Politics.
Technical hurdles (getting / massaging data) might be challenging or time-consuming but they're straightforward problems.
Dealing with people who feel possessive of their data or fear having their laundry aired are the real challenge - I want to say 90% of the problem even.
This might be a non-issue in a small company, but it has been consistent death in the places I've dealt with it (10,000+ employees).
For example...
Company 1:
The public is clamoring for access to data that the company has been holding tight to its chest and disseminating sporadically on paper and later through a painful website that only returns one "result" at a time.
We track down the data sources, create some shims to get it in an easier format and even write some code that will make the output more useful to the public.
First, the director in charge of 70 or so people who deal with the traditional entirely manual system of handling this data freaks out.
The director claims that what we're presenting must be wrong because it doesn't match the stats that someone has been delivering to him manually. We demonstrate that our stats are actually more accurate - he doesn't care - we try to move forward with the raw stats as he likes them.
Next, It comes out that certain external agreements are based on juked stats. If we present the data as is the company is afraid that they will end up having to deal with very difficult questions or a lawsuit.
The project gets killed.
Later, we're asked to come up something in the same area, but purely for internal use. As requested, data is combined and averaged down to a short list of half a dozen percentage based metrics and a single "traffic light" for overall health.
Again, someone in the chain has been lying about their 100% stats and flips out when their numbers only add up to 97% and they get a "yellow". We refuse to fake the stats so they ended up tasking some administrative person with cleaning up our data to the liking of the management before every meeting in which it would be reviewed.
Company 1 - Situation 2:
Generally similar to Situation 1, except this time to jilted director goes for the trump card - says revealing his data would be a threat to national security.
Discussion over.
Company 2:
Major public-facing failure. Executives think a dashboard will help them spot trends and keep the rest of the company honest.
Departments directly responsible for the failure simply refuse to allow access to their data (actually just make the process so slow and bureaucratic as to be pointless). The project dies on the the vine when the pain of the failure has sufficiently faded from the executive memories.
-----
In the end, dashboards are just information, they can reveal problems directly or indirectly, but they won't solve them.