Produce meaningful research to stand out in AI search

Andreas says: “Produce scaled insights in your content, to get your website AI-recommended.”

Is that the same as scaling content production?

“No. Scaled insights is about producing research that has achieved a significant sample size, which means that the insights contained therein, within your content, have significance.

That's really important for AI because it means that it's likely to generalise well, and whatever you're describing is likely to apply in the real world.”

Is it more linked to enhancing your own personal and brand authority?

“That's a very succinct way of putting it.”

How do you go about doing this?

“Let’s first discuss how you don't go about it.

The first thing is that, if you're using AI to produce research for your content in the hope that it will get featured or surfaced in AI results, that's not going to work. Even if you use AI on Deep Research Mode or pay for the $200-400 option, that's still not going to work because the first issue is that AI already knows about it.

Trying to tell AI something it already knows is not going to work. It’s not going to cut it. You need to produce something that provides ‘information gain’ – that favoured phrase.

It has to be something AI doesn't already know about. It has to add value. It has to be something that AI deems will help it become smarter, as a result of using your content. It must be something that adds value to its existing models.

That is what's going to position your company, brand, or non-profit as an authority on the given subject that you want to be featured or cited for as a source in AI.”

How does AI determine that your content adds value to what already exists out there?

“AI is already modelling conversations that are happening online.

The way AI works currently is it's synthesising memory. If it compares what you're producing to what it has in its modelled memory, then it will quickly be able to check whether it correlates – or how likely what you're producing is to be true and how likely it is to generalise.

On that basis alone, it can tell whether you're providing something truly game-changing or you're just another also-ran trying to get into AI.”

How do you get AI to start considering your content, to begin with?

“There is plenty of documentation online showing that, if you're in the top 20 in Bing (and more recently Google), you're in the running for AI results.

While SEO is not what it was, the SEO basics absolutely matter. Your content still needs to be discoverable and parseable, etc., but that's something everybody can do pretty easily.

After everything is technically optimized, the real game-changing feature here is whether your content tells the world something new. This is where scaled research insights come in.”

Are experience, expertise, authoritativeness, and trust less essential now?

“It's more the case that EEAT very much matters, but it now needs to be enhanced via scale. It needs to be more data-driven, and it needs to be data-driven at scale.

It's not that EEAT isn't important anymore – it's more important than ever – it just needs to have scaled data research behind it. That's what's going to get you into AI.”

Why do you believe that LLM outputs won't cut it, even in Deep Research Mode?

“If you're attempting to fake EEAT, or scaled EEAT, to get into AI, it's not going to work because AI generally hallucinates. It's just that, 80-90% of the time (assuming that we're asking it about something it would know or is known to the internet), the hallucinations happen to be useful and correct.

However, if you're using those hallucinations, or the outputs of AI, to try and add EEAT, it's not going to work because those outputs are derivative, diluted, and summarised. It is something that AI already knows about, so you're not really adding value to AI by telling it something it already knows.”

If you use AI in any form to enhance your content, are AI search engines less likely to feature that content?

“I would say so.

I want to make a distinction here. You can absolutely use AI, or a form of machine learning, to help you make sense of the data that you collect from your research. That's a legitimate use of AI, but using LLMs (AI platforms like ChatGPT) to do the research for you and make your content authoritative and expert, etc., is not going to work.

If you're using AI to help you produce rigorous statistical conclusions that then get used when it's written up by a human in your content reports or insight feature article, then that's fine.

However, for me, using AI to write content is a massive no-no.”

What type or structure of content have you had success getting featured within AI search engines?

“The kind of content that I find has worked really well, and got clients recommended within 90 days, is the type of content where you produce reports that the target buyer can learn from.

The internet has seen more than its fair share of ‘What is Target Topic X?’ Your target buyers already know what that target topic is. That content looks like it's aimed at every single country in the world, every single persona type – students, retired people, etc.

AI search has raised the quality bar and the threshold. Producing reports that your target buyers can learn from, that's a big one.

Where do you get the data? Well, we're data scraping the internet on a daily basis, and we're able to subset those conversations for the target buyers, and then do some data enrichment to maximise the confidence interval.

That is what AI seems to really like, and it's pretty reliable. I've never seen it fail – and that comes at a fraction of the cost of going to Gallup or YouGov and taking out surveys. That could become very expensive very quickly, but that will work.”

If you produce content for your target buyers through this extensive research, will that naturally be appealing to LLMs?

“That's a really great question. ‘What is Topic X’ SEO-style guides were primarily driven for search. Whereas, if you're producing reports for target buyers, not only does it have value for AI search and traditional search engines, but it also has value in social media because you're discussing things that your target buyers are truly interested in.

You can even make videos of individual sections within the content reports that you're producing. It can go not just on LinkedIn (assuming you're B2B), but it could go on YouTube, Instagram, TikTok, or wherever your target buyers hang out.”

Does producing high-confidence, low-error-margin surveys, studies, etc., give LLMs confidence that your content is better than anything else that exists out there?

“100%.

If we could contrast that with using LLMs as a content research or insights tool, you would never get the raw data, or anything meaningful and scaled, from an LLM output. It's designed to give you a summary as opposed to the 10 blue links.

Whereas, if you're getting your own data, then it's not just enough to have the data; you have to do something with it. This is where we do a lot of cross-validation to make sure whatever we collect is likely to land in the real world. When AI can see that, it passes those fact-checking algorithms and, therefore, it's seen as something that is value-adding and worthy of being cited in response to a prompt by a target buyer.”

Does AI build confidence that your domain is likely to be authoritative in a particular sector, or does it take each piece of content on its own merit?

“I think it builds confidence. I built my own LLM five years ago, and what I learned is that AI search is really a synthesis of memory. The more you feed it, the more of a memory it has, and the more it would score your domain or brand as knowing more about a certain area.

There'll be greater weight placed on your domain when it comes to a certain domain of knowledge/expertise. Yes, each piece of content is taken on its own merit, but there's a compounding and cumulative effect.

If we were to talk in SEO terms, it's a bit like Domain Authority and Page-Level Authority. Obviously, the page has to stand out on its own merit, but the more of those you have, the more the overall domain benefits – even for pages that have a lot of intrinsic authority at the page level.

In the age of AI, you want to stay in the target buyer’s lane. That's where you want to be. Whatever your target buyers are interested in, that is the lane to go in. If they're discussing it, then it's relevant and you want to be covering it.”

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

“The key takeaway is: you want to be conducting research on what your target buyers are discussing.

If you can do that, then you've got the raw ingredients for getting your content featured by AI.”

Andreas Voniatis is the CEO at Artios. Find out more over at Artios.io.