AI sends you fewer visitors. Here is why that can be good news

If you compare raw numbers, an AI assistant sends a business far fewer visitors than Google does. Owners who watch that number drop assume something has gone wrong. It is worth asking a different question first: fewer visitors doing what, and are they worth more or less than the ones you had before?

Fewer visitors is the whole point

A Google search hands you a page of ten options and lets the visitor do the comparing. An AI assistant does the comparing first and hands the visitor two or three names, already filtered by the questions it asked on their behalf: who is nearby, who has the right reviews, who actually does the job being described. By the time that visitor reaches your site, most of the decision is already made. They are not browsing. They are checking the one detail the assistant could not answer for them, usually your price, your availability, or a way to contact you.

That is a smaller number of visitors doing a more advanced job than most Google traffic ever does. Judging it by volume alone measures the wrong thing.

A simple way to work out what it is worth

You do not need special tools to estimate this for your own business. Three numbers do it: how many enquiries you get from a channel in a month, what share of those actually become paying jobs, and what an average job is worth. Multiply an enquiry count by a close rate and by a job value, and you get a rough figure for what that channel is worth to you per month.

Run the same formula against your AI-referred enquiries and your Google-referred enquiries separately, and compare not the enquiry count but the number that falls out the other end. A channel that sends you five enquiries a month at a 60% close rate can be worth more than one sending you thirty at 10%, and the volume figure alone would tell you the opposite.

Why the higher close rate is not a coincidence

It follows from what an assistant is actually doing before it names you. It has already checked that your service area covers the customer, that your reviews hold up, and that you plausibly do the job being asked about. A visitor who arrives after that filtering has fewer reasons left to walk away. A visitor who arrives from a general search has not been filtered at all, and a share of them will click through and discover you are not what they were after.

How to actually see where an enquiry came from

None of this is useful if you cannot tell your channels apart. Three ways, roughly in order of effort. Ask new customers how they found you and log the answer, even informally. Watch your analytics for referral traffic from chat.openai.com, perplexity.ai and similar, though this only catches visitors who click through rather than acting on the answer alone. Track your branded search: a rise in people searching your business name directly often follows being named in an AI answer, even when no click ever gets attributed to it.

A channel that sends nobody your way is worth zero, whatever its potential close rate looks like on paper. Being named at all is the number that has to move first, and the value only shows up once it does.

What to actually measure going forward

Sessions and click counts were built for a web made of ranked lists, where more visitors reliably meant more chances at a sale. That assumption gets weaker every time an assistant sends you three qualified visitors instead of thirty browsing ones. Track enquiries and jobs by channel instead of raw traffic, and a falling visitor count from search stops looking like a crisis and starts looking like what it usually is: the same or a better number of customers, arriving by a different door.

Before any of this matters, an assistant has to name you at all. If you are not sure how often that is happening, our guide on how Australian businesses get found in AI search covers what decides that, and an AI visibility audit tells you exactly where you stand today, before you spend anything working out what a mention would be worth.

Written byDima SmetaninCo-founder, SEENMETRIC