So far, we have discussed some actual use cases that satisfy the goal of validating hypotheses. Here, we look at a use case in which we do not start with a predetermined hypothesis, but we keep an open mind and try to discover and understand what the data is telling us. In these types of cases, the project team is usually not sure of what to expect. Team members are truly open to discovering new information from the analysis. The content utilized in this project cannot be considered as social media content; however, the content is unstructured text and the techniques used and conclusions drawn are equally applicable to text captured via social media.
One of the divisions in IBM had initiated an award for the best-managed project in that division. A number of project managers submitted their projects for consideration of this award. A panel of judges evaluated the applications and designated one project as the best project of the year.
As part of this process, the project management profession leaders of that division wanted to analyze the project submissions to discover attributes of top projects based on the information submitted. What made one project stand out against others? What were the common attributes of projects that were perceived to be better than others?
As input to our analysis, we started with the content that was submitted by each of the 100 projects. The format of the input file contained the following
Name of Project:
Email Address of the Project Manager:
Name of the Manager:
Business Value: Free-form text
Business Value of Projects
To give you an idea of the content available for this project, we include here one random example of an entry for a project that we utilized in our analysis. This is basically the text that the person who submitted the application included in the Business Value field.
This Project implemented best practices in Scope Management, Planning, Monitoring including Status Reporting. In addition to using Fasttracking wherever possible, the PM created an Excel-based visual dashboard for displaying project scope up to the task level on each of the multiple components (Work Breakdown Structure), status of the task, and the date when it was expected to complete. It facilitated stakeholders and the PM to get a quick understanding of which tasks and which components were behind schedule.
This sped up monitoring as well as control and communication to stakeholders, and thereby identified action items that could control the deviations in the schedule. This saved quite a bit of time during reviews with stakeholders and enabled the team to focus more on resolving issues, minimizing risks, increased planning, and better decision making and implementation. The quantitative business value was the setting up of a new I/T infrastructure of 22 servers from scratch (including capital and hardware procurement) and deploying IBM’s business-critical CRM (Customer Relationship Management) application which produces reports that are utilized by executives and our CEO in just 10 weeks, which normally would have taken close to a year.
Analysis of the Information in the Business Value Field We then analyzed this information using a text analytics tool called IBM Content Analytics (now, Watson Text Analytics). IBM Content Analytics (ICA) is an advanced search and analytics platform that enables users to derive insight and create better decision-making models from any type of content (enterprise content, social media, social network analysis, plain text, and so on). IBM’s Content Analytics solutions allow users to understand the meaning and context of human language and rapidly process information to improve knowledge- driven search and surface new insights from that content. Content Analytics uses the same natural language processing (NLP) technologies as IBM Watson DeepQA, the world’s most advanced question-answering machine.
Once the model was defined in this tool, the tool was able to analyze all of the content and, based on the frequency and co-occurrence of the words, come up with themes suggested by the content. For example, there was a lot of discussion associated with projects being executed “on budget,” “under budget,” or with “cost savings.” Our analyst collected a list of such words and categorized them in our model under the attribute “Financial Planning” of a project. This process was continued, and nine attributes were identified across all the projects that were submitted for consideration. The following list shows the name of the attribute and the key words that we used to include a particular textual description under the given attribute of project performance:
■ Financial Planning — On budget, under budget, cost savings, accurate revenue-cost forecasting, financial management cost savings, decreased costs, no cost overruns, cost reduction for the client, under estimated cost
■ On-Time Delivery — On time, on schedule
■ Strong PM — Implemented strong PM practice, strong PM, PM capabilities, PM discipline, PM resolution, PM approach, WWPMM, The IBM Worldwide Project Management Methods Project MGMT Disciplines
■ Agile Methodology — Agile methodology, agile development framework
■ Control Management — Agile Scope Control, monitoring, tracking, tight control framework
■ Risk Management — Risk management plan, minimize the risk of failure, risk management, reduce risk
■ Defect Resolution — Eliminate constraints, address clients’ issues, fix incapability and dependency, minimize disruption, cutover, zero escalation, manage defect, reduce defect, fix framework
■ Milestone Adherence — Meet key deadlines, key project milestones, meeting of planned deliverables, deliverable management mobilizing resources in a timely fashion
■ Target Adherence — Project performance measured against the project plan on a weekly basis, systematic tracking of goal, target adherence, ongoing evaluation results, refine requirements at the beginning of each spiral We then calculated the percentage of projects that referenced these key attributes, as shown in Figure 7.4.
This analysis showed that the top three attributes of successful projects were strong project management discipline, on time delivery, and good processes for defect resolution. These results were fed back to the Project Management Education and Training program for that division to further reinforce the importance of these attributes in successful execution of projects.
Using Iterative Methods
Many times when we start out on a project, we have a clear vision, and with a little forethought, we establish a course of action that enables us to achi eve our desired results. Of course, as Robert Burns once said: “Sometimes the best-laid plans of mice and men oft go awry.” Mice, men, or even social media analytics strategy analysts it’s all the same! Sometimes the ideas we have, the questions we’ve posed, or the plan of attack that we’ve devised needs to be modified. We have to be flexible because sometimes we’re just plain wrong, or we’ve picked up a screwdriver to drive a nail. Things just aren’t right.
An iterative method of working, be it the creation of a model, the writing of software, or even of writing a book, is one in which we do not attempt to start with a full specification of requirements or plan. We don’t define steps 1 through 100 and then proceed in order creating our solution. Instead, we begin by specifying and implementing just part of our solution, which can then be reviewed and evaluated in order to identify further requirements or changes. This process is then repeated, producing a new version that is hopefully an improved version of the previous release. We continue cycling through iterations of our product, software, or solution until we believe we’ve achieved success.
As we work iteratively on a project, we create a rough draft or rough set of results in a single iteration. We then review it, decide on changes to (hopefully) improve it in next iteration, and continue until we’ve finished. Figure 7.5 visually describes the iterative method.
In the iterative model, we are building and improving the final analysis step-by-step. In this way, we can address deficiencies in our model or analysis early on in the process. This allows us to perhaps change our model, looking to additionally analyze social media data, or perhaps change the question in a way that is more relevant. The most important aspect is the ability to obtain feedback as we progress.
A feedback loop is something we use to gather feedback about what we’re doing, learn from the feedback, and then make changes based on that feedback. The sole purpose of a feedback loop is to improve a project based on the current plan of attack. These loops are important because they allow us a systemized approach to observing our results and learning from them.
Using this approach helps us to avoid those awkward moments that can happen when we present our analysis and the customers simply look on with a blank stare and comment, “That’s not what I asked for.” If they can see the progress (and perhaps help shape the final analysis), there are no surprises at the end.
Originally published on Blogger