Needs to every marketing professional be putting 80 % of their effort right into AI visibility? Should we be throwing billions at tracking AI suggestions? Should firms restructure their budget plans around showing up in AI tools? Possibly not.

Three new research studies came out in the last week showing us some genuinely concerning information:

1. GEO is a Diversion from Marie Haynes– https://www.mariehaynes.com/geo-is-a-distraction/
2. How much website traffic do AI Assistants send from SEOWorks UK– https://www.seoworks.co.uk/downloads/ai-referral-traffic-benchmark/
3. Does AI drive even more direct web traffic from Kevin Indig– https://www.growth-memo.com/p/does-ai-drive-more-direct-traffic

The verdict: Web traffic from AI assistants (ChatGPT, Claude, CoPilot, Deepseek, Perplexity, etc) is almost non-existent. And worse, even if you represent dark traffic and sessions that AI referrals influence indirectly (e.g. an AI suggests a brand and you later on Google that brand or type in the domain), it’s vanishingly small.

What we believe: AI is, probably, quite influential on some buyers in some markets. As Mike King intelligently kept in mind in a follow-up to these, AI is much more like a brand network , a digital billboard if you will, and it ought to be invested-in and gauged because of this. This week’s 5 -Min White boards goes into the three researches, their final thoughts, and the risk that lots of marketing experts may be taking by buying AI like a website traffic channel.

Records:

Is AI exposure a lot of malarkey? Is it a marketing boondoggle? There’s three current researches that suggest maybe.

So, this is from Marie Haynes; she composed on October 21, that “GEO is a disturbance.” So right here’s what she took a look at. Twelve months of information throughout 100 different sites and whether they got any kind of significant AI website traffic. And the response is 0. 62 % from ChatGPT. If you consist of all LLMs, it’s 0. 73 That is much less than DuckDuckGo. If you’re consumed with AI web traffic, you better be consumed with DuckDuckGo web traffic too, because boy, howdy. You know, we’re discussing a few dots out of out of a thousand.

Conversions. Well, conversion price, yes, ChatGPT site visitors on all these websites she considered were greater, yet Google still draws 97 % of the complete conversions, 27 times more than all LLMs incorporated. Again, vanishingly tiny.

Marie created this graphic taking a look at essentially what was anticipated during those years of 2223, 2425, like that expected growth contour. And afterwards what’s really taken place over the last year, year and a half.

Once again, it just does not look great. I assume this is why her verdict is that AI presence, what she calls GEO, is a distraction. Okay. Another study.

How much website traffic do AI assistants in fact send out? Does it convert better? I understand you’re about to ask, well, it’s not concerning web traffic. Do not worry. I’m going to obtain there too.

This study took a look at a different part of web sites and again, natural search 39 times even more traffic, AI web traffic converted even worse than natural on 23 of the 63 internet sites that search engine optimization Functions considered.

They’re looking at 95 sites, so virtually the very same number that Marie is, twelve months from September 2025 to to August 2026 You you can see once more, same conclusion that, boy, if you’re counting it on a traffic basis, AI just does not look wonderful.

Kevin Indig did something various with his study a couple of days ago, just released on Internet search engine Land as well, but he released in his Development Memo e-newsletter. And he is taking a look at at essentially the concern of what if website traffic sent from LLMs is not the number, those like 0. 5 or 0. 6 of traffic. Possibly that’s not the number. Maybe what occurs is individuals to an AI, they become aware of a brand name, and then they go to Google and search for it, or they key in the domain name directly, and they do that with greater tendency than they used to because of the LLM’s answer. So you can undergo his entire report. He actually, truly attempted here to discover a silver bullet.

The correlation initially maybe looked excellent, but he’s unsure. And, and so here I’m gon na read you the answer. I found that the quantity of straight traffic to long tail pages in my example driven by AI is not observable yet. We can not yet locate a trustworthy between AI exposure and top quality search or straight website traffic.

Some business with impressive AI development could be able to see it, yet throughout the 12 websites in various sectors, he couldn’t discover it. Now, I did do a little bit much more research dug around in Gemini and Google’s AI mode. I found the most effective that AI might create in regards to various other research studies were points such as this, which doubtful. They are case studies in that they declare to show rises in exposure, however no boost in conversions or sales or web traffic.

There are things such as this. This one’s in fact from Go Fish Digital, who I’ve I have actually long admired and thought extremely highly of. They don’t say that the company is, yet they assert that there was considerable development in AI presence and considerable growth in conversions from the AI references. We just do not understand what the standard is. So if that baseline is small and Google’s still sending, understand, 40, 50, 30 times as much, well, how much is truly going on right here? Perhaps Marie is best and GEO AI exposure is a distraction.

I intend to dig into this much deeper, yet I thought that I should appear these three studies because they’ve all appeared in the last week and they are all asking a pointed inquiry, which is, have we been chasing our tails below? Is AI exposure a boondoggle? Should you be reconsidering where you place your initiative? Is that most likely Google if you appreciate search and website traffic and very well might be.