Optimize your AI shadow

Rose says: “Your website casts an AI shadow: a distilled version of your content that AI search systems present to users. It's often this shadow, not your site, that people now meet first.

It might look like a two-sentence summary, a list of bullet points, or an AI-curated comparison. It could be on ChatGPT, Perplexity, SGE, or any of the other LLMs or chatbots. Increasingly, though, this is a user's first engagement with your brand, before they ever reach your pages. Sometimes it's the only interaction they have with your brand.

This matters because it changes the unit of competition. For years, SEOs have thought in terms of positions and ranks, but in an AI-driven world, the real battleground is representation. How is your brand represented when an AI introduces you to the user? Is it accurate? Is it authoritative? Or (and this is happening more and more), are you simply fueling the answer while a competitor gets the credit?

SEO is evolving from being about rankings to being about representation signals. It's not enough to publish content and hope it ranks. You need to actively manage how your content is summarised, how your expertise is attributed, and how your brand shows up in that new AI shadow layer.”

How do you find out what's in your shadow, and what do you do about it?

“Something you can do is pick 25 to 50 topics and branded questions that are applicable to your company and your industry. Then, for those topics, check how you appear in Google AI overviews, ChatGPT, SearchGPT, Perplexity, and Claude.

Log whether you're cited. Log how you're summarised, and log which competitors are credited. Also, check for any errors or gaps, such as factual errors or places where there's shadow leakage, which is when your information is present but it is being credited to a competitor.

Then, you can begin to map each gap to a fix, such as adding more schema, increasing your off-site corroboration, or adding quotable snippets on your key pages.”

How can you determine when the content you've published is attributed to a competitor, and what can you do to change it?

“You can do this during your shadow audit.

If you run a hotel, for example, you can see whether any of your page info and what you offer is being listed, but being credited to a competitor.

The key way to combat this is to create highly original content, such as benchmarks, original studies, and things like that, which AI has to credit you with. It's very much about integrating your brand and your USP into your content, so that your content becomes an extension of your brand.

Make sure that your content isn't something generic that could easily be copied by other companies.”

You say that the signals that shape shadows are structured data, EEAT, entity consistency, and content clarity. How do you decide what to focus on first?

“It has to be a combination of things. This is where it gets practical. If you want to optimize your AI shadow, you need to think beyond the rankings and start engineering for summarisation.

For example, you need to write shadow-ready content. AIs love clarity. They need unambiguous structures that they can turn into neat answers. That means building FAQs or answer capsules into your pages. These are short two to three-sentence summaries right at the top that directly address the query, add the answers in plain language, and break processes down into steps or lists. Essentially, AIs are lazy, so you need to feed them the clearest possible material to reuse.

Next, you need to strengthen your machine-readable trust. Schema markup is essential because it helps AIs disambiguate your content and understand context. If you're a hotel, that means marking up reviews, pricing, and amenities. If you're a publisher, it means using article and author schema. If you're a SaaS company, it means using product schema. Structured data doesn't just win you rich results, it also makes your shadow more accurate and more likely to carry your name.

Another step would be to double down on EEAT. Generative systems are highly sensitive to authority cues. Make sure your authorship is clear and credible. Build consistent signals of expertise across third-party sites. Push reviews, citations, and mentions. Remember, the shadow doesn't just reflect what's on your site; it reflects the broader perception of your brand, and that's really important.

You also want to engineer attribution opportunities, which is where most people miss a trick. If your content is fueling answers but you're not credited, you've got shadow leakage. To plug it, you need to publish unique assets that AI can't paraphrase without citing you.

Then, the final step would be to inspect and iterate. Just as you check the SERPs, now you need to check the shadows. Search your priority queries in Google AI Mode, Perplexity, etc. Look at how you're represented and see if you're present. Are you credited? Is the description accurate? That's your shadow audit.

Fix the gaps, then keep rechecking.”

What kind of unique assets could a brand publish?

“It could be a proprietary survey, for example. Collect your own data and turn it into a study that can only be attributed to you.

Use original data. Use some of your own case studies or your own user-generated content – anything that is tied to you and unique to you.”

Is there any style, type, or length of content that is more likely to be picked up and used by AI?

“It goes back to framing the content in a way that is easy for an AI to scan.

Nothing too wordy or poetic, for example. Headings with simple questions and simple answers. Things like tables and comparison data are very useful – especially as we begin to move into the next stage of search, agentic browsing, where we'll be sending bots out to get the information for us.

You need your information to be clearly understandable by a machine, more so now than ever before. A good side effect of that is, if it's clear for a machine to read, it will also be clear and obvious for a user. It's not a choice between one or the other.

The things that machines look for (like an answer capsule at the top that summarises what's in the article) are actually very on brand with what users are looking for, in an age where everything is delivered very quickly. Our attention spans are shorter. We don't necessarily want to read a narrative essay about what a product offers. We want to get to the point, and that's very much what Google also wants to do.

You don’t have to choose between writing for AI and writing for humans. They go hand in hand, in that way.”

Should you publish content specifically to be published and consumed on a platform like ChatGPT?

“I think it's important to have an eye for that. However, it should be multimodal content.

It should be content that’s suitable across the internet. It used to all be about Google search, but now attention is spread across all these different LLMs – ChatGPT, Perplexity, Claude, etc.

It’s not all happening in one place.”

Should you try to publish in as many different places as possible, or should you focus on where your audience is most likely to interact?

“It very much depends on your industry.

For example, I work in the hotel SEO industry, so third-party corroboration is especially valuable from places like review sites, OTAs, and community pages.

You need to have an understanding of your user persona and where they go. Have a vague idea of their user journey, and then try to meet them at those separate digital touchpoints as much as possible.”

What are representation signals, and how do you measure them?

“Representation signals are essentially the cues that AI systems use to decide how to introduce you.

These might be things like facts, citations, author, entity credibility, recency, and corroboration across the web. Classic ranking factors decide where a page appears, but representation signals decide what the AI says about you and whether you get named and linked at all. There is a subtle difference.

Measuring these signals comes from the AI audit, and having a look through your priority prompts to see how you are represented in the LLMs.

Ask those key questions, type them in, and see if your name comes up. If it does, you are looking at how it's being described and how it's being explained. From there, you will see the gaps that you ought to be filling, the brand understanding that is missing, etc.”

Is there software that you can use to help you conduct an AI audit, and how often should they be carried out?

“The tools aren't quite there yet. There isn't a perfect tool for doing an AI audit, but I think that's being worked on.

The most thorough way to do it at the moment is manual. You pick your top prompts for your client or your industry, and you search them yourself. Don't choose just one LLM; choose a couple to see how you're appearing across different places.

You can do this semi-regularly, perhaps every month or so, just to get an idea of how you're coming across in those spaces and keep plugging those gaps.”

What does plugging those gaps mean?

“It depends on what the gaps are. It could be that AI is struggling to understand what's in your content, which could mean using more specific schema.

It could be improving your third-party corroboration. If there are gaps in your entity optimization, you need to make sure that there's consistent information on your brand across lots of different places. That can be fixed within the content. Essentially, you need to have a really strong About page, or brand fact area, that gives all the important data on your company: your address, USP, etc.

You need to make it super clear to AI who you are and what you're about. If there isn't that clarity, you'll be able to see it quite clearly when you do these searches and look at how you're represented in that shadow layer.”

Rose, what's the key takeaway from the tip you shared today?

“Don't just chase rankings. Engineer your representations.

Stop chasing vanity rankings for broad generic keywords. Shadows don't care about single keywords; they care about entities and authority. Focus instead on building content that feeds the AI accurate, trustworthy, and brand-consistent information.

Shadows can be shaped. Those who fail to adapt will face a visibility crisis or watch their authority drain into the shadows of their competitors.”

Rose Tero is a Senior SEO Strategist at Formula. Find out more over at thisisformula.com.