How We Selected Our Initial PAs

So far we've:

• decided to adopt the CMMI for Services, and
• determined that because of our small size, we really can't focus on more than one or two process areas (PAs) at a time.

The Tie-in to Our Business Goals

We also feel strongly that:

• the first set of process areas we tackle must deliver immediate business value... not value in 2010, or even in the second half of 2009. We want bang-for-the-buck over the next several weeks.

One implication of our "immediate return" approach is that to understand our initial PA selection, you need some insight into our two key short-term business activities:

1. Overhaul website. We'd like to do a major reworking of our website. (For competitive reasons, we can't say much more.)
   
2. Fill public classes. We're teaching a series of public CMMI courses starting in April, and we'd like to fill those classes -- cost-effectively. (Although it was great to have 19 students in our last public Introduction to CMMI class, our current profit margins on public classes are too slim to be sustainable.)

The only CMMI-SVC PAs of interest to us right now are the ones that make it easier to accomplish at least one of those two objectives. Period, end of statement. Other PAs? Sure, we may tackle them... eventually. We just can't afford (literally) to be too concerned about them right now.

Our Selection Process

The process we used to select our PAs is hugely dependent on two key pre-requisites. If you're going to try something like this, make sure you:

• Understand the CMMI-SVC. At a minimum, take the CMMI-SVC class, either from us or another SEI Partner -- we've heard rumors we're not the only company that teaches it. (Note: the three-day CMMI-DEV class is currently a prerequisite for the one-day CMMI-SVC class, since the former covers basics that the latter assumes you already know.)
• Understand your business. Specifically, have a knowledge of organizational "pain points" -- where you believe your business is hurting right now.

Given those prerequisites, here's what we did.

1. Listed all 24 CMMI-SVC process areas.
2. Set up four categories - Probably, Maybe, Probably Not, and Definitely Not.
3. Made a preliminary assignment of each PA to one of the four categories.
4. Took a deeper look at the Probably's and Maybe's (down to the specific practice level).
   
5. Chose the PAs of most immediate value to us.
   

One simple question was used to categorize each PA (step 3):

• If we were to adopt significant portions of this PA in the next few months, would it play a major role in helping us accomplish our two key near-term business objectives?

The assignment went quickly and smoothly, helped in large part by some no-brainer additions to the Probably Not or Definitely Not categories. For example, we really wouldn't want to target any of the "advanced" PAs (like Quantitative Project Management) that build upon "basic" PAs (such as Project Planning or Project Monitoring and Control.) We also decided it wouldn't make business sense to initially target the organizational PAs (Organizational Process Focus, etc.), because they simply wouldn't give us the immediate return we're seeking.

Once we had this first cut at categorization done, we took a closer look at the Probably and Maybe PAs (step 4). We examined each of them down at the specific practice level, and came up with our final list of just a few PAs (step 5). Our next blog entry will reveal the results.