i work at a top 50 fortune company that have insane monthly, quarterly and super-sized end of year batch data loads. i manage the dept. that makes sure we are properly scaled in non-prod to predict prod load handling during busy times w/ our 3rd party vendor installed in our system. we started using AWS in-addition to our existing in-house hardware. i even brought in Exadata 6 years ago which was costly but more than made up for perf issues w/ the vendor's software on the 10 app server mode cluster (we started w/ 4 app server nodes)
bottom line is that like me, IT dept leaders at places such as amazon, netflix, ebay, comcast xfinity, and walmart learned to scale properly. while it's technical, that's also a method of the "crazy made scientist" madness for making decisions on when and how to scale.
when the ACA website opened shop, they learned the hard way. well, the lego, hbo-go and ticketmaster web sites still function like the old ACA one which itsself looked like in was built in 2005.
there's no excuse for poor performance just like poor security. only diff is you can blame performance and inability to take orders on your tons of customers and excellent products. can't use that excuse when it comes security as that's a diff beast.
thus, i think honestly lego wanted to brag about 75192 being so popular, consumers overwhelmed the web site and then imply their web site is strong enough to handle it but couldn't due to way too much 75192 demand,