In refactoring this site, I accidentlally removed my analytics script. And then proceeded not to notice for about a week.
I've always been pretty torn about analytics. On one hand, it just feels good to have numbers. On the other hand, those numbers may have nothing to do with my actual goals and yet influence me to change my goals to increase them.
If I want to have analytics that work for me, I need to start with my goals and work backwards to them (duh).
This website is a place for me to work with the garage door up. Especially this fast section, but really everything here, is in an unfinished state, that's slowly progressing towards something more.
There are a couple sides to this. One is it forces me to shape my work in a way that's legible to others, which is very effective at teasing out the ambiguities and inconsistencies I'm holding in my head.
It also opens me up to serendipity, to someone out there seeing my work, and reaching out. That connection could be anything, words of appreciation, or a link to something I should read, or a conversation.
Separate from me, working in public can also create value for, well, the public. If I write about the problems I'm having, or the unfinished things I'm working on, someone might get something out of seeing those. It could be concrete knowledge on how to solve some problem, or even just solidarity in seeing someone working on something they are thinking about.
I'd divide existing web analytics systems into two categories. There are massive systems like Google Analytics, and Matomo, and then there are smaller simpler systems like Goat Counter, Simple Analytics, and Fathom (what I currently use)1.
The latter generally have way better privacy and way fewer features, but still fundamentally give you the same kind of information. Things like views, time on site, etc. These things generally capture "engagement", how much you hold people's attention.
This isn't an inherently bad metric. Attention, to a certain extent, correlates with a lot of other things, like information learned, or connection to the things people are interested in. However it correlates a lot stronger with things like how "clickbaity" your website is, or how trendy the topics you're talking about.
They're also inherently about scale. More people = more views. I don't want to optimise for the most people, but for the specific subset of people who would add the most value to my life, and I to theirs.
So what would analytics for working with the garage door up look like?
Well, the first (and simplest) option, is it would look like no analytics at all. Ditch the numbers, just keep working, and get data by doing regular self checkins and retrospections.
Another option could be tracking something like "conversations started". For my newsletter this could be replies, for the website maybe backlinks. The idea here is to track how much people want to meaningfully engage, as opposed to superficial clicks.
Another option is to translate my "north star", of drastically improving the way people learn, into some sort of metric a la Patrick Mckenzie's success function:
Fifteen years into this career thing I definitely know what my success function for the next thirty is: the product of the number of people I’ve helped in the community I serve (software people, broadly writ) times the delta in the average life that my efforts uniquely caused.
I think my own is pretty close to that. Something like: the total delta in the the knowledge and skills that my efforts uniquely caused.
The hard part of course remains. Patrick approximates with income as a metric, but I'm not sure what that is for me.
Ultimately this isn't really a analytics problem, it's a self-reflection problem. I have my north star (Improving the systems the world uses to educate people), but I don't have a clear way of measuring it, let alone measuring it in realtion to my site. Slapping a view counter on my site isn't going to fix that.
Ideally what I want is a system for iterating on metrics, a way for me to reflect on my metrics, and switch them out easily. A very programmable dashboard. Sounds something like fancynote.