In November 2011, the lead of the Accelerated Solution Delivery (ASD) practice within IBM Global Services, Paul Goras, and I ran a survey on the Agile state of the art. The survey had two goals:
- To explore which adoption strategies, such as executive sponsorship and assigning people to a single agile team, were related to agile project success.
- To explore the affect of two common scaling factors, team size and geographical distribution, on agile project success. There were 168 respondents to the survey, which was announced in several agile mailing lists and on my Twitter feed.
Of the respondents who believed they were on Agile teams, 42% believed they were on successful teams, 28% believed they were on challenged teams, 3% said they were on failed or failing teams, and 26% said it was too early to determine the success of their project. So, excluding the last group, 57% were on agile teams deemed successful and 38% on challenged teams. (There were too few failed agile teams to warrant analysis). In our 2011 Dr. Dobb's IT Project Success survey, we found that agile teams reported a success rate of 67% and a challenged rate of 27%, although that survey had a more robust definition of success and challenged than did the present one.
This survey explored several potential adoption accelerators. When it comes to environmental factors, we found that successful Agile teams were more likely to have dedicated business expertise involved with the team, have a working environment and tools that support agile development, and were organized to support agile development. None of this should be a surprise. Regardless of paradigm, it behooves you to make people with domain knowledge available to your development teams, indicating a need for you to educate your stakeholders as to the importance of this and to motivate them to be actively involved. Clearly, Agile teams are going to need tools that reflect the Agile way of working — an observation that will surely frustrate organizations that have made a large investment in tools for traditional development approaches.
We also explored potential adoption accelerators that were management or governance focused. We found that successful agile teams were more likely to be in organizations with effective executive sponsorship of agile approaches. When it came to governance, I was happy to see that successful agile teams were more likely to be measured based on value creation (compared with challenged teams) and less likely to be judged based on traditional metrics. The survey also found that successful agile teams were more likely to be in organizations with an agile path in their governance strategy than were challenged teams. I've written in the past that agile teams are easier to govern than traditional teams, much easier in fact, so I was saddened that many respondents indicated that their organizations still hadn't improved their governance strategies.
The survey explored two of the eight scaling factors of the Agile Scaling Model (ASM); namely, team size and geographic distribution, and their relationship with project success rates. First and foremost, respondents indicated that they've been successful on large agile teams, including teams of 250 or more people. We also had responses of success at all levels of geographic distribution, including teams spread out across the globe. These results were consistent with the Dr. Dobb's November 2009 Agility at Scale survey. So, don't let anyone tell you that you can't scale agile.
When it came to team size, there was a definite correlation between success and smaller teams. When considering teams of 100 or fewer people, the average size of successful agile teams was 12 people, and of challenged teams 16 people. (Note, however, that because there were several respondents indicating their teams were very large, inclusion of that data unreasonably skews the average team size.) This is similar to what we've have found in the past, particularly in Dr. Dobb's July 2010 State of the IT Art survey, which found that regardless of paradigm, larger teams we less successful than smaller teams. It also found that on average, Agile teams were as likely or more likely to be successful than traditional teams regardless of team size.
We had some interesting results concerning geographic distribution. First, only 26% of respondents indicated that their teams were entirely co-located. We've also seen similar numbers in previous surveys, belying the belief that the majority of agile teams work in the same room together. The good news is that 41% of respondents indicated that team members were near located (within reasonable driving distance), and only one third far located (some team members are flying distance away). For successful teams, roughly one third were co-located, one third near located, and one third far located. For challenged teams, only one in seven were co-located, three sevenths near located, and three sevenths far located. The greater the level of geographic distribution, the greater the risk due to communication and coordination challenges, resulting in a lower success rate.