Is AI visibility worth it? I think it is, as long as you stop trying to justify it with referral traffic, because on traffic alone the numbers that came out over the last month are small enough to get the line item cut. Across 100 websites and more than 145 million sessions, ChatGPT sent 0.62% of the traffic Google organic sent (Marie Haynes), and that number has been making the rounds ever since.
Rand Fishkin asked the question out loud in a post called "Is AI Visibility a Marketing BoonDoggle?" and his answer to the traffic version of it was "probably not" worth a big budget (SparkToro). Then Search Engine Journal pulled four of these datasets into one column on November 6, and the argument there is that the visibility is real and most of the ways we measure it don't hold up (Search Engine Journal). I mostly agree with both of them. If you run marketing for a brand, somebody above you has probably seen the 0.62% figure by now, so I want to go through what the data says and where I come down.
The short version
On traffic, AI assistants are small next to Google. Organic search sent 39 times more traffic than AI assistants across 95 sites in one benchmark (The SEO Works).
Conversion rates are all over the map, from more than four times better than Google organic to a little worse, depending on whose sites you look at.
Nobody has found a reliable link yet between AI visibility and later direct or branded search, which is where a lot of us assumed the hidden value was sitting.
So I'd treat AI visibility as a brand and consideration channel, measure it like one, and fix conversion tracking before arguing about any of it.
What the traffic data says
There are three datasets worth knowing, and they're pretty consistent with each other even though they come from different sites and different countries.
Marie Haynes looked at 12 months of GA4 data from 100 websites. ChatGPT came in at 0.62% of Google organic's volume, and adding every other LLM only moved it to 0.73%. Google drove 96.5% of the total conversions across the channels she analyzed, and Google organic drove 27 times more conversions than all of the LLMs combined. She also found the steep growth in ChatGPT sessions ended in August 2025, around the GPT-5 release (Marie Haynes).
The SEO Works ran a benchmark on 95 GA4 properties from September 2025 to August 2026. AI assistants sent 2.6% of the traffic organic search sent, and that share crept up to about 3.9% by August (The SEO Works). Their AI numbers leave out Google AI Overviews, which matters, and I'll come back to it.
Salience, a UK agency, ran its own client data and got AI assistants at 1.2% of Google organic sessions, with Google organic producing 91 times the purchases (Salience). They also saw AI sessions grow 4.4 times year over year while Google organic sessions fell 17.9%, so the trend line is going the right direction for AI, just from a very small base.
So somewhere between under 1% and around 4% of what Google sends. If you were hoping the traffic would pay for the program on its own, it won't, at least not yet.
How big is AI traffic next to Google?
Three datasets from fall 2026. Switch views to see traffic next to conversion rate.
Sources: Marie Haynes (100 sites, 145M+ sessions, 12 months; conversion rate from the 55 sites tracking conversions), The SEO Works (95 GA4 properties, Sep 2025 to Aug 2026, excludes AI Overviews), Salience (agency client data, Oct 2025 to Sep 2026).
The conversion rate argument, and why the baseline matters
The pushback you'll hear is that AI visitors convert better, so the small numbers are worth more than they look. That's partly true and it depends a lot on whose data you pick.
In the Haynes data, ChatGPT visitors converted at 13.8% against 3.02% for Google organic. That's a big gap. Still, only 55 of her 100 sites were tracking conversions at all, and the other 45 had no key events set up (Marie Haynes). My guess is that ratio isn't far off for the industry as a whole, which is its own problem.
The SEO Works numbers are tighter. AI converted at 7.18% against 6.08% for organic, so about a fifth better overall, and AI converted worse than organic on 23 of the 63 sites with enough traffic to compare. For B2B product companies, the pooled AI conversion rate was 0.62 times the organic rate (The SEO Works). Salience's client portfolio had AI slightly behind, at 2.04% against 2.22% (Salience).
The SEO Works also did a bit of math I'd urge you to run on your own site. On their average site, doubling AI traffic added about 6.6 conversions a month, and growing organic by 5% added about 10.3. So a small lift on the big channel was worth more than doubling the small one, and that's why I wouldn't lead a budget conversation with the conversion-rate argument, even though I'd still want the AI numbers in the deck.
Double AI traffic or grow organic 5%?
Put in your own monthly numbers. The starting values use the pooled conversion rates and AI traffic share from The SEO Works benchmark on a site with 10,000 organic sessions a month.
Benchmark values: The SEO Works AI Referral Traffic Benchmark (95 GA4 properties, Sep 2025 to Aug 2026). AI sessions set at 2.6% of organic; conversion rates 7.18% AI and 6.08% organic. The 10,000-session site is an example, not a figure from the study.
The hidden influence nobody has found yet
This part bugs me a little, since it's the argument I would have made. The theory goes something like, "Sure, the clicks are small, but people see us in ChatGPT, then come back later by typing our URL or searching our name." That would show up as direct traffic and branded search.
Kevin Indig went looking for exactly that. According to Search Engine Journal's summary of his work, he ran 632 correlation tests across four brands, using data from 12 GA4 properties, and found no reliable link between AI visibility and later direct or branded search. Total direct traffic on his sites rose 140.9% from January to September while engaged direct traffic moved by about minus 1%, which points to bots. When he lined up the data against the wrong dates on purpose, the correlations looked about as strong as the real ones (Search Engine Journal; Indig's original is at Growth Memo).
I wrote in October about ChatGPT traffic showing up as direct and recommended building a direct-traffic baseline for the pages that used to get ChatGPT referrals. I still think that's right. Indig's numbers are a good reason to filter that baseline down to engaged sessions, because a raw direct line going up and to the right can be mostly crawlers.
Now, the tools we have are pretty bad at finding this kind of effect, so I wouldn't call it disproven, though I also wouldn't promise a CFO it exists.
Where the buying decision happens
So why do I still think it's worth it? Mostly it comes down to the referral click being a small part of what happens in these tools.
When I ran the AI Verdict Study, I scripted a seven-turn buyer conversation for nine pairs of brands and ran it across four engines. Category leaders got rejected in 35% of those conversations (75 of 215). None of that shows up in anybody's analytics, since the buyer gets steered toward or away from you inside the chat. When I sorted the 32,000 citations from those conversations, the brand's own site was 4% of the classified citations on the first unbranded question, so the engines were mostly reading about brands on other people's pages (Where AI Engines Go to Decide About Your Brand).
So when I say brand visibility in AI answers is the thing to measure, this is what I mean, and the referral numbers are a separate conversation. Ahrefs found branded web mentions correlated at 0.664 with brand visibility in AI Overviews across a study that started with 75,000 brands, a stronger relationship than branded anchors or branded search volume (Ahrefs). That study is from 2025 and covers Google only, and it's correlation, so don't lean on it too hard. Still, it lines up with what Rand suggests, which is to treat AI as something closer to a billboard and measure it that way.
Remember too that most of these traffic datasets leave out AI Overviews and AI Mode, where Google answers a lot of questions before anyone clicks. If your buyers start in Google, a good chunk of your AI visibility problem is sitting there and none of it shows up as an "AI" channel in GA4.
How to make the case to your boss
If you're the one who has to defend this budget, here's how I'd do it.
Number one, fix conversion tracking before anything else. If you're one of the sites with no key events, every other number in this post is a guess for you. Set up the lead, purchase or booking event, then build an AI assistant channel group so you can see those sessions at all.
Two, compare AI against the right baseline and split it by business type. AI against organic is a different number than AI against all traffic, and a B2B product company should expect something very different from a retailer. Don't borrow the 13.8% figure for a deck unless your own data backs it up.
Three, measure presence in the answers directly. That means checking what ChatGPT, Gemini, Perplexity and AI Mode say when a buyer asks about your category, whether you get named, how you get described, and which sources the engines lean on. Mentions and citations are different things, so track both, and if you're reading a vendor's study to set targets, check how it was run first.
Then last, budget it like brand and don't rob organic search to pay for it. Most of the work overlaps anyway. Google's own guidance says AEO and GEO are "still SEO" for its AI features (Search Engine Journal), and the off-site work that gets you described well (reviews, comparisons, press, the pages other people write about you) is the same work a brand team should already be doing.
If you want a fast read on where you stand, the free AI Visibility Quick Check takes about 20 minutes and shows you which layer is weakest. The self-check below is shorter and focused on the measurement side.
Can you defend your AI visibility budget?
Six questions about measurement. Nobody sees your answers but you.
Answer all six to see your score and what to fix first.
If you want the longer version of the brand side of this, it's what my book, Explainable, is about.
I'd love to hear what your own numbers look like, especially if your AI conversion rate is nowhere near the 13.8% everyone's quoting, or if you've found a way to tie AI visibility to pipeline that holds up. Send me a note or find me on LinkedIn.
Jarred Smith is the author of Explainable: Why AI Recommends Some Brands & Ignores Others, an Amazon bestseller on AEO, GEO, and SEO. He's a marketing leader with nearly 20 years of experience across healthcare, public media, retail, and environmental services. Find him at jarredsmith.com.