Did you know that teams where all the members are dedicated to that one team produce half as many defects than teams that have even 25% "multi-taskers"? A relatively simple decision (dedicate folks to one team) for such a huge gain in quality.
Agile coaches know the above statement is true, but they have never before had the data to quantify HOW TRUE. When you pile on how much productivity, responsiveness, and predictability are impacted by this "dedicated team" decision, it becomes a lot easier to overcome objections of "specialized skills".
For the last year, Rally Software has allowed me to crawl around in the data of over 20,000 Agile projects. We can learn a lot from this non-attributed data but we also sent out surveys so we can correlate individual agile practices (daily standups, sprint planning, TDD, pair programming, etc.) with the goal of confirming (or refuting) the most common agile recommendations and practices. Rally also sponsored the partnership with the Software Engineering Institute (SEI) at Carnegie Mellon to help validate this research. Come to this session to have your most passionate convictions about what Agile practices work and which ones do not confirmed with real data.
WARNING: Do not come to this session if you are not prepared to hear data that refutes your most closely held beliefs about Agile.