The goals of planning, scheduling, and tracking are universal. We need to understand what work needs to be completed, determine a good way to do it, then make sure it happens while making adjustments as new information emerges. However, in the last 30 years, we have started tackling more projects with higher degrees of uncertainty and change. These characteristics help us determine if we should use traditional, predictive approaches or rely more on newer, adaptive techniques.

When our projects undertake defined, repeatable work using technologies and approaches our organizations have experience in, then uncertainty and change rates are typically low and manageable. Here, traditional project management approaches work great. It is safe and effective to use Gantt charts, detailed work breakdown structures (WBS), network diagrams and earned value analysis.

Yet when projects use new (to us) technology and tackle problems our organizations have not solved before, then risk, uncertainty, and rates of change will be high. Traditional approaches have plenty of tools for handling risks, uncertainty, and change; but modern, adaptive approaches were purpose-built for these types of projects and have proven to be effective in these circumstances.

Work Characteristics, Not Industries
It is important that we examine work characteristics, not just the industry domain we are operating in. It would be easy (and wrong) to classify all construction projects as candidates for traditional approaches, and all IT projects as needing an agile approach. Instead, there are plenty of experimental construction projects using new designs, materials and assembly approaches. Likewise, there are defined, repeatable IT projects that can (and have been) successfully managed using meticulous planning, detailed estimation, and formal change control procedures.

So we need to dig deeper and see if we are dealing with low-variability tasks or more consensus gathering and problem-solving. These work types often change depending on which phase of a project we are working on. Designing something is typically a consensus-gathering and problem-solving exercise. Here, formal planning and estimation are difficult because we don’t know what we will encounter.

Consider the process of designing a new car or home. We have a combination of creative goals (produce something new and appealing) and engineering goals (meet specifications, constraints). The process is likely to be iterative and adaptive. We are looking to build consensus between stakeholders, who include sponsors (concerned with value, schedule, quality), designers (aesthetics, features) and engineers (performance, practicality).

This design phase requires collaboration between subject matter experts and probably iteration on prototypes to confirm understanding and validate ideas. Approaches like lean, kanban and agile work well in these uncertain, high-change environments. Their tools for experimentation, rapid feedback, reprioritization, and improvement help generate consensus on designs and drive uncertainty out of models.

Then, once the design is agreed upon, the process of production is typically more defined and repeatable. Unless our car or home is using new technology, materials or assembly techniques, the process of actually turning designs into completed examples should be less uncertain and iterative so more predictive approaches to work management can be used.

Physical projects—which manipulate tangible materials like concrete, steel, and plastic—have significant production phases where predictive approaches can be employed effectively. Digital projects—which manipulate intangible data and algorithms—have no production phase since the process of turning code into executable software (the process of compiling code) is automated.

So, software projects remain in the design phase—that early, upfront, uncertain period where subject matter experts are collaborating to create something that has likely not been done before.

There will be many trade-offs between design goals and implementation practicality to be made. All the divergent stakeholders will have divergent goals. The sponsors want fast, cheap and high quality; the users want their work simplified and streamlined so they can focus on their goal. The development team wants interesting work using new technology and skills to further their craft.

Once we understand what work types suit predictive and adaptive approaches, we can make better sense of our projects. Having said software development is design phase focused, it’s important to understand most IT projects do more than just software development. Tasks like ordering equipment, installing hardware and training users can all benefit from predictive planning and management techniques.

A Case Study
A couple of years ago, I worked on a project to develop and install routing software for truck drivers. This combined custom software development, integration with commercial off-the-shelf (COTS) software, and hardware installation that required wiring into the truck’s engine management system and installing GPS transmitters, tablets, etc.

The custom software development was easy to plan (but not easy to do). It was new, uncertain and benefitted from an agile approach. Integrating with the COTS software was a hybrid process. We worked with the vendor to iteratively tackle the highest risk and highest business value portions first. However, being just one customer of many, the vendor did not have the availability to serve our needs as quickly as we would have liked.

We worked on a monthly delivery cadence maintaining a backlog of issues and features to tackle next. Due to previous disputes about charges, the working relationship with the vendor was quite adversarial—so detailed statements of work and a formal change control process was followed. This consumed quite a bit of time for both parties, time that could have been focussed on feature development—but that reflects the reality of many commercial projects. We have to make the best use of what we have, given the current circumstances.

Installing the equipment in the trucks demanded precision timing, OCD levels of planning and copious communications. Semi-trucks cost hundreds of thousands of dollars and are carefully scheduled to make the most of their time. Bringing them in from a wide distribution area and having them out of commission while installs are performed (and drivers trained) is an expensive exercise.

When there are several hundred trucks spread across the United States and Canada, scheduling install teams and trucks to come to install/training centers becomes a variation on the traveling salesman math problem. Minimizing the total cost of lost trucking time, travel costs and staff time is a classic traditional planning problem.

In summary, like most real-world projects, the environment was complex and required using a variety of approaches. The custom software development was in-house and under our control. We used an agile approach with team-led sprints, demos, retrospectives, adaptation etc. The integration with the third-party package software was more of a hybrid approach. There were monthly deliveries based on a backlog we prioritized, but also statements of work and formal change requests.

Finally, the hardware installs and driver training was handled in a traditional, predictive way. Schedules for installs, equipment, and labor were planned and communicated well in advance. We did adjust these plans based on findings from early installs, but traditional, waterfall-style plans have always been amenable to minor adjustments. Software updates were delivered to the trucks over-the-wire as the trucks communicated back to base, so the agile teams could get new versions distributed once the equipment was installed.  

Making Informed Choices
Assessing uncertainty and consensus-gathering needs are important factors in determining the most appropriate approach to use. Thinking first about uncertainty, well-understood, often repeated work (such as building a new Costco store) represents much less uncertainty than rare endeavors such as building an underwater hotel:

If we add to this uncertainty view the dimension of approach focus, we arrive at a framework for assessing project approaches (shown below):

The “Approach Focus” Y-axis describes if techniques (approaches) are technical, such as creating a work breakdown structure (WBS); or people-focused tasks, such as team decision making or conflict resolution.

Using this framework of project uncertainty and approach focus, we can see that traditional, predictive approaches cover the bottom left quadrant of the graph well. They are great for work we are able to define and provide good process guidance:

Agile approaches tackle the problem space from the opposite corner. They are best suited for projects with high degrees of uncertainty and offer good people-based guidance:

There is a large overlap, too; it represents areas where we could use a traditional approach or an agile approach. Usually, it is recommended to use the approach the stakeholders will be most familiar with. So, if we are running with an agile team, a risk-adjusted backlog and risk burndown chart will likely be an easier sell than traditional risk management approaches. Likewise, if we are in a traditional, formal contracting environment, then statements of work and bills of materials will be accepted more readily than agile alternatives.

Summary
We can use the project environment to help determine which execution approach to use. Obviously, there will be organizational standards and guidelines to be aware of, too. However, even within traditional or agile guidelines, we can tailor our approaches based on uncertainty and task focus.

New project managers should understand that traditional project management has a wealth of process-oriented guidance for well-defined tasks. Likewise, agile offers much for uncertain, high-risk work that focuses on collaboration and people-based tasks.

We should also be aware that real projects are messy, complicated affairs. We often use a combination of approaches at macro and micro levels to try and be successful. It sounds complicated, but it forms the mastery of being a great project manager and is a journey worth pursuing.

[Note: I first wrote and published this article on ProjectManagement.com here]