Citibank is cited as an example where the shift is from measuring the returns to maximizing them. Long article but with some interesting trends.

If you can't measure then how can you be sure that you are not maximising some kind of dead end with a big brick ceiling just high enough to hit after you start project?

To me that "maximising" sounds more along the lines of fine-tuning of a project, but it would certainly be good to know that project has got underlieing positive ROI.

That process of maximisation might not be cheap either - internal and external resources could have been used on something more productive.