Choosing the "best" platform for the client has both a technical aspect and a financial aspect. In most cases that I've seen in my 20 year career we probably could have gotten the job done with nearly any technology. The choice was ultimately down to whatever platform was most comfortable for the developers or, to be honest, just plain snobbery on the part of one or more of the team.
Of course there may be an ideal platform from a technical sense. I may have a project and I decide that it's 100% ideally suited for Scala. Using PHP may be only 85% ideal. However, is that really best for the client? Are they going to be forced to blaze new trails for things that are already solved in PHP? Is the client going to be able to find Scala developers when we leave? How about in 5 years? How much are they going to pay for that? Is it worth the 15% of efficiency that was gained?
There is no real correct answer here, but I guess my point is that when you are talking about what's "best" for the client, it's not always just a technical question.
Of course there may be an ideal platform from a technical sense. I may have a project and I decide that it's 100% ideally suited for Scala. Using PHP may be only 85% ideal. However, is that really best for the client? Are they going to be forced to blaze new trails for things that are already solved in PHP? Is the client going to be able to find Scala developers when we leave? How about in 5 years? How much are they going to pay for that? Is it worth the 15% of efficiency that was gained?
There is no real correct answer here, but I guess my point is that when you are talking about what's "best" for the client, it's not always just a technical question.