Loghound: an Open-Source Bot Detector and Traffic Analyser That Runs on Your Own Server
Roughly half of the sessions a conventional analytics tool reported on this site carried a hard bot signature — identical header fingerprints across unrelated consumer ISPs, one page each, never two, every one of them reporting exactly ten seconds on site. Loghound was written because that capture did not survive being looked at properly. It is free, open source and MIT licensed, it runs on your server, and it reads your logs read-only.
01 Three independent planes, cross-checked
A User-Agent blocklist catches what announces itself. It does not catch dozens of addresses on unrelated consumer ISPs sharing one byte-identical header fingerprint, each fetching a single page — from inside any one plane of observation, that is indistinguishable from a crowd of real visitors. Loghound watches three planes and compares them against each other.
02 Time on site, measured honestly
Conventional analytics pings while the page is merely loaded, so a tab left open in a background window reports as an hour of engagement. Loghound keeps separate clocks and never conflates them.
- Wall time — how long the session lasted, start to finish.
- Visible time — how long the page was actually on screen.
- Engaged time — how long somebody was doing something in it.
Kept apart, those three numbers say something. Averaged into one, they say whatever you want to hear.
03 Your logs stay yours
Loghound installs on your own server and reads your access log read-only. What it learns lives in two Opensolr indexes it provisions in your account during setup — yours, queryable with the standard Solr API like any other index you own. Everything except the address is hashed. There is no telemetry, and no vendor copy of your traffic.
Your visitors' behaviour is measured on somebody else's infrastructure, and the raw record of it is theirs to keep.
The reader runs on your server, the data lands in your own Opensolr indexes, and you can read the source that decided every verdict.
04 Installing it
Clone the repository and run the installer, or finish the setup in your browser. There is no Composer, no npm and no build step. An Opensolr account is required — Loghound needs to create and shape its own indexes, which is not something it can do against a Solr server it does not manage — and the free plan is enough to start.
If you would rather look before you leap: the source is on GitHub, the verdict rules are plain and readable, and the documentation says exactly what each signal means and what it does not.