The Practical Application of Big Data for the PMO

As part of this month’s Ideaca blogging network challenge, we were tasked with discussing our thoughts on Big Data.

This is the second post of a 2 part post:

  • The first part covered how you, as a project manager, should approach a project that carries the mantle of “Big Data.”
  • The second part will cover how you, as someone in a Project/Program Management Office, can use Big Data without getting snookered by the hype.

Part 2 – I’m a PMO Manager. Why do I care about Big Data?

Some Assumptions

If your organization is large enough to have a PMO, you likely have more than 2 program managers with more than 10 project managers working for you on at least 40 projects per year.  You probably work in an organization that has a level 2 or 3 of maturity in Information Management practices.  You have likely had some train wreck projects that have helped influence the development of a project stage gating process.  You’ve been burned a couple of times, so you are relying on your Program Managers to keep a close watch on your Project Managers.

Your Data volumes likely aren’t “Big”, but the Unstructured Nature is “Big”

Not like Facebook big, or Amazon big, or even SCADA data volumes big.  But you are collecting both structured and unstructured data in the form of status reporting, budget monitoring (either through AFE draw-down reports, active budget management reports, or another formal reporting mechanism), and resource planning reports from your Program Managers.

People Lie, Data Doesn’t

In reading through the multitude of status reports that you are collecting, how can you really be sure that the budget indicator is Green without a correlation performed on the AFE draw-down report, and comparison with what was initially approved for the project?  Moreover, what if the text in the comments box says something to the effect of “we foresee having to wait for a decision” but the budget indicator is Green, not Yellow?

This sort of correlation doesn’t exactly take a rocket scientist to perform, but it does take effort to do.  Could you imagine doing that for 40+ projects each week?

However, if you blindly accept that “people lie, data doesn’t”, you may be setting yourself up for trouble.  If a project manager is able to hide something from their Program Manager, you can be that they will be able to game what is written in the status report to reflect a rosy picture.  You should still be talking to your project managers, program managers, and client sponsors as to how the projects are faring.

You need a little Intelligence with your Business

As a PMO manager, your number one priority is to not be surprised when a project (or project indicator) starts to trend down.  Being able to triangulate all of the standard reporting measures with the freetext comments entered by each Project Manager on their status reports, as well references to the initial project demand data, is challenging, but is worth the challenge if it reduces the time you have to spend reading status reports.

Think of it this way – if a project is being tracked by its ability to satisfy scope, schedule, and budget, shouldn’t a PMO be measured on the aggregate measures of Schedule Performance Index, Cost Performance Index, Budget at Project Initiation vs. Actuals at Project Closure, Resource Utilization, and Change Requests?

The ability to take all of this disparate, unstructured, and unrelated data, put it into a usable format, and prepare it for presentation back to the project management community within your organization would be of huge value for predictive analytics of project success and demonstrating the efficacy of your PMO.

How do you get started?

Before even thinking about the technology, think about the business case. If it costs you $1MM to implement a toolset, and it only provides you with $60K of value per year, the business case is just not there.

If the business case is there, though, as my Business Intelligence (BI) colleagues have stated, start with capturing the data (but don’t spend more than 20% of your effort doing so), focus on developing structure, and then begin the predictive analysis.


Are you a PMO manager that is currently struggling with projects that get by on best intentions?  How have you put some semblance of control in place?  Let me know in the comments below.




About Jason H Zalmanowitz
I am a nerdy Management Consultant / Project Manager with a MBA, have spent the majority of my career working for consulting firms in Calgary, and I race in triathlons because of (and thanks to) my wife. As a Project Manager, I have managed field implementations, strategy development projects, software development projects, and hardware implementation projects. As a consultant, I have helped companies articulate how and why they are going to implement and interact with sustaining technology to support their business.

7 Responses to The Practical Application of Big Data for the PMO

  1. Pingback: Project Management and Big Data – as a project | Unnatural Leadership

  2. Catharine Hortsing says:

    Jason, I noticed the link to your blog on Linkedin news. It’s a good read – I’ve passed it along on our intranet at Garner. Great to read what you’re thinking about. Warm regards, Catharine

    • jzalmanowitz says:

      Thanks Catharine! I’d be keen to know what you and the other smart folks at Garner think about using data for better managing a PMO. As VP of Major Projects, are you using any predictive analytics to help manage your portfolio?

      • Catharine Hortsing says:

        Thanks for your interest, Jason. One of the key things Garner does is help our clients better manage their capital project portfolios by capturing data and using predictive analytics. We use analytics in our own PMO as well. `I’ll ask Garner people to respond on your blog directly with any comments they have.

  3. One thing Garner has been doing for a long time is to track detailed timesheet and estimate data, at a granular level (as small as an hour for some tasks). It’s impossible to really know how long a piece of new software will take to create, but analyzing the data over time can give a feel for how long something will actually end up taking, given an estimate, that is better than the estimate itself. This can also be used in a feedback loop, to improve estimating, but one has to do it carefully, as an improvement in the quality of raw estimates can mess up the accuracy of the raw estimate to final estimate model. Think of it as Heisenberg’s uncertainty principle, as applied to software estimating.

    • jzalmanowitz says:

      Hi Mark, interesting approach – sort of a practical application of the Delphi Method (aka – I’ve done this before, and here’s what we saw). It’s always interesting to see how the different corporate cultures use data to drive performance or conformance. Similarly, it’s always interesting to see how different organizations manage by facts, or manage by feeling?

      Some questions that immediately come to mind to help understand the management by fact or feeling notion are:
      – Does Garner have a formal system to track these estimates to improve the quoting process?
      – Do you ever do formal lookbacks/post-implementation reviews/project closure lessons learned type sessions to see how accurate you were?
      – Are people incentivized to be more accurate with their estimates (i.e. year end bonus is multiplied by degrees of accuracy) or dis-incentivized to be less accurate (i.e. go with your hat in hand to ask for more money, which results in a negative perception of one’s estimation ability)?

      • We provide feedback to the team at various scales, either per iteration or on a more occasional basis for big project wrapups, and accuracy/consistency of estimates is a regular topic in our iteration restrospectives. However, we don’t have any specific incentive to encourage the developers to be more accurate: it has worked better for us to rely on their professional pride plus their understanding that other people are depending on the reliability of estimates for their own work.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: