In the past, I’ve wondered what % of people upgrade immediately vs. how many wait, and how long they wait. With the introduction of https://analytics.home-assistant.io/ in 2021.4, it became possible to kinda see it, but that site doesn’t show the upgrade trends/history. So, I started collecting a snapshot once a day beginning when 2021.5 was released.
Same data as above, stacked:
Some comments:
- About 10k installs are upgraded within 2-3 days of EVERY release (major or dot release).
- Yet it isn’t the same 10k people doing all that upgrading. Many thousands will upgrade once and stay parked on a version for a long time…
- … meaning that even now as 2021.7 is about to be released, well over 30% of the installs are NOT running /any/ version of 2021.6.
- Two weird things happened right before the 2021.6.2 release (about 7 June):
- The rate (slope) of “total installs” shifted drastically
- After having been very stable for a month, 2021.4 installs abruptly started falling. I assume they were upgrades, but there’s no way to be certain. It’s strange in that it’s consistent: about 500 per day And double strange that it’s staggered, meaning that first it was 2021.4.0 systems , then 2021.4.1, then 2021.4.2, and so on, all the way up to 2021.4.6 – as you can see in the graph below.
I really haven’t come up with a great theory to explain either of the weird observations.
Note: the above is using a log scale.
If you’d like to see or copy the raw data, it’s on the last tab here:
Raw data notes:
- The green data is the staggered upgrades happening
- The “upgrades” columns assumes that nobody is uninstalling or turning off analytics. This obviously isn’t true, but presumably it’s small enough it can be ignored
- I still have the daily .json files that I could upload or transfer if someone is interested in this going forward. This was interesting, but I don’t have any plans to do further graphing.