I’m wanting to suggest Matomo for plugins.jenkins.io, and potentially the same for jenkins.io. I’ll admit I’m mostly curious what things are being clicked on (Like for example, does anyone care about new plugins?). We use matomo internally at current and past jobs because it was able to be self hosted, so for internal apps, no data got leaked out. As such I kinda fell in love with it. Its simple and straight forward, at least for non marketing purposes. I’ve never used the advanced GA functionality so I can’t speak how it compares.
The big selling point of Matomo is its privacy first attitude. Cookie trackers are single site, so not tracking you across the internet. The default visitor hash has a date so people are not tracked over multiple days. And there’s super easy support for adding in opt-out support/ui.
Which brings me to point number two. As Canada isn’t one of the countries we need the cookie banners for, I havn’t been following it, but based on Don’t irritate visitors with cookie consent banners - Analytics Platform - Matomo
I think we should have a cookie banner for jenkins.io because of GA. And if we add a cookie banner, we might as well not really have it because most people won’t opt in.
So, things I’m looking for from people:
Reasons why matomo is a good idea
Reasons why we shouldn’t run matomo (Past pitfalls you’ve encountered)
Alternatives you have heard of that I should invesigate
I like the privacy sensitivity that you described for Matomo. No objections from me for using it. I’m not experienced with site metrics or their use, but trust your judgment that Matomo is a good choice.
I was thinking about a similar topic a week ago. I was wondering about “if we do not really use Google Analytics as a data source for the community, why do we even keep it?”. I was rather thinking about disabling Google Analytics for good on our resources.
Switching to Matomo makes sense though I assume we would like to ask for sponsorship and SaaS then. I doubt we want to run our own analytics service and to store data. I like Matomo, because it is an open source solution with a diverse list of contributors and end user companies. I have no practical experience with it though.
Whatever we do, I think we should target having a single web analytics solution. I see 3 options we could vote for:
Evaluate and adopt Matomo or any other FOSS web analytics
Keep using Google Analytics
Ditch web analytics all together
The only advantage of Google Analytics is that it helps users to find right resources on jenkins.io thanks to the collected analytics being used for Google Search. It was essential when we had no embedded site search, and it is somewhat important even these days.
I’m not sure what that data will provide, especially since I have access to the search results from google into the site, and algolia on what people are searching on the site. That might be useful for jenkins.io, since knowing how they get there means the content might need to be changed, but not useful for plugins.jenkins.io, as the content isn’t controlled by anyone other than plugin authors.
What I was wanting to know is what features are being used. Is the issues tab ever being clicked. Do people ever click on the new/trending/updated things in the footer. What are the common links out. Etc. Behavior things, which is exactly what the analytics provides.
Google can potentially adjust its search outputs based on the collected data and user cohorts (e.g. “People in the US prefer using Declarative Pipeline” while “People in Russia prefer hacking everything with Scripted Pipeline and Pipeline Libraries”). Not sure they do something reasonable for jenkins.io, I am not a SEO expert or a Google employee. I can only say for sure that I get completely different results when searching for open hardware things (VHDL for my Google account, Verilog for anonymous search).
What I was wanting to know is what features are being used. Is the issues tab ever being clicked. Do people ever click on the new/trending/updated things in the footer. What are the common links out. Etc. Behavior things, which is exactly what the analytics provides.
If somebody is going to use analytics to improve our websites, I totally support using Matomo or any other solution
oh man. This is one of the biggest show cases for why doing things in public is so important. I hadn’t even heard of that before. It looks pretty cool. I may install both and do a side by side comparision, but if I can track what people are clicking (not links) on with plausible, that might be better. Matomo has a lot of features which I can’t see me ever using
I’m not sure what that data will provide, especially since I have access to the search results from google into the site, and algolia on what people are searching on the sit
I’ve found it a little bit helpful to see the Algolia reports for plugins.jenkins.io. They show that people are still searching for tfs and those searches are not finding the tfs page. Unfortunately, I don’t know what to do with that information to better help users see the current state of the Jenkins TFS plugin.
However, I assume that information from Algolia search will continue to be available to us.
yea, because people are searching on algolia, its not going to go away. I wasn’t planning on removing anything.
Matomo has nice flows, so you’d be able to see someone came in from google searching for “tfs”, which led them to /plugin-foo then clicked the button labeled “github”, and left the site.
I’ve never managed to see that flow easily on google before.
Matomo Cloud support got back to me today. I’m not really good at asking for sponsorship, so I think that part was missed. They pretty much told me:
Matomo Business Cloud - 12,900 USD per year (excluding discount)
up to 5M pageviews per month
up to 30 websites
Learn more: https://matomo.org/pricing/
Honestly, I don’t think its worth it for this. I could host it on my home cloud for a month or so. Honestly its super easy to setup on the infra k8s (just needs mysql).
I don’t really have strong feelings on what to do next. I was just hoping to find out if the footer is used.
Self-hosting is always an option. Also it might be a use-case for using asking smaller public cloud sponsors like DigitalOcean, IBM, or maybe CIVO Cloud if someone wants to experiment
I am all in favor to switching to something else than Google Analytics for plugins.jenkins.io .
I am a happy user of plausible for personal projects.
It provides less information than Google analytics but it shows me what I want to know, where the traffic comes from, bounce rate, time spent, and page visited.
What I like with plausible is that we can make dashboard public.
I have a paid account with enough quota so I can quickly add plugins.jenkins.io but I doubt having enough quota for jenkins.io as well but if we are happy then we can explore solutions.
I would avoid hosting such services ourselves because then we would have to be sure that data are correctly persisted otherwise it lose it’s value.
With jenkins.io I like that we have data to compare current traffic with what we had 5years ago.
Once we add <script defer data-domain="plugins.jenkins.io" src="https://plausible.io/js/plausible.js"></script> to plugins.jenkins.io<head>, then we will see data to Plausible · plugins.jenkins.io
Thanks @halkeye and @halkeye, it’s working I especially like that the analytics becomes public with this change. I do not think we have anything to hide, and it is another valuable tool we could share with maintainers and adopters/vendors.
It doesn’t solve any of my needs, but I have no objections to it. After the conference tomorrow I’ll see if work will let me throw up a VM with matomo on it. Failing that i’ll run it on my own account for a few months. I want to know what they are doing, not who they are.
I especially like that the analytics becomes public with this change. I do not think we have anything to hide, and it is another valuable tool we could share with maintainers and adopters/vendors.
