Author here. I build this to learn and ship something with Elixir.
I know it's not a new idea, there are plenty of alternatives to track unfollowers, but couldn't find any that works inside twitter.
I got the idea recently reading some political tweets about a situation in my region (Catalonia, Spain). I usually follow tech people and this days they're tweeting political stuff and I wondered how many people unfollow them after those tweets.
Could you elaborate how this works? I imagine you're comparing your followers and then scanning for each, prev followers list and new one, so notifying when a diff detected.
How do you think a good digest could be made so a twitter user with a long list of followers can receive notifications not bothering too much?
I have a process to always follow my followers (and unfollow unfollowers) so I can DM them.
Regarding the checks, the initial free plan (just following my bot) comes with weekly checks. That means I'll fetch your followers once a week and compare with last week's, find out who is missing and sending you a DM with the list of all unfollowers. So you'll basically a weekly DM with unfollowers.
Daily, hourly and realtime plans work in the same way but with a higher frequency, but without any kind of digest, just one DM with the list of unfollowers in the time window.
6 comments
[ 3.8 ms ] story [ 20.4 ms ] threadI'd appreciate any feedback :)
How do you think a good digest could be made so a twitter user with a long list of followers can receive notifications not bothering too much?
Regarding the checks, the initial free plan (just following my bot) comes with weekly checks. That means I'll fetch your followers once a week and compare with last week's, find out who is missing and sending you a DM with the list of all unfollowers. So you'll basically a weekly DM with unfollowers.
Daily, hourly and realtime plans work in the same way but with a higher frequency, but without any kind of digest, just one DM with the list of unfollowers in the time window.
I don't have a monetization model yet, but yours looks really fair.