Knowing what works, without tracking anyone

What it is

Umami is open-source web analytics. A small script on each page reports which page was viewed, where the visitor came from, roughly what kind of device and browser they used, and which country they are in. Umami turns that into a plain dashboard: visitors, page views, top pages, referrers, and how those change over time.

What it does not do matters as much. It sets no cookies. It does not fingerprint browsers to recognise people across visits or across sites. It does not build profiles, and it does not share anything with an advertising network.

Why I run it instead of the SaaS alternative

Google Analytics is free, powerful and deeply integrated with Google's advertising tools. If you run paid campaigns and need conversion tracking tied to ad spend, it does things Umami does not attempt, and that can be the right choice.

I do not run ads, and I do not want visitors to my sites feeding anyone's advertising profile. I want to know which pages are read, where people arrive from and whether a change helped. That is a much smaller question, and it does not need anyone tracked. Because Umami sets no cookies, there is no cookie banner to put in front of every visitor just to count them.

Hosted privacy-first services such as Plausible and Fathom take the same approach and are a sensible choice if you do not want to run a server. I already run one, so hosting Umami costs almost nothing extra, and the data stays on my disk.

How it is set up here

1. One instance for the estate. Umami and its database run in their own containers on the estate server, with each site registered as a separate website inside it.
2. A small script on each page. Sites load the tracking script from my own analytics address, not a third-party domain, so it is not caught up with advertising trackers.
3. No cookies, no fingerprinting. Visits are counted without identifying anyone, which is the point of choosing it.
4. Server logs for the rest. Some visitors block scripts, and some traffic is not a person at all. The nginx edge logs every request, so crawlers, API calls and script-blocking visitors are still visible in aggregate when I need them.
5. Backed up like everything else. The analytics database is part of the nightly dumps that leave the server.

What it replaces

• Google Analytics for sites that do not run advertising.
• Cookie consent banners that exist only to allow visitor counting.
• Third-party tracking scripts loaded from someone else's domain on every page.

What it costs to run is a small container, a database and routine updates. It is one of the lightest services on the estate.

What I would do for a client

First, find out what is actually loaded on your pages. Many sites carry analytics and marketing scripts that nobody remembers adding, each sending visitor data somewhere. The audit lists them, says what each one collects, and asks whether anyone still uses it.

Then the recommendation. If you depend on advertising and need conversion tracking, keep the tools built for that and configure them properly. If you just need to know what is being read and what is working, I would set up Umami, hosted by you or as a hosted privacy-first service, remove the scripts you do not need, and give you a dashboard that answers your real questions without a consent banner in the way.

Questions people ask