Where did we drop the ball? The point of this isn't to point fingers or assign blame, but rather to identify where our processes and strategy are not serving us well.
This time, I will actually resolve the market to a rough approximation of what I think actually didn't go well. Place your bets!
Last time: https://manifold.markets/JamesGrugett/manifold-team-janq4-retro-what-coul
[j] better segmentation here would be good - is there a way to tell which users came from an ad
[t] many referred; but one piece stocks - they found it on a Twitter ad. [s] we track the campaign, but it doesn't necessarily get everyone
[d] Could just do (ND1 / signups) ratio. As long as that ratio goes up (or is relatively) [j] people from one piece are less good users than normal, then it should go down
[jf] Should put all of the one piece into their own group chat [t] retention would be a lot better if there was a community/subreddit for these
[d] status of shorts?
[t] put up some shorts, and one person who sponsored to watch it; it was overproduced, and the musical stuff was hard to follow without knowing the rationalist terms. Spent $5k; got 5k viewers, but the production itself wasn't a great ad (just a good show)
Would have been better on Twitter; we scheduled stuff on the restream, but it got blocked on the Twitter side. few problems that interfered with the success of the marketing
[c] if it's good and you change it, people are unhappy. [s] if we already had something decent, people will care a bit more
[j] power users didn't want this at all, unfortunately
[i] people who are on the dev site checking out are extreme power users
[t] good to have shown it to some new users -- power users will complain about anything that's a ibit different [jf] yeah
[s] if power users say this thing is too complicated, that's bad. if they say "I want these new features", you should ignore them
should be measuring outcomes more - a simple A/B test could have gone a long way here. Still don't know what the sign & magnitude of the change would be
[i] if that trades off against just doing a user test, that's not worth doing. The user test test is much more informative
[jf] could just ask users if they like the change [a] good, but user test surfaces many more bits of info
[j] not that bad in the grand scheme of things; they're temporary. They're also very distracting to our team's dev workflow
[i] bad that we have opaque backend processes that only marshall understands
[a] wasn't there a plan to offload knowledge. [j] Biggest thing is how he configured the deploys
[c] some of us need to be doing more devops, process-y, backend-y stuff to have the skills in-house. bit intimidating but worth stepping into it
[j] turning point was the feeling of "there's a lot of low-hanging fruit" after user interviews.
think there's a lot of other things - creator program, numeric markets -- there's a lot more still to do
[s] updated in favor of, rather than a separate site, targeting a different audience and reshaping the site to work well for them. Politics betting is the next largest thing by market size
@strutheo oh ok, just wanted to check if there was a specific reason u put it as an option haha