AI Crawlers Came 608 Times, Referrals Are Still 0: The New Ledger's First Reading

The new ledger opens

We fitted our own site with a new measuring instrument. It came online at 2026-08-10 07:52 UTC. The ugly part first: this ledger is very young, and the sample covers only the short stretch since switch-on. Every reading below describes "now" — none of it represents a trend.

But precisely because this is the first page of a new ledger, it is well suited to the question readers ask most: does AI search actually bring people in? Rather than wait until the numbers look good before opening the books, we would rather record the ugly readings honestly from day one.

The readings since switch-on

The new ledger sorts visitors into two kinds: those who haul goods away, and those who walk in the door.

• Hauling: 608 AI crawler visits, from 7 different crawlers.
• Walking in: humans arriving via AI referral — 0.

608 versus 0. If this is your first time reading such a ledger, you would probably suspect the instrument is broken. We want to say the opposite: this 0 is the normal reading for an ordinary day.

Why 0 is the norm

Our earlier piece went through five months of the old books: AI companies hauled our goods away roughly 30 million times, while humans who actually walked in on an AI's introduction numbered just 476 — and that is a visible lower bound, counting only arrivals whose origin can be seen (source: https://km.idaeo.ai/ai/ai-crawler-shop-ledger).

The key phrase is "five months". 476 was not the score of some breakout day; it accumulated slowly over five months. Zoom in on any ordinary day and the referral column will most often read 0. Referral scarcity is the normal state of this business, not a malfunction of your site. Put the other way: if you grade yourself daily on referral numbers, you are measuring a second hand with a calendar — the scale is wrong, and even accurate readings will only mislead your mood.

An analogy: you run a shop at the market. The wholesaler's truck comes every day to load goods, and the counter at your door clicks away; but retail customers introduced by that wholesaler may not show up all day. Truck runs are buzz; retail footsteps are referrals. A shopkeeper would not block the trucks just because retail customers have not arrived — but neither should a shopkeeper report truck runs to themselves as sales. The buzz ledger and the referral ledger must be kept separately; one must not impersonate the other.

Which metric to watch, and when

The key to managing expectations is recognizing that each metric lives on a different timescale.

Our earlier piece measured each crawler's time to first crawl: after new content goes live, ClaudeBot comes to haul it at a median of 4.2 hours; OAI-SearchBot takes a median of 332.8 hours (13.9 days) to show up for the first time, with indexing and referrals later still (source: https://km.idaeo.ai/ai/ai-crawler-marketing).

Crawls run on hours, indexing on days, referrals on months. So the rhythm for reading the books should be:

• Daily: crawler visits — they tell you whether your goods keep getting seen.
• Weekly: signs of indexing and citation.
• Monthly: referrals. Staring at the referral number every day will only break your heart every day.

The fine print (limitations)

Before any conclusions, a few lines of fine print must be read aloud.

• Scope: we can only count referrals whose origin is visible (referer present). When an AI finishes the answer inside its app, or a user reads an AI answer and then searches their own way in, this ledger records nothing. So the 0 is a visible 0, not zero impact.
• Young instrument: the new instrument only came online at 2026-08-10 07:52 UTC, so the sample is extremely short. Today's readings describe today; they must not be extrapolated.
• Do not mix scopes: 476 and roughly 30 million belong to the old books, accumulated over five months; 608 is a new-instrument reading over a very short period. Different periods, different instruments — they must not be added together, and must not be divided into each other.

What to do next

An action list for ourselves, and for readers:

• Check crawler visits daily, just to confirm goods keep moving.
• Allow the referral column to stay at 0 for a long time and let it appear on its own; write the 0 into your expectations as the background value, so that every non-0 reading later becomes a signal rather than a consolation.
• We will report this new ledger's readings publicly on a fixed rhythm, to be announced in this column once it is set. Next time we open the books, you will see when the first entry after 0 arrived.