Friendster is frequently cited as a potentially massive company that failed due to its poor performance: https://www.techinasia.com/awesome-startups-failed
Note that I’m specifically talking about improving bad performance. If I was talking about having good performance, largely meaning designing the app to be fast from the start and never letting it get bad, the figures would be a bit more favourable, though they would still be very heavily skewed against large teams or companies. My conclusion is simply that the more people you have, the harder it is to make fast stuff.
I’m quite serious about these numbers. Not all small teams can produce fast apps, but over time no large teams can produce fast apps unless they very deliberately design it in from the start, in which case it’s merely extremely difficult to maintain (again, over time).
You can still opt in to keep using their old interface thankfully, the difference in both performance and UX is staggering. When that option disappears from the site, so will I.