Recommendations to others considering ROOT:
Try it first, and best is on Windows - with version 5, not 6 (still very raw). Best if used from python (e.g. import ROOT) as this takes away lots of c++ overhead and helps to start fasted.
Once on linux, always compile from source, NEVER do 'make install' as it scatters the files int he system and will not allow for easy upgrade. 'source thisroot.sh' is there to set all variables needed in the terminal.
read examples, try simple things, then read the manual - its width discourages new users often.
ROOT libraries are fast - when imported from python, they speed up the execution of the task significantly. Review collected by and hosted on G2.com.
What problems is ROOT solving and how is that benefiting you?
Main use is the data analysis. visualization of data, proper fits producing the ch^2/likelihoods and correct uncertainties on fit parameters that are PROPERLY dependent on both y and x error bars (note that a lot of software I saw treats x-bars incorrectly or just ignores them). Different minimization engines allow for flexibility and cross-check in difficult fits that are done in batch mode.
Another feature is the TSpectrum, allows to find peaks in the data, useful in finance and final report analysis. All libraries are fast and are fit for on-the-fly analysis in the world of finance.
Fast and compact storage file format is good for backups. Datafiles can be opened from server via different protocols with no need for a local copy.
Making tools with GUI for clients use is very beneficial as they all have same-looking interface and are portable to windows-linux-macos platforms without changes to code that deals with ROOT. of course there are some differences in general but at least this saves major headache of rewriting GUI for different graphical libraries that typically is the case. Review collected by and hosted on G2.com.