Three numbers first.

This client has been in its local market for 15 years. Its monthly digital marketing budget runs into six figures, and has done for years. Over that time it has been through several agencies.

Trackable customers from the web: zero.

Not few. Zero.

Every agency it hired delivered beautiful traffic

Each one had a report ready: impressions climbing, clicks climbing, keyword rankings moving up. The data was not fabricated. It was real.

The trouble is that none of it answered the question the owner was actually asking — how many customers came from the web this month?

When nobody could answer, the owner did the only thing left to do: switch agencies. He switched, and switched again. Each new agency arrived with a fresh dashboard, the numbers looked just as good, and the customers stayed exactly as absent.

Somewhere around the third switch, this should have become visible: the problem was never the agencies’ execution. It was the thing they were selling. Their product is traffic. Sign the traffic contract with someone else and what you get is still traffic. Traffic does not turn itself into customers.

This is not an attack on the industry. It is a structural point: if you have already changed agencies three times and nothing moved, the fourth one will not move it either. The thing to change is not the vendor. It is the acquisition channel itself.

What “trackable customer” actually means

Every number in this article hangs on that phrase, so let us pin it down.

A trackable customer is a real customer you can attribute to a specific source channel. You know who the person is, which road they walked in on, and what they were asking before they arrived.

That is a different object from traffic, impressions or clicks. Traffic tells you how many devices loaded your page. Trackable customers tell you how many people actually came to you. The gap between those two is exactly where this client’s six figures a month disappeared.

Most marketing reports cannot produce a trackable customer count, because doing so means connecting behaviour on the website to a human being who walks through the door — and for most agencies the job ends the moment the visitor leaves the site. They do not have the number.

This client does. Because he actually does the work: every single customer who walks in gets asked “how did you hear about us?”, and the answer gets written down. Every number in the rest of this article was counted that way, one customer at a time.

So here is one question you can put to your agency today:

“How many customers came from the web last month?”

If the answer starts drifting toward impressions, rankings, reach or engagement rate, you already have your answer. They are not hiding the number. It does not exist.

Month one: inbound customers up 109%

In the first month after we took over the managed GEO work, the client’s inbound customer count grew by 109% — it more than doubled.

One thing needs to be said plainly here, because it is easy to misread: that 109% is the growth in total inbound customers. It is not the growth of web-sourced customers. Web-sourced customers started at zero, and zero cannot grow by 109%.

What actually happened is this: the extra customers — roughly a whole additional set — came almost entirely from the web channel that had just been built from nothing. And because that new channel brought in about as many people as all his existing channels combined, it is now his single largest acquisition channel. Steadily so.

For a 15-year-old brand, doubling the customer count is hard — not because doing more is hard, but because its existing channels had long since hit their ceiling. What opened up was a road six figures a month had never managed to buy.

And this was not a one-month firework. We have been running his GEO for several months now, and the web has been his largest and steadiest channel throughout.

More than half his customers come from AI — a number he counted himself

Start with where that number comes from, because it decides whether you should believe it.

We did not estimate it from a dashboard. He counted it, one customer at a time. Every person who walks in gets asked “how did you hear about us?”, and the answer goes into the record. Across everything he has logged so far: more than half say they were recommended by an AI engine.

That is his operating data, not our marketing claim. And it beats any analytics tool on the market — GA cannot see a visitor who arrives with no referrer. A human mouth can.

The mechanism is straightforward. His potential customers go to ChatGPT, Perplexity and other mainstream AI engines, describe their situation in their own words, and ask who they should turn to. His name comes back in the answer. They follow it to his door.

The contrast with traditional search is the whole point. SEO puts you near the top of a results page and waits for the user to evaluate and click. An AI engine says your name out loud — and only after the user has finished describing what they need.

So the engine has already filtered once before it hands the person over. People who walk into his shop know what they want.

That is why the web went from zero to his largest channel so fast. It does not send traffic. It sends people who have already made up their minds.

You can test this yourself today

Do not take any number above on faith. Spend three minutes finding out whether you are standing in the same hole:

  1. Open any mainstream AI engine — ChatGPT, Perplexity, Gemini, whichever.
  2. Do not search your brand name. Ask what your customer would ask. Not “is company X any good”, but “I have this problem, who should I go to” or “recommend a service for this”.
  3. See whether you appear in the answer.

If you do not, then the budget you spend every month is not buying you that position — because that position is not among the things those agencies sell.

Those three steps are the quick version. To do it properly — an incognito window to avoid personalisation bleed, several engines, results written down — we have a full walkthrough: Screenshots lie, backtests don’t: verify your brand’s AI visibility by hand, in one incognito window. Run it once and you will have a record you can take to someone and argue with.

Then take the question from the previous section to your agency: how many customers came from the web last month?

Can you do this yourself?

Honestly: some of it, yes. Some of it, no.

Here is what you can do on your own. Run a free health check on geoweb.tw and find out where your GEO health score sits. Ask the AI engines a few times yourself, using the method above, and see whether you get named. Put the basic structured data your site should already have in place. All of this is worth doing and you can start today.

These three, on the other hand, are hard to carry alone:

That is why this client’s result came out of a managed engagement: each of those three needs someone watching it, continuously.

What to do next

If the story sounds like your own company — budget in place, agencies changed, reports gorgeous, no customers arriving from the web — then there are two concrete things to do:

  1. Today: run your site through the free health check on geoweb.tw and see where you land across the 12 dimensions. It costs nothing, and you walk away with a report you can take to any agency and use as a cross-examination.
  2. Next: if the score is low, or you already know what is wrong but have nobody to fix it, write to [email protected] or book a demo. We will run a backtest first and show you how the mainstream AI engines currently talk about you. Once you can see that, we can talk about whether to work together.

Where these numbers come from: all of them were compiled by the client himself — he asks every customer who walks in how they found him, and writes it down. These are not our estimates; they are his operating data. The 109% is the change in inbound customers during the first month of managed GEO. “More than half from AI search” is the standing position to date, not a single-month spike. The client’s identity is de-identified under NDA.