In-Depth

App Development Survey Shows Most Projects Still Running Late

Application development delays of a few weeks are common. On the other hand, the vast majority of projects wrap up within three weeks of their target dates.

• By Stephen Swoyer
• 03/08/2011

A recent survey from application development consultancy Evans Data Corp. underscores what most programmers already know: software development projects tend to run late. According to Evans Data, just under half (49 percent) of all application development projects don’t finish on time.

That’s specific to North American software development projects. App development projects in other regions run even later: Asia-Pacific programmers, for example, trailed their North American counterparts in the on-time category, while coders in the EMEA -- a region that encompasses Europe, the Middle East and Africa -- did much worse, on average. Only a little over one-third (36 percent) of EMEA programmers finished their projects on time.

There’s “late” (e.g., two or three weeks) and then there’s really late -- e.g., delays of several months. In all regions, thankfully, most app dev projects tend to finish within three weeks of their projected deadlines.

Developers in the Asia-Pacific region actually have the best track record in this regard: almost three-quarters (71 percent) of Asia-Pacific app dev projects finished within three weeks of their target dates.

North America was a close second, with two-thirds of all efforts wrapping up within three weeks of a target deadline. Programmers in the EMEA once again fared much worse: less than three-fifths of app dev efforts in the EMEA finish within three weeks of their target deadlines.

What’s striking is that everyone -- from developers to project managers to executives further on up the organization -- seems to expect some acceptable degree of tardiness. According to Evans Data CEO Janel Garvin, this has to do with what might be called the bifurcated nature of application development: it’s at once a science and an art.

“[These results] might not seem so bad to people who have been active as product managers or development managers,” writes Garvin, on her company blog. “Fact is it’s hard to predict how much time a software project is going to take. How can you know how the bugs will be injected, where they’ll be, and how long it will take to find them? Software development is in many ways more an art than a science.”

In every case, coders tasked with writing “internal corporate apps” were most likely to finish on time. In both North America and the EMEA, projects were most likely to be canceled while still in their coding phases; in Asia-Pacific locales, projects were typically canceled once they went to (external) beta testing.

Canceled projects aren’t always for naught, according to Evans Data. In many cases, shops are able to reuse discontinued code in other projects.

Intriguingly, agile projects tend to have higher cancelation rates than do traditional (or “measurement-based”) app dev efforts. “[W]hile developers love [a]gile, it was measurement-based models like [S]ix [S]igma or ITIL or projects using a waterfall model that were least likely to be discontinued,” said Garvin.