Human visitors | HoneyLog Documentation

Human visitors are the requests left over when everything automated has been taken out. They are the number you would put in front of your team or your client, so it is worth knowing exactly what had to be true for a request to end up here.

#What makes a request human

A request is counted as a person only when nothing about it suggests otherwise. It does not carry the name of any crawler we know. It does not look like a script, a library, or an automated browser. And it arrives from the kind of connection people actually use, a home line, a mobile network, an office.

That last condition does more work than the other two. Traffic that looks exactly like a browser but comes from a hosting provider is not counted as a person here, it is counted as fake, because nobody reads your site from a rented server. It is a large amount of traffic on most sites, and it is the main reason your human number in Honeylog can be lower, sometimes much lower, than in tools that count anything with a browser signature.

That is a feature rather than a disappointment. A smaller honest number is worth more than a bigger one you have to apologise for later.

#Why the count is complete

Because Honeylog reads what your site actually served, a person is counted whether or not their browser ran anything extra, and whether or not they blocked the usual trackers. Visitors who are invisible to page based tools are visible here, which pushes in the opposite direction from the strictness above.

The two effects do not cancel out neatly, and the difference against another tool is rarely a single clean percentage. If you compare, compare trends rather than absolute totals.

#Where a visit gets credited

The second half of the question is what sent the person to you, which matters most when you want to know whether AI assistants are actually sending you anybody.

A visit is credited in one of two ways, and never both:

By the tag on the link. Some sources add a marker to the link they hand out, saying where it came from. When that marker is present it wins, because it is precise and it survives the messy journeys real visits take, through native apps and redirects that lose everything else.
By the site the visitor came from. If there is no marker, we fall back to the site that referred them, cleaned up into a recognisable name.

Doing it in that order, rather than adding both up, is what keeps a single visit from being counted twice.

#The honest limitation

There is a gap we accept on purpose. A visit that arrives carrying a marker from something unrelated, a newsletter for example, while having been referred by an AI assistant, is credited to neither of them. Counting it in both places would be worse, and guessing which one deserves it would be inventing data.

You will also see visits with no source at all. Someone typed your address, used a bookmark, or came from an app that passes nothing along. That is ordinary, and it is reported as unknown rather than quietly distributed among the sources that can be named.

Bot Conversions is where this attribution turns into a story you can follow: which assistants read your pages, and which of them then sent you people.