You want to test and develop your applications against real data. But it's hard to get good staging data.
The typical development or staging environment has very poor data in it - out of date, unrealistic, manually entered. Or worse, it'll be data copied straight from production - risking privacy.
Generally it won't be big enough to be able to properly test performance and code correctness before features hit production. And it won't be realistic enough to get a sense of how things will look and feel with real data.
What you really want is anonymized, up-to-date copies of production data - ideally, you want "subsetting". You want staging and development databases big enough to develop and test your features, but not full copies of production that take hours to create and move about.
We're building exactly this - fresh, anonymized, subsetted development and staging databases for every branch, automatically.