Should we add tracking to the guide?

[GBKS]

We have always been strictly against any sort of tracking. No tracking is a 1000x better starting point than just throwing in Google Analytics and another thousand scripts by default.

Now I'd like to revisit this decision. The primary reason is that understanding what pages get visited, and how often, can help us learn which content is the most interesting and useful. And this can help us decide what to prioritize next.

While every page has a very visible "Feedback & questions" prompt, fact is that absolutely no one uses this. From conversations, we also hear that many people refer to the guide. It would just be helpful to have a rough idea of what is going on. Do we have 10 visitors per month or 1000? We have no idea.

The Bitcoin Dev Project is using Umami, which is GDPR-compliant (no cookies) and open-source. You can see their dashboard here. They self-host it, ensuring that the data is not shared with anyone else. That could be an option (so we would also make it publicly visible).

If we are not 100% sold, it could also be an option to install Umami (or another tool) for 3 months, and then remove it again. So we would get some understanding, but not commit to it.

Generally, I never fully trust analytics, as I know that a huge amount of people use ad-blockers (Safari even has it built-in by default). And I think bitcoiners are especially sensitive to this. So even if we add tracking, we will not get super reliable numbers.

What do you think?

[sbddesign]

cACK. I remember raising a discussion like this several years ago, so I'm glad to see the topic being revisited.

Yes, it's good to not just blindly give over all the data to Google, but we also don't want to miss out on the pretty clear benefits of knowing how users interact with a product. There's a lot of nuanced gray area between invasive tracking cookies on one end of the spectrum and not tracking at all on the other end of the spectrum. For example, just keeping track of which page are most frequently visited at the server level would be hugely beneficial.

Anyways, maybe I can provide some more detailed input later, but just wanted to drop a concept ACK.

[swedishfrenchpress]

ACK.

Based on a brief review, I believe we can use Umami responsibly and in a way that respects privacy. Starting with a three-month trial, followed by reassessment, seems like a reasonable compromise.

[paulosacramento]

ACK. This makes perfect sense. Having data on which pages are visited more and so on helps to make important decisions. It can also help provide numbers that accurately communicate the relevance of the project to supporters.

I think Plausible should be considered as a candidate. My understanding is that it is as privacy-preserving as Umami.

The main advantage of Plausible is that there are clear instructions on how to proxy the scripts, so that adblockers do not affect tracking as much. More on this: https://plausible.io/docs/proxy/introduction#are-you-concerned-about-missing-data

[yashrajd]

concept ACK

I agree we'd like to see pages visits and clicks. But my default position is against any further/deeper tracking. Also:

• we should only do this if we can self-host and none of the data collected is available to even the analytics provider
• we should not be collecting/keeping any PII data or anything else (device model, country etc.) that can be used to pinpoint users in any way

I took a look at both umami and plausible, I like Plausible more but I don't know much about choosing or using trackers.

I also like trying something for a couple months first...

edit: I also like the idea of making (only the highest-level) stats like view/visitor count, visit time etc. publicly available.

[GBKS]

The main advantage of Plausible is that there are clear instructions on how to proxy the scripts, so that adblockers do not affect tracking as much.

I would prefer if we respect people's choices to use ad/tracking blockers and not try to circumvent them.

I also like the idea of making (only the highest-level) stats like view/visitor count, visit time etc. publicly available.

All data should be publicly available, IMHO. Not sure about Plausible, but Umami doesn't even track that much detail in the first place.

we should only do this if we can self-host

I would definitely self-host, probably with a dedicated DigitalOcean droplet on a 5$/month plan. I should be able to set that up pretty quickly.

[GBKS]

Alright everyone, so I dug into Umami a bit further and managed to set up an instance of it on DigitalOcean (under my account). It is mapped to the info.bitcoin.design subdomain, and I created a public dashboard right here. Now this is just the Umami setup and it does NOT collect any data yet. To start that, we need to create a pull request on the guide repo that adds the tracking script to the site itself. If/once we do that, then the dashboard will start populating.

How does that look/sound to you? Should I move forward and create the aforementioned pull request?

[GBKS]

I created PR #1129 for adding the tracking script to the site. So, are we doing this?

[swedishfrenchpress]

PR #1129 LGTM. I think making the dashboard public is good choice too.

[GBKS]

The tracking script is live now, you can see the stats on the dashboard.

I also have a small follow-up fix for the dashboard link added to the join page. See #1130.

[GBKS]

It's been 4 months or so now. Generally, I find the data helpful. Are there any strong thoughts on removing it? If not, I'd like to keep it running, and potentially also add it to the sites for the Bitcoin UI Kit and Bitcoin Icons so we get some insight there as well.

[yashrajd]

It's been 4 months or so now. Generally, I find the data helpful. Are there any strong thoughts on removing it? If not, I'd like to keep it running, and potentially also add it to the sites for the Bitcoin UI Kit and Bitcoin Icons so we get some insight there as well.

sounds good