Modeling and Documentation
Practices on IT Projects Survey Results:
The survey results are summarized in my November 2008 DDJ column "Newsflash: Agilists Write Documentation"
Some findings include:
- Agile teams are more likely to model than traditional teams. In
Figure 1 you can see that the black bar, the one that
indicates that the team isn't modeling at all, is much small for agile teams
than it is for traditional teams.
Iterative teams are most likely to model.
When traditional teams used software-based modeling tools (SBMTs), it
was for documentation.
A small number of agile and iterative teams were using SBMTs for
Although there was some slight differences, agile and traditional teams
are equally as likely to create deliverable documentation. See
. For more information, read my
For all the talk in the agile community about acceptance test driven
development, few teams are actually doing it in practice. The
2010 How Agile Are You? survey
also found that acceptance TDD was
only being done by 44% of teams claiming to be agile.
Over half of agile teams, and 60% of traditional teams, write detailed
requirements specification documentation.
A little less that one fifth of agile respondents indicated that they
were doing design-level TDD. However, the
2010 How Agile Are You? survey
found that 53% of people claiming to
be on agile teams also claimed to be doing developer TDD.
The most popular approach to architectural modeling for both traditional
and agile teams was to do high-level diagrams. See my
article for greater detail.
Traditional teams are twice as likely to create detailed design specs
(arguably a bad practice) as agile teams. See my
article for advice.
Figure 1. Primary modeling strategy.
Figure 2. Deliverable documentation creation.
You may use this data
as you see fit, but may not sell it in whole or in part.
You may publish summaries of the findings, but if you do
so you must reference the survey accordingly (include
the name and the URL to this page). Feel free to
with questions. Better yet, if you publish,
please let me know so I can link to your work.
I was disappointed with the low number of respondents. I
suspect that the timing of the survey, I sent it out in July when
many North Americans and Europeans are on summer vacation, had an
impact. The topic, modeling and documentation, was likely
considered to be rather boring.
- I am very interested to see the research community do some empirical
research surrounding actual modeling and documentation practices on IT
projects. This survey has revealed some significant differences in the
rhetoric that we hear, from both the traditional and agile communities,
surrounding modeling and documentation. It would be great to have
conversations based on actual fact instead of the religious discussions that
we seem to have. The traditional view of modeling has been foisted
upon us for several decades, yet my experience, and the results of this
survey, are much different than the "software engineering vision" which the
professional modelers among us promote.
- This survey suffers from the
fundamental challenges faced by all surveys.
Links to Other Articles/Surveys
- My other surveys
Why Share This Much Information?
I'm sharing the results, and in particular the source data, of my surveys for
- Other people can do a much better job of analysis than I can. If
they publish online, I am more than happy to include links to their
- Once I've published my column summarizing the data in DDJ, I really
don't have any reason not to share the information.
- Too many traditionalists out there like to use the "where's
the proof" question as an excuse not to adopt agile techniques. By
providing some evidence that a wide range of organizations seem to be
adopting these techniques maybe we can get them to rethink things a bit.
- I think that it's a good thing to do and I invite others to do the same.