Elevator Talk: Is AI Traffic Really Hiding in Direct?

October 22, 2026

If AI answers rarely send a click, where does all that AI influence actually show up in your analytics? The go-to answer lately has been "it's probably hiding in direct." But is that a real explanation, or just a convenient one?

Brad Porter dropped a post from Rand Fishkin into Slack that sums up a lot of the team's attribution worries. Rand was reacting to a Semrush study of AI search in manufacturing, which found that AI referral traffic is a sliver of the total. Brad's question: if AI traffic contributes so little, is it fair to assume direct is the catch-all for everything AI influences? He's seeing direct traffic climb for nearly every client, so maybe there's something to it.

The data was limited to manufacturing and industrial brands, but the team agreed the pattern looks familiar across industries. Here's how they think about it.

"Some People Might Use, Maybe, But We Don't Know"

Samantha BrownVP of Client Services

As usual, Rand nailed it. It's not the main thing Brad called out, but Rand perfectly put words to a frustration we all share with prompt performance analysis. That goes for the prompts we custom track, and especially for synthetic databases like Brand Radar. It basically boils down to: here are the most visible brands, based on a bunch of prompts some people might use, maybe, but we don't know.

So instead of looking for one AI number, we have to watch every path someone could take to the website once they've finished their off-site research:

Direct traffic, with the caveat that it's a catch-all
Branded impressions and clicks in Google Search Console
Homepage traffic, since that's where people land when they already know your name
Branded paid campaign performance, which often improves when more people are searching for you

I just wrote this up for a client project that shows the new user behavior next to the old, traditional SEO behavior. The difference is clear once you lay them side by side.

Before AI Overviews	Today’s AI-Era Search Journey
1. Buyer searches for information Google is the starting point.	1. Buyer searches Google or uses an AI tool
2. Your brand appears in search or an ad The buyer sees a result and clicks through.	2. Your brand appears during that research The buyer sees your brand in an AI answer.
3. Buyer clicks through to the website The website becomes the place where deeper research happens.	3. The buyer continues researching without clicking They ask follow-up questions about their specific needs, comparing fit, pricing, strengths, weaknesses, use cases, and more.
4. Buyer evaluates and narrows their options Much of this activity happens on the website and is measurable.	4. The buyer narrows their options They may form a shortlist or preference without ever visiting the brands’ websites.
5. Buyer converts Analytics can connect the visit and conversion back to the acquisition channel.	5. The buyer finally reaches the website The visit may show up as direct or homepage traffic, an LLM referral, organic branded search, or a paid branded click.
The journey and attribution largely happen in the same place.	6. The influence shows up somewhere else You may see more branded search impressions, more homepage and direct traffic, or stronger performance from branded paid campaigns, even as non-branded organic or paid traffic declines.

AI Visibility Isn't the Same as AI Traffic

Paxton GrayCEO

I agree that "it might be hiding in direct" is a weak answer. Still, direct is a valid channel, and AI search is affecting it. It's just made murkier by the fact that it's the catch-all bucket.

I'm with Sam that we need other measures of success: overall conversion rates, branded impression volume over time, and AI traffic plus AI traffic conversion rates. As fuzzy as it is, I'd also show clients how they compare to competitors on an agreed-upon prompt set. That set can change over time, but it's still motivating to watch.

Where I think Rand is off is this line: "IF it were the case that AI visibility and direct traffic were correlated, we'd see their trend lines looking similar." The statement itself is true. But AI visibility doesn't equal traffic from AI. It's reasonable to imagine referral traffic from AI platforms going down while direct traffic goes up, because more people are seeing your brand in AI answers and coming to you on their own.

I might be missing it, but the study never compares branded mentions against direct traffic. That's the overlap you'd need to show that more AI visibility drives more direct traffic and branded search, or that it doesn't.

Attribution Was Always More Art Than Science

Blake NielsonHead of Accounts

Marketing attribution has always been more art than science. I think as marketers we just got comfortable with how easy attribution was the past few years. But conversions and revenue have always been the gold standard of truth.

I love the suite of metrics Sam called out. We'll look at all of them together and make strategy decisions from there. Ultimately, though, those metrics should get us closer to revenue.

Does Perfect Attribution Really Matter?

Alondra MeloSe

Reporting on AI search has been the biggest learning curve in all of this. Sam's metrics are the closest thing we have to validating performance. The nice thing is that if you're doing the right AI SEO work consistently, you're putting yourself in a really good position.

The tricky part is that you probably won't always know exactly which source influenced every lead. But if visibility, traffic, and leads are all moving in the right direction, does perfect attribution really matter that much?

At some point, the focus has to be on doing the marketing work, showing up in the places your audience is looking, and paying attention to the overall results. That's something I'm still learning to communicate with clients, too.

The Takeaway

The team agreed on the short answer: "it's hiding in direct" isn't good enough on its own. Direct is a catch-all, so a rising line there can't be credited to AI by default. But AI search is plausibly part of why direct keeps climbing for so many clients. And as Pax pointed out, low AI referral traffic doesn't prove low AI influence. Someone can read about your brand in ChatGPT, never click, and show up a week later by typing your URL.

So stop looking for one AI number and watch the full set of paths a buyer takes after off-site research. That means direct traffic, branded impressions and clicks in Search Console, homepage visits, and branded paid performance. Layer in AI referral traffic and its conversion rate, plus share of voice against competitors on a prompt set you've agreed on with the client. None of these proves AI influence by itself. When they trend up together, the story gets hard to ignore.

And as Blake and Alondra put it, the scoreboard hasn't changed. Conversions and revenue were always the truth. Attribution will stay fuzzy. Do the work, show up where your audience is looking, and judge success by whether visibility, traffic, and leads move in the right direction together.

If your AI search reporting feels like guesswork, let's talk about building a measurement framework that connects visibility to revenue.