Project Management Techniques
In the last few decades, project management has evolved and adapted itself to changing market needs. We will review the pros and cons of Waterfall, Scrum, Kanban, SAFe, CRISP-DM, and hybrid approaches.
Photo by Daria Nepriakhina on Unsplash
Project management approaches vary from predictive techniques on one end to empirical iterative practices on the other. Each approach has a valid use case. Therefore it is important to understand the concept behind each approach and its pros & cons. Using the correct approach to manage an initiative is key to successful delivery.
Waterfall
The Waterfall approach is a predictive model and the most common form of project management approach. As the name suggests the flow of activities from conceptualization to delivery follows a well-defined set of steps, with stage gates after each step to verify completion so that the project can move to the next step.
Use it when:
The outcome of the project can be defined in very certain actionable terms, well in advance. For example, upgrading infrastructure across the organization.
Skillset in the team is siloed. Meaning you have a team of specialist, like Business Analysts, Testers, Developers who cannot do each other’s job at, at least 50% efficiency.
Pros:
Since the end state of the project is defined well in advance, stakeholders have a clear understanding of the project deliverables including the end result.
Each step is well defined so the intermediate teams involved know exactly what to expect and when. This helps in planning, managing resource usage, and maintaining bandwidth.
Cons:
On a long duration project, the business needs can change during the implementation period. This model does not handle changes well, any substantial change can derail the project.
If projects in the pipeline are not sequenced properly, it will lead to a feast or famine situation. Certain members of the team will be overwhelmed with demand from multiple projects and then once the step is over, they will not have any work.
Agile
Agile approaches are iterative by nature. They are designed to be malleable and pivot quickly as business needs change. The degree of agility can vary from few weeks like Scrum to extreme programming, which advocates live code development.
Use it when:
If project details are uncertain, it’s best to use Scrum or Scaled Agile practices. With each iteration of these practices, the team can learn and decide on changing or pursuing the direction.
Team members are capable and willing to wear multiple hats as needed. Everyone on the team must be multi-skilled and should be ready to execute tasks necessary to meet the team objectives.
Pros:
Agile practices allow for frequent retrospection and are self-healing. The cost of changing direction is minimal.
Practices like Kanban, works very well with tasks like break-fix, with a very clear and precise backlog management.
Cons:
While agile practices can handle change management very well, it does poorly at managing time and budget. These practices cannot answer “How much will it cost to build x?”. At its best, we can come up with a range for both time and budget with a high degree of variability, making it difficult to get project approvals.
Team members and stakeholders with a non-agile mindset can make agile adoption a painful experience.
Data-Driven
Standard project management techniques Waterfall or Agile do not fit well with data-driven projects (AI, ML, Data Science). Standard practices are geared towards application development along with stakeholder participation. Whereas on data-driven projects, it is the learning from data analysis & ML models that have to be infused into the application, thus demanding a different approach.
CRISP-DM: Cross-Industry Standard Process for Data Mining
CRISP-DM is an industry-proven way to guide your data mining efforts.
As a methodology, it includes descriptions of the typical phases of a project, the tasks involved with each phase, and an explanation of the relationships between these tasks.
As a process model, CRISP-DM provides an overview of the data mining life cycle.
Use it when:
The project is primarily about analyzing data, building ML models, and eventual deployment & integration, to realize value. CRISP-DM (CRoss Industry Standard Process for Data Mining) methodology fits well data-related projects,
The data-driven effort can be part of a large project which is following a standard project delivery model. Don’t follow the same project delivery model for the data-driven portion of the project. It is perfectly fine to use multiple delivery approaches, based on the task at hand.
Pros:
This process is designed to iterate over the learn-model-train-test cycle multiple times before deployment.
In a data-driven project, ambiguity is a given, it will exists. The iterations help us learn from the ambiguity to derive a model that will predict with a certain degree of confidence.
Cons:
Since we are dealing with a high degree of ambiguity here, a considerable amount of time can be spent on the drawing board. You should plan for unexpected delays and/or level of confidence in the results.
Unreasonable expectations from stakeholders who are used to the traditional delivery model will build undue stress on the delivery team.
Hybrid Model - In Practice
In practice, you will use a model that fits your work environment and it might not be pure. Hybrid models are a reality and it works well. While you might not know everything about a project at the start, but knowing enough is wise. This helps manage: delivery expectations, scheduling, and cost.
After an overall understanding of the project, execution can be done in a phased manner using any one of the agile approaches. If any component of the project is data-driven, CRISP-DM is the best model to follow for that piece of work after the basic interface design is completed.
