Deploy a poorly-tested, monster feature on launch day?
Ted Schaefer
9 replies
I've been working on math support, so StudyWise can quiz you on your calculus or offer next steps for your geometry studies. It's a huge feature because math has to be integrated with our interactive capabilities.
And until yesterday, it didn't work at all.
I'm crazy, don't test me. I'll freakin' do it!
Replies
Gong Zijian@zijian
VIVA
Deploying a poorly-tested, monster feature on launch day might seem bold, but it's a risky move that can lead to unforeseen issues and user dissatisfaction. It's often wiser to prioritize thorough testing and incremental updates to ensure stability and positive user feedback.
Share
Zappi Ad Predictor
Are you using feature flags or equivalent? We've used them for the last year or so - it's amazing knowing that you can generally push things to production and test them there...
Albert
@nik_hazell Nope, not at the moment, but we're using Vercel so it's really easy to select an older deployment if we need to.
Zappi Ad Predictor
Albert
@nik_hazell definitely, it's a great product! i used it at a previous job π
Sugar Free: Food Scanner
Think twice! The first impression is very important!
I'd better keep it for later. No need to risk it at the last moment! π
This is where I see much mixed opinion and I'm trying to find the right approach as well. The wisdom with launching MVP seems to be lauch as early as you can and iterate. I'm just not quite sure how literally to take the 'as early as you can' bit.
Perhaps the happy medium in your case would be to include a very clear 'test feature' or 'beta feature', to at least show the user you are aware it might not be working as it should.
Albert
@jake_saul I'm definitely in the "early means EARLY" camp - but I am holding off on this until I can do a little more testing later today. It's pretty easy for us to undeploy if necessary, so that makes it a little easier to experiment.