The Rise of “Dark Traffic” — And the Data Blind Spot It’s Creating

Photo By: 1981 Digital

For years, marketers have treated “direct traffic” as a sign of strength. It meant brand awareness. Intent. Demand already in motion. But what if a growing share of that traffic isn’t direct at all?

What if it’s the result of decisions being made somewhere your analytics can’t see?

That’s the reality many B2B teams are beginning to confront as artificial intelligence reshapes how buyers research, compare, and choose vendors. The shift isn’t just changing behavior — it’s quietly breaking the systems companies rely on to measure it.

The Part of the Journey That Happens Before the Click

A typical buyer journey today doesn’t start on Google. It starts inside tools like ChatGPT.

A potential customer might spend 20 or 30 minutes asking questions, comparing vendors, and refining their understanding of the market. They leave with a shortlist, a mental model of the category, and often a clear frontrunner.

Then they close the tab.

Days later, they type a company’s URL directly into their browser.

From the outside, it looks like a clean, high-intent direct visit.

But the real decision was made somewhere else.

That’s the gap Shane H. Tepper has been tracking. As cofounder of Resonate Labs, he works with companies trying to understand why their analytics no longer reflect how buyers actually behave.

“The LLM session that built their conviction? Your attribution model has no idea it happened,” he says.

When the Data Isn’t Wrong — Just Incomplete

This is where “dark traffic” enters the conversation.

Not as a technical glitch, but as a structural shift.

AI-assisted research doesn’t generate clean attribution signals. There are no referral paths, no campaign tags, no visible journey from question to decision. Instead, that influence gets absorbed into existing categories — direct, branded, sometimes even organic.

“The AI influence gets laundered into familiar-looking channels,” Tepper explains.

That’s what makes it dangerous. Because companies aren’t just missing the impact of AI — they’re misreading it.

The Illusion of Performance

On paper, everything can still look fine.

Direct traffic holds steady. Branded search grows. Conversion rates remain healthy.

But those signals no longer mean what they used to.

If AI-driven decisions are being recorded as direct visits, then marketing teams may be attributing success to the wrong channels — and investing accordingly.

At the same time, the AI traffic they can measure appears minimal, often too small to justify attention.

So it gets ignored.

Meanwhile, its real influence continues to scale — just hidden behind metrics that feel familiar.

Showing Up Isn’t the Same as Being Chosen

The gap between visibility and influence is where this becomes more than a measurement issue.

In one case Tepper points to, a company appeared in roughly half of the AI-generated responses tied to its category — a seemingly strong position.

But when it came to actual recommendations, where the AI explicitly named vendors, its share dropped to just 14.7%.

That’s a 35-point difference between being present and being selected.

And none of it showed up in traditional analytics.

“Traditional attribution captures what happens after the click,” Tepper says. “It has no visibility into what shaped the buyer’s conviction before they ever came to your site.”

The Decision Is Already Taking Shape

By the time buyers reach a website or speak to sales, much of the decision has already been made.

They arrive informed. Framed. Pre-conditioned. They reference specific features, raise targeted concerns, and compare options with unusual clarity. But they rarely explain where that perspective came from.

Increasingly, it comes from AI.

Tepper describes it as an “invisible committee member” — a silent force shaping how decisions are formed long before companies have a chance to influence them directly.

It doesn’t appear in dashboards. It doesn’t show up in CRM fields. But it’s there.

Learning to Measure What You Can’t Fully See

This doesn’t mean performance has become impossible to measure. But it does mean the rules have changed.

Instead of relying on clean attribution, teams are starting to look for patterns: shifts in branded search, changes in direct traffic behavior, and more informed prospects entering the funnel.

None of these signals tell the full story on their own. But together, they start to outline one.

“They’re fingerprints, not footprints,” Tepper says.

And in a world where the most important part of the buyer journey happens outside of measurable channels, those fingerprints may be the only evidence left behind. The uncomfortable reality is this: the system hasn’t just become more complex — it’s become partially invisible.

The data is still there. But it no longer tells the whole truth.

And the companies that fail to recognize that may find themselves optimizing for a version of reality that no longer exists.