What the website counts

Gigahost counts how its website is used, to see which pages people read, where they come from and how many sign up. The counting happens on Gigahost’s server as it answers a page. It uses no script, no cookie and no third-party service in your browser, and it keeps counters, not a record of who visited what.

What a count holds

For each page the website answers, Gigahost adds one to the counters of:

• The page, by its route. A page of yours is counted as /instances/{id}, never as the instance itself, and the part of the address after a ? is left out.
• How it was answered: found, not found, or something else such as a redirect.
• Whether you were signed in, and nothing about who you are.
• The site you came from, when the link that brought you was on another site: its host name, such as example.org, and never the page it was on.
• The kind of device, desktop, phone or tablet, which your browser’s user agent tells.

An address the website has no page for, such as a mistyped one, is counted as a page that was not found, under its path with anything that looks like an ID, a token or someone’s own text replaced by {id}. Each website server lists at most 200 of these a day and counts the rest together.

How a visitor is counted without being identified

To tell how many people visit rather than how many pages they open, the website makes a hash of your network address and your browser’s user agent together with a random salt. The salt is new for each website server and each day (UTC), lives only in that server’s memory and is dropped when the day ends or the server restarts. The hash is kept in memory for the rest of the day, only to know that you were counted already, and then it is gone too. What Gigahost stores is the resulting number.

This means a visit cannot be turned back into an address or a user agent, and a visitor cannot be followed from one day to the next. It also means a visitor counts once a day for each website server’s run, not once a day in all: Gigahost runs more than one website server, so someone who reaches two of them on the same day counts as two visitors, and each server starts over with a new salt whenever it restarts, which it does when Gigahost deploys an update, so someone who comes back after a restart counts again. The number of visitors is therefore an upper bound rather than an exact count, and it is somewhat higher on a day with updates. The figures say so. Gigahost does not keep the salt past the server’s run to avoid this, because a kept salt would put something that can be matched with an address into storage.

What is not kept

No address, user agent, account, session or ID goes into the counters, and nothing links a count to an account. The sign-ups on the statistics page are the number of accounts created each day, taken from Gigahost’s own account records, and the first logins the number of accounts that signed in to the website for the first time that day, which is counted at the moment of the login and added to that day’s number. Neither names an account.

What is left out

The counts leave out visits by bots and crawlers (told by their user agent), the website’s health check, images, styles and other files, the calls your page makes to Gigahost’s API, and the operator’s own pages and sessions.

How long counts are kept

Counts by the hour are kept for 30 days and counts by the day for two years. Gigahost deletes them after that.

Cookies on the website

The counting sets no cookie. The website’s own cookies keep you signed in and remember your theme and time zone, and Gigahost does not use them for counting.

The Privacy Policy lists each of them with its purpose and lifetime, and says what else Gigahost keeps about you and for how long. This page has the detail of the counting alone.