Hacker News Reader

Hacker News

Comment by DeusExMachina | Hacker News Reader

Parent Full thread

DeusExMachina

I think (but I have no idea) that browsers are tracked by their user agent, which they always send.

View on HN

Many browsers start "freezing" the user agent.

But the user agent isn't that important when we talk about tracking.

https://github.com/w3ctag/design-reviews/issues/467#issuecom...

Indeed, however Firefox often masks its user agent as websites falsely report broken and missing features for non-chrome browsers. User agent is growingly meaningless as a metric.

One of the first things I do on a new Firefox install is add a user-agent switcher and set it up to tell sites I am using chrome for exactly this reason.

Somehow we have gone back to the dark ages of web where sites complain unless you are using one specific browser.

It was the more technical minded people who spread the word and mindshare of browser choice back then to get us out of that mess, and now the same group has gotten us back into it worse than ever.

I'll never understand how this happened.

User agent override is a builtin feature in firefox for ages.

Thanks for that info! I had no idea about this.

A quick (though not very user friendly) method to change Firefox's User-Agent string is to create a new about:config pref (string type) called "general.useragent.override" wit the new User-Agent string.

You can see Firefox's current list of User-Agent overrides in Firefox's "about:compat" page.

The website has the user agent but it cannot send it to the analytics platform because requests are being blocked.

The user agent header is passed to whatever server your browser originally talks to, that server is free to do whatever it wants with the header (including forwarding it or storing it in logs that are analyzed somewhere at some point). Blocking on the client or even client network level has no effect on that. If you receive any content at all, there is a chance the UA header will be looked at.

But as others pointed out, the header will be rather meaningless in the foreseeable future anyway.

But Google Analytics is typically (always?) done through JavaScript in the browser. And if the browser blocks any call to the Google analytics servers, Google won't know the user agent.

Theoretically what you're saying is true, but practically speaking many websites are behind CDNs, like Cloudflare, Cloudfront or Akamai, with static content being cached aggressively to reduce costs and latency and it's pretty cheap and easy to do so nowadays, so most requests aren't even going to hit a backend.

You can work around this by installing a tracking pixel on every page you have, a pixel that always hits your backend, which can then generate a log line that can be analyzed. But this requires extra development and it's much easier to just install Google Analytics.

So I'm pretty sure that most browser stats are not coming from analyzing 1st party data that's logging the user agent.

And speaking from experience in this space, the error margin for such analytics is somewhere around 10% - 20%, which is roughly the percent of people having ad-blockers installed and this number is growing — you can extrapolate of course from those that don't have ad-blockers installed, but then you have a selection bias issue, because you're not talking about the same kind of user; e.g. people that use ad-blockers are the people that are more likely to be computer literate and capable of installing their own browser.

While you in turn are of course correct as well I wasn't trying to imply that every business out there is sitting in front of their server farm and analyzes httpd logs anymore.

With the prevalence of ads being served from the customer domain on larger sites these days I'd just think that enough parties have incentive enough to implement a tracking pixel or put a few lines of measurement protocol code somewhere to make sure that request data ends up in GA. But granted, I'm not in that space, the assumption that this data does not end up in the hands of some interested party for B2C sites just seemed weird, even in modern architectures.

Thanks for the error margin btw, that's far higher than I would have expected.

Tools like GA will be blocked by Firefox in tracking protection mode. They won't register a Firefox user.

The server is free to do anything with UA, but it doesn't, because google analyticts already takes care of that.

Sure, but most people don't analyze their server logs, they rely entirely on analytics, and if not done correctly (noscript tag) won't fire even a tracking pixel. Even then, most blockers will block those requests too.

Reply on news.ycombinator.com