AI Founder Weekly

Marc Lou says bots hit TrustMRR 5x more than humans

Thirty days of logs from one directory show agents fetching pages far faster than they send anyone back.

Marc Lou posted TrustMRR's bot traffic for the last 30 days and the headline number is 1.23M verified bot requests against 278K human visitors. That is about 5x more machine traffic than human traffic, or 4.4 bot requests per visitor by his count. OpenAI on its own accounted for 606K of those requests.

Marc Lou
@marclou
X
So AI agents do 100-1000 fetches to refer a visitor on TrustMRR.
Nov 6, 2026 · View on X

The more useful half of the post is the conversion math. Lou says ChatGPT made 238K page fetches and sent back 2,591 tracked referral visitors. Claude made 68K fetches and sent back 58. Same site, same month, wildly different return on being crawled.

What the ratio actually means

Lou's own read is that agents do 100 to 1,000 fetches to refer a single visitor. Treat that as one founder's logs rather than an industry benchmark, because it is exactly that. But the direction is hard to argue with if you run anything public facing. Agent traffic shows up on your bandwidth bill and your database before it shows up in your analytics, and for most sites the gap between those two moments is large.

That reframes a question plenty of founders have been arguing about all year. The debate has mostly been whether to let AI crawlers in at all, framed as a content ownership fight. Lou's numbers suggest a plainer operational version of it. If your serving costs scale with crawls and your revenue scales with humans, you want to know which bots are actually paying you back in traffic and which are just reading.

The business read

There is a real difference between the two assistants in his data. ChatGPT returned a visitor roughly every 92 fetches on TrustMRR. Claude returned one roughly every 1,170. Whether that holds on other sites, nobody knows yet, and one directory's thirty days is not a trend.

It is worth remembering what TrustMRR is in Lou's portfolio. He said earlier that it did $427k in its first year at 90% margin, which makes it exactly the kind of product where serving costs are small enough to ignore right up until they are not. A site that sits on 90% margin can absorb 1.23M bot requests without blinking. A heavier product with real compute behind each page view cannot.

The practical move is boring. Look at your own logs, split verified bot requests by provider, and compare them against referral visitors from the same provider. You will either find a channel worth feeding or a cost worth capping. Lou just published his version of that exercise.

Steal the numbers, not the vibes

One email a day. Under three minutes. Real MRR, real pricing, real launches from founders building with AI.

Get the next one by email

Founders building with AI, every day. Real revenue, real pricing, real launches and what they would redo. No hype sludge. Every number links out.