Handicapping Level 3's Provisioning Fix

Yesterday, I had a question on one of the investor message boards asking how long I think it will take Level 3 to fix their provisioning issues, and even though nobody outside of the company really knows the answer to that, I thought it might be worth prognosticating about it, and that folks might be interested to hear what I think since I have had a long career in software development of telecom systems, and since I worked at Level 3 for six years.

Let me start by taking a guess at the size of the problem.  Level 3 reported problems in provisioning orders due to difficulties in integrating the disparate systems of their half-dozen acquisitions over the past few years.  However, if experience is any guide, Level 3's integration problems are not likely to be limited to only the provisioning systems.  I think there is an 80% chance that Level 3 is also experiencing problems in integrating other systems from their acquisitions, such as the billing systems, network management systems, accounting systems, and test labs.  So far, these other systems may not have impacted revenue to the same degree as the provisioning systems have, but I'm betting there is a lot of real work left to do on these systems, too, and that some of Level 3's resources are battling those issues as we speak.

Second, I want to point out an "inconvenient truth" about software development that was eloquently described by Fred Brooks in his classic book The Mythical Man-Month way back in the 1970s and still holds true today.  Brooks pointed out that you usually can't accelerate a software project by adding more people to it.  The classic adage is "nine women can't have a baby in one month".  So, it is really hard to accelerate software projects, other than by asking your engineers to work longer hours, something the engineers were probably already doing even prior to Level 3's announcement.  You might get a 20% acceleration that way, but it is probably already figured into Level 3's estimates. (The link to The Mythical Man-Month is to the left in my Previously Read books section of this blog).

Third, software development is not the only way to fix provisioning issues, because to some degree you can substitute manual processes for an automated system.  Here is where you can throw The Mythical Man-Month out the window and just brute force your effort to get orders installed by manually entering orders on many different systems, taking into account the chance for human error, and correcting those mistakes when they are discovered.  Of course, you can't use just anybody to do this kind of work...you need people who are trained on the systems they are working with.  To some degree, Level 3 is also engaged in this kind of activity, but they have already reported that they made mistakes in letting people go who were experts in some of the systems they needed to work with.

Fourth, any large integration project such as the one Level 3 is facing has a lot of different moving parts, different systems, and the changes needed can be broken down into a lot of different sub-projects.  Herein lies Level 3's best opportunity:  you can triage the sub-projects so that the highest priority is given to sub-projects that yield the greatest near-term benefit.  For example, Level 3 might identify one sub-project that allows 35% of the delayed orders to be entered successfully, more than any other sub-project, and would direct its engineers to work on this sub-project before any of the others.  Level 3 is clearly already engaged in this kind of triage effort.  This also points out that the big "Project Unity" (Level 3's name for the integrated system) is not really a binary project where you are either finished or you aren't.  Level 3 will be aiming to get incremental benefit from partial completion at each step along the way. 

Fifth, I believe that whenever a management team has to admit mistakes to the street, it has a big incentive to sandbag, or overestimate, their estimates for the time it takes to turn the situation around, and underestimate the revenue they are likely to achieve in the future.  I think Level 3 is no exception, and would expect that the 9 to 12 month schedule Level 3 shared with the street is longer than internally developed schedules.   Does that mean that Level 3 will have "finished" Unity in that time frame?  I don't think so, because I think that there will be Unity sub-projects that extend into 2009.  However, I think there is a 70% chance that Level 3 will beat the 9 to 12 month schedule for addressing the biggest pain points they are now experiencing.   In fact, some of my clients that are using Level 3 services have already reported some improvement in the ordering experience for some services, while other services remain horribly painful.

When I throw all of these factors into the pot, I come up with a hypothetical timeline that shows a 20% improvement in three months, an additional 35% improvement in six months, and an additional 15% improvement in nine months, meaning I am guessing Level 3 will see 70% of the operational benefit from Unity within 9 months.  The final 30% of the benefit will never be completely achieved but Level 3 will chip away at it for many years.

Of course, if Level 3 does more acquisitions in the meantime, all bets are off, and I'll need to re-calibrate!

Comments

[CK]

Ike, Many times I ask two pronged questions, which always lead to more questions, quite possibly, even more questions after that. I apologize, but sometimes it's the little boy inside of me always curious to know things which just might not be knowable. However, I am still interested in your guesstimate surrounding any out of the blue purchase before this year's end? Expanding upon that further, while tying it together with your horizontal disaggregation epiphany this morning, I wonder how deep into the communications radius' LVLT might go, at some point during their next acquisition, please!? Will they stop at the enterprise or go straight for the consumer next?

[Ike Elliott]

CK, I stand by my estimate that there will still be about 30% more integration work to do in 9 months to "fully integrate" the systems of the acquired companies. I'm guessing Crowe would explain away the difference between his estimate and mine by conveniently defining the scope of Project Unity to be smaller than the set of everything you might want to do to integrate the companies. The fact is, in any big integration effort, you get down to the end of the triage list and you start having diminishing returns for the effort you might put in, so in the end, you decide not to finish the job on purpose because you have bigger fish to fry.

[toddforthree]

i think you pretty well summed up what mgt. is telling the outside world. it will be brought up in pieces and the benefit will be gradual. nice blog.

[CK]

How would Jim Crowe reconcile his target and shareholder commitments to have "Project Unity" at one hundred percent "seamless" by 2008's end, compared to your own probability assessments-70 percent nine months into 08 and forever?

As importantly, how probable is it that, "Next Acquisition, please" does not occur by year's end inclusive of the current integration mess?

I cannot forget Jim's quote that, "Money is never a problem for LVLT."

You are doing a fine job!