AI Assistant Traffic in Your Analytics Is a Floor, Not the Result

AI assistant traffic is the first number most DMO teams look at when they start taking AI seriously, and it is usually small. In one DMO client’s 28-day baseline it was about 1,600 sessions, around 0.2% of total traffic. A board looking at that figure could fairly ask why it deserves attention.

The answer is that analytics measures the wrong moment. This post explains why AI assistant traffic is a floor, how to read it, and what to measure alongside it.

What AI assistant traffic actually measures

AI assistant traffic is the set of sessions that arrive on your website from AI tools such as ChatGPT, Perplexity, Gemini or Copilot. In analytics they usually appear as referrals from those domains, if they are tracked at all.

That number captures one specific behavior: a traveler read an AI answer, saw a link to your site and clicked it. It is real, and it is worth tracking. But it is the last step of a journey that mostly happens somewhere your analytics cannot see.

Why AI assistant traffic is only the floor

Most of the influence happens before the click

An assistant answers the question in the chat. The traveler reads which destinations were named, what to do there and when to go. Many decide right there, with no click at all. This is often called zero-click behavior.

Later visits are credited elsewhere

Some travelers come back later, from another device or a fresh search for your destination’s name. Analytics credits that visit to direct traffic or organic search, not to the AI answer that started it.

Some assistants do not pass a referrer

Not every AI tool sends a clear referral source. Some visits arrive as direct traffic, which hides their origin entirely.

Put together, AI assistant traffic is a minimum. It proves AI systems are sending people to your site, and the trend line is useful. It does not show how often your destination was recommended, or left out.

How to set up an honest baseline

Before reporting AI assistant traffic, set up a clean baseline:

1. Create a segment or channel for known AI referrers in your analytics tool. Google’s Analytics help center explains custom channel groups.
2. Pick a fixed window. A 28-day window smooths out weekly swings and is easy to repeat.
3. Record sessions and share of traffic. Both matter: one shows volume, the other shows scale.
4. Repeat on the same schedule. The trend matters more than any single reading.

4 measures that show the real result

Because AI assistant traffic only counts clicks, pair it with direct measures of what happens inside the answers:

• Presence: across a fixed set of real travel questions, how often is the destination named?
• Accuracy: when it is named, is the description right and current?
• Attribution: does the answer cite the official website, or only third-party sources?
• Readability: how well can AI systems read the official site? A Machine Readability Snapshot score out of 100 gives a simple, repeatable measure.

These are measured directly, by asking the same questions on a regular cycle and reading the site the way a machine does. Our list of 5 questions to test AI visibility is a good starting set.

How to report AI assistant traffic to a board

Keep both kinds of number in the same report, and label them clearly:

Measure	What it shows	How to describe it
AI assistant traffic	Clicks from AI tools	A floor and a trend
Presence	How often you are named	The core outcome
Accuracy	Whether the description is right	Brand protection
Attribution	Whether your site is cited	Authority
Readability score	How well machines read your site	The lever you control

A board can follow that story: the small number is real, and the larger picture is measured another way. We go deeper in how to report AI visibility to a DMO board.

Reading the trend over time

One reading of AI assistant traffic tells you little. A series of readings tells you a lot. Look for three patterns as the months go by.

First, steady growth in sessions while total traffic stays flat suggests AI tools are sending more travelers your way. Second, growth that follows a fix, such as adding destination identity or event markup, suggests your site is becoming easier to cite. Third, a sudden drop can point to a technical problem, such as a firewall rule that blocks AI crawlers, which is worth checking straight away.

Compare those patterns with your direct measures. If presence and attribution rise while AI assistant traffic stays modest, that is not a failure. It means more travelers are getting accurate answers that name you, even when they do not click.

What not to do with AI assistant traffic

Two mistakes are common. The first is dismissing AI because the traffic is small. That confuses the visible part of the journey with the whole of it. The second is overstating it, projecting huge growth from a small base. Both damage trust.

The honest position is simple: AI assistant traffic is growing and worth watching, but the real outcome, being the answer, is measured directly. Saying that plainly builds more credibility than any forecast.

Frequently asked questions