Yes you're right; I started working full-time on it 2-3 months ago, and it's a real pain in the ass to get all the data (especially to find the dates of new product/service launches and so on). Moreover, I needed to develop a framework (some sort of service categorization) of the ecosystem to analyse the emergence of new sectors and market segments over the years. This framework is of course constantly adapted during the research process, it's really fascinating to see the countless applications that have seen the light in recent years.
So far, I managed to include some 350 companies/projects in the database with "full" information (at least the minimum data I need for my analysis, which so far are the foundation date, launch date of each product/service in case they are diversifying, geographic location, funding, possible mergers & acquisitions, as well as the "exit date" and the reason of the firm going out of business). The main sources are the company's websites (difficult if they are out of business for several years), company blogs, news articles, bitcointalk forum, reddit posts, linkedin, twitter etc.
It's a really time-consuming project, but also incredibly exciting since it gives fascinating insights in the development of the ecosystem. So although it's a lot of work and sometimes annoying, the experience so far has been really worth it (and I hope it will be useful for other people too once it will be published).