April 18, 2007

In a previous post I offered the question, “Is Ireland a Talent Magnet?” and from various statistics about the health of the software industry, the answer was a resounding yes. Being called Europe’s Silicon Valley is no small moniker, with Ireland exporting billions of euros worth of software each year, while the country’s entire population of 3.7 million is only about 1/3rd that of New York City alone. Goodness!

In light of that, I think back on a conversation I had with Seamus Gallen of Enterprise Ireland a couple of weeks ago over dinner. He told me that as attractive as Ireland is for technologists, it’s still difficult to attract and recruit young people, and have them stay. As a result, some companies have no choice but to outsource work to places like Eastern Europe or India in order to design and build products. That means companies have to establish a core competence in managing outsourced development, which is no easy task.

During a panel discussion, I was asked a very pointed question by an audience member. In light of the many negative stories about outsourcing gone awry, could I share a maxim or two about what companies do when outsourcing goes well? I had to be thoughtful about this, and I hope I succedded, but what came to mind ironically, was a U.S. based client of mine that had outsourced to – of all places – Ireland. (I realized it could have sounded like I was sucking up to my Irish audience, but really I wasn’t 🙂

In this example, I recalled two distinct characteristics of their projects, which showed lower defects than other projects that we had benchmarked, along with higher productivity. First, they were very careful about how much parallelism they attempted between the requirements/analysis phase and the design/code/test phase. This was important because these two disciplines represented the teams split across continents. Secondly, they did not use large teams. Instead, the offshore Ireland team was a small(er) cadre of engineers, compared to the monstrous army that was applied on other projects that were outsourced.

Tricky thing about these two attributes is that to choose them gives the appearance of increasing schedule risk – just the thing that management abhors. But in actual reality, when all was said and done, the projects came in faster and with fewer defects than projects which overlapped the analysis and build phases, and used large teams.

So, my advice to companies that feel they must outsource is to consider applying this tactic, even if it seems to be the opposite of what every cell in your body is screaming for you to do.