That Under-1% Traffic — Why Does It Convert 3 to 15 Times Better?

A few days ago, a friend in marketing was venting to me.

He said, "Liu Run, my organic search traffic keeps dropping, paid ads are absurdly expensive, and my boss chases me every single day for growth. I'm about to lose my mind."

I said, "Hang on. Have you ever checked whether there's a kind of traffic on your site — tiny in volume, but with conversions so high it's scary?"

He froze for a second. "What traffic?"

I said, "The kind that comes from AI. ChatGPT, that sort of thing."

He dug around in his dashboard. For a long time, he didn't say a word.

Then he sent me a screenshot. Even he couldn't believe the number.

A Counter-Intuitive Number

Let me give you a data point.

In November 2025, someone ran the numbers. Traffic from AI answer engines made up less than 1% of total site traffic.

Less than 1%.

But its conversion rate was 3 to 15 times that of traditional search.

You heard that right. 3 to 15 times.

When I first saw that number, my instant reaction was — wow, this isn't traffic. This is sifted gold.

Here are three more specific ones.

Someone looked at an e-commerce site and found that visitors referred by ChatGPT converted at 11.4%. On the same site, what was the organic search conversion rate? 5.3%.

More than double.

Someone else looked at Microsoft Copilot. Its subscription conversion rate was 15 times that of traditional search. Even direct traffic and social media were left in the dust.

A recent global survey even found that whether a person uses AI search is the single strongest predictor of whether they'll buy CRM software. Stronger than any other behavior.

What does that tell you?

It tells you that the person who clicks through has already made up their mind.

Why Is This Happening

Think about how you normally use Baidu, or Google.

You search a question. Ten blue links come back. You click the first one, glance at it, back out. Click the second one, glance again. Back out again.

You might bounce between five or six sites for a whole morning, researching — and still not decide what to buy.

That's traditional search. It shatters the act of "research" into one click after another.

But AI is different.

You ask AI a question, and it doesn't hand you ten links to pick from. It first breaks your big question into several smaller ones. Then it goes off and searches on its own, stitches the answers together, and tells you directly.

This move has a name: "question fan-out."

Fan-out — like a fan, spreading out from a single point into a wide arc. AI takes all the sub-questions you'd otherwise have to search one by one, and searches them all in one breath.

So by the time someone has wrestled with the AI chat box and then clicks into your site, they already have an answer in their head.

Their research phase already ended inside that chat box.

They're coming to you to confirm one last thing. They're coming to place an order.

Small Volume Doesn't Mean Weak

I know what you're about to say.

You're thinking, "If the volume is that small, what good is it?"

Fair enough. The absolute volume of AI traffic really is small. Compared to the organic search you've spent years building, compared to the paid ads you're pouring money into, it's still the new kid on the block.

But you have to look at it from a different angle.

If there were a channel where every single visitor "already had an answer in their head" — would you really turn your nose up at it for being small?

High volume but loose, versus low volume but precise — which is worth more?

The math speaks for itself.

How to Surface It from Your Dashboard

Having a feeling isn't enough. You have to be able to prove it with data.

And that's the tricky part. Because in most people's dashboards, there's no row for "AI-sourced traffic."

It's either mixed into referral traffic, or it simply gets counted as direct traffic. You don't even know how much of this gold you're sitting on.

What to do? Three methods, fastest first.

The first is the fastest. Open your dashboard, find the traffic-source column, and just search for ChatGPT, search for Perplexity, search for Claude, search for Gemini. One glance and you'll have a rough idea of how much you've got. You don't have to build anything — five seconds, done.

The second takes a little more time. You build a channel grouping yourself, name it "AI Search," and then use a regular expression (regex) to capture the URLs of the main AI players. The benefit: from then on, it sits as its own row, side by side with organic search and paid search — crystal clear. And you can backfill historical data.

The third is the most effortless, but it depends on luck. Dashboards have recently started natively supporting an "AI assistant" channel. If that row shows up for you — congratulations, you don't have to configure anything. Just use it.

But one thing I have to warn you about.

No matter which method you pick, it will undercount.

Because some of the visits that click through from AI don't carry a source attribution at all. A tap inside an app, a link copied into a browser — they all end up counted as direct traffic. So the number you see is a floor, not the full picture. Treat it as a signal, not a precise count.

Finding It Isn't Enough — You Have to Chase It to the Money

Finding the traffic is step one.

The hard part is this: you have to prove that this traffic turns into money.

Your dashboard can tell you that someone showed up, looked at a few pages, stayed for so long. But it usually can't tell you one thing. Did they end up buying? Did they sign the deal? How big was that deal?

For that, you have to go chase it in the CRM.

The principle isn't actually complicated.

The first time a person lands on your site, they haven't filled out anything — they're an anonymous visitor. But the system has already quietly recorded where they came from this time.

Then one day they fill out a form and become a Contact, and the system sticks that earlier anonymous-visit source onto this Contact. And from there on, the deals they sign automatically inherit that source.

This way, you can see a complete chain: clicked through from AI, became a Contact, signed a deal. Wherever a link in that chain is broken, you can see it clearly.

Some CRMs have already started flagging AI sources separately. A click from ChatGPT, a click from Claude — automatically grouped into one category, no configuration needed on your end.

The value of this is that you can finally answer the boss's hardest question: did this traffic actually make any money?

Stop Competing on Volume — Compete on Quality

Alright. Now the numbers are all there. How do you compare?

My advice: don't go toe-to-toe with AI on volume. Its volume is small — that's a fact.

What you should compare on is quality.

How do you compare on quality? With one thing, called intent scoring.

In plain terms: you give every visitor's behavior a score. Stayed long enough — add a few points. Scrolled to the bottom — add a few points. Looked at the pricing page, looked at the comparison page — add a few points. Completed some key action — add a few more points.

Each channel ends up with an average score.

Doing this has two benefits.

First, it lets you see, before anyone has even converted, which channel's visitors are more serious. AI traffic is small, and its conversion rate can swing wildly — hard to read. But the intent score is stable.

Second, it puts every channel under the same ruler. You say AI visitors have higher intent? Fine — let's score it and see. This isn't just talk. It's checkable.

Lay the scores out, and who's high and who's low is obvious at a glance.

How to Get AI to Cite You More

One last thing. Since this kind of traffic is so valuable, how do you get AI to cite you more?

I thought about it, and really, it comes down to three things.

First, you have to think through that "fan-out" on AI's behalf.

Like I said earlier, AI breaks a big question into several smaller ones. You have to anticipate all those smaller questions — and then answer all of them in a single article.

Don't split the answers across five thin posts. Combine them into one thick one.

Prioritize answering the sub-questions that carry buying intent. How much does it cost, how does it compare, can it integrate with my existing system, how do I decide. Those are the questions that actually make people pull out their wallets.

Second, what you write has to match the way buyers ask their questions.

In the old days of SEO, you'd think: what keyword will this person type into Baidu? Now you have to think: how will this person ask ChatGPT?

The asking has changed. It used to be "best CRM." Now it's "we're a ten-person company, already using a certain system, and we want to find something that works alongside it — any recommendations?"

Longer. More conversational. Closer to the decision.

You have to hit the nail on the head.

Third, the pages that get cited have to be tied to revenue.

This is the one a lot of people overlook.

An article gets cited by AI, and the traffic rolls in. But have you gone back to check: how many deals did that article actually bring in?

Some pages get little traffic, but every visitor brings a big contract. Other pages get tons of traffic, but not a single person buys.

Which one would you prioritize writing?

My answer is: rank your content by closed-deal revenue, not by traffic. That page with the smallest traffic but the biggest order — that's the direction you should keep writing in.

Finally — How to Make This Comparison Hold Up

You'll probably want to take this whole framework to convince the boss, the finance team, and the rest of the marketing department.

A few numbers aren't enough. You have to make the entire comparison withstand scrutiny.

I suggest you do three things.

First, nail down who owns each piece. Who maintains the channel definitions, who sets the scoring criteria, who manages the dashboard. Don't let it get to the end and have three people pointing fingers, each thinking it's someone else's job.

Second, write the definitions down on paper. What regex your AI-search channel uses, how many points each scored behavior is worth, which conversion goal you use when comparing — write it all down. Get three departments looking at the same version, not each picturing their own.

Third, review it on a regular cadence. Every quarter, go through it once and check whether your channel setup is still capturing new AI sources. This field moves fast — a new engine can pop up every few months.

Do these three things well, and your comparison becomes something that can actually convince people — not a chart anyone can poke holes in.

So Now — Where Do You Start

I know. Doing all of the above all at once isn't realistic.

Just do these five steps first.

First, define the intent signals. Figure out which few behaviors mean "this person is serious."

Then, separate the AI traffic into its own row in your dashboard.

Next, put it side by side with the other channels and compare.

Then, trace the source all the way to the deals in your CRM.

Finally, pull those numbers into one dashboard and show it to the boss.

One week, give or take, and you can have a preliminary result.

One more thing I want to say.

The essence of this — it isn't really a technical problem.

It's a problem of perception.

While everyone is busy agonizing over declining traffic, a small group of people has quietly started paying attention to something else. These people coming in — how badly do they actually want to buy?

Traffic will keep changing. Search engines will change. AI will change. But one thing won't change.

Whoever is closer to the person holding the wallet, wins.

Here's wishing you get a little closer to the people who want to pay.