader knows who is speaking. If the placement was paid for, we'd want the page to say so. Readers weigh a publication's own verdict differently from a contributor's, and the label tells them which one they're reading.

Lean Labs' off-site AEO program shows how this work gets packaged. The listing covers expert articles each month on outside sites, mentions of the brand on other platforms, and reporting that tracks citations. It's priced at $5,000 per month for 10 months, which comes to $50,000 in listed payments. The company's on-site package, $9,000 one time with the first year included, has a system that collects customer reviews. The page's off-site step also names directory listings.

The expert articles are the specialist-publication kind of evidence, and the review system and the directory listings support two of the others. The citation tracking does a different job, because it records whether assistants cite any of those pages. A delivered article is a page that now exists, and the tracking is how you find out whether it gets used.

The four kinds don't carry equal weight in every market, and you can get a first look at which ones show up in yours. Put the questions your buyers ask to the assistants they use, and write down the sources each answer cites. Then sort the list. Some pages you can edit yourself, and some belong to publications you could contribute to. Other kinds include reviews and independent write-ups, where the author decides what gets said, and some sources will fit none of these.

That list shows the sources that appeared in the answers you sampled. It isn't a record of every page an assistant read, and it doesn't measure how much any one source counted. Treat it as a place to start work, and then watch the results. Citations go in one column, and visits, contacts and deals go in others, which we explain in how to track traffic and referrals from AI answer engines.

An assistant naming your company is also a different event from an assistant recommending it. The second depends on the buyer's request and on what the sources say about fit, which is the subject of what makes AI engines recommend one brand over another.

Run the same questions again after a listing is corrected or an article is published. You'll be comparing the same prompts before and after the change.