Stadia definitely had software engineering related issues. The latency and stream reliability might have been better than competing services, but I found the UI and store to be quite lame. Features that were promised at launch (e.g. YouTube integration) took years to show up and weren't as good as advertised. Launching a game took too many clicks and way too much waiting. It should have been instant and that alone would have been huge. Porting games to Stadia was too hard leading to an anemic catalog; meanwhile Valve knows that the catalog is everything and engineered Steam Deck to run unmodified Windows game binaries on Linux! Google engineers ultimately failed to take advantage of the unique advantages that streaming should have given them, and without any compelling unique features Stadia was just a worse version of consoles with a subscription fee.
The pricing model was probably the biggest single issue, but I believe better engineering could have made a huge difference. This is a common pattern for Google products: a technically impressive core is held back by bad engineering and business decisions at higher levels. Duo is another good example IMO.
The YouTube integration launched after one year. The games were a single click, had direct links and could be launched by voice assistant on any compatible cast device. It was also as easy to port games too as any Linux port would be. Online play was free, subscription unnecessary unless you needed 4K and more than 50 games in the subscription on sign-up.
Launching a Stadia game was never truly one click for me. It always required several, and then you'd have to endure the same old load screens and splash screens and menus as any other platform. The load times should have been faster, and they should have had platform requirements to make it faster to jump into gameplay without nagivating menus after load. I heard them talk about deep links directly into gameplay but I never saw it, it must not have been implemented in many games. But I did see a bunch of Stadia account related clickthrough BS that added more clicks and more load time (and sometimes bugs) when all I wanted to do was play.
> It was also as easy to port games too as any Linux port would be
And nobody bothers to port to Linux either. It isn't trivial in many cases. And I believe Stadia wasn't just as easy because of a Vulkan requirement along with other platform integration and certification stuff. An unhappy medium of too many platform requirements to be trivial to port to, but not enough to actually make Stadia a better experience than other platforms.
> Online play was free
Only if you bought games at full price, trusting Google to run the service forever. Clearly a risky bet, and the risk was obvious to everybody at the time, not just in hindsight.
> more than 50 games in the subscription
The selection was not compelling compared to other platforms' subscriptions. If they had had a larger catalog to draw from, that could have helped.
The pricing model was probably the biggest single issue, but I believe better engineering could have made a huge difference. This is a common pattern for Google products: a technically impressive core is held back by bad engineering and business decisions at higher levels. Duo is another good example IMO.