Introducing the Concept of Predictive Planning
Introducing the Concept of Predictive Planning
Do you remember the number of times you have had to manually check your Word document for spelling errors? How many times a day do you explicitly type a recipient’s email address when compiling an email? How often do you mentally calculate how many miles a journey is when planning a road trip? The answer, of course, is never. Today, we rely on computers to store and, more importantly, recall knowledge, eliminating the need for us to carry out repetitive and mundane tasks.
Why then do traditional project planning tools provide next to no guidance to a planner when building a project schedule? Suggestions regarding durations, guidance on whether the sequence of work is correct, warnings about erroneous building blocks such as open-ended tasks, etc. For the past twenty years, project planning tools have been reactive – waiting for you to enter data – instead of proactively offering suggestions and helping users input meaningful data. Project planning needs an overhaul, and I believe that should be in the form of what I am going to call “Predictive Planning Definition.”
Think about predictive planning and predictive text for a moment. Predictive text is called predictive for a reason – as you type, the computer is predicting what you want to type based on what you’ve already typed. Well, why can’t planning tools do the same? As I build out my plan and define activities, it shouldn’t be beyond the realms of reasonableness for the planning software to at least make suggestions. If they are good suggestions, then I’ll use them; if they are bad suggestions, I can tell the planning tool, and it should be able to learn from its mistakes in the same way we can train a spell checker to ignore certain words or use specific spellings.
For predictive planning to really be effective we need two key capabilities:
- Ability to capture and store sufficient historical data in order to establish trends or benchmarks
- Ability to search this data storage and extract relevant patterns and make accurate suggestions
Traditionally, data storage has been expensive and limited to highly structured data (e.g., modeled in a relational database or the likes of Excel). Today, data storage is not only cheap but a lot more flexible in terms of storing less structured data, e.g., natural language.
Planning organizations pride themselves in hiring the smartest, most savvy planners, yet do little to try to capture and retain that expertise and knowledge. It really shouldn’t be all that difficult to take historical as-built plans (that, of course, have that expertise embedded in them), or corporate standards and benchmarks, and develop a knowledge library.
If an organization can establish a knowledge library, and re-use the information in it, predictive planning is a no-brainer.
Making Sound Suggestions
For a computer or software tool to accurately predict an outcome (or in regard to predictive planning, make a suggestion to the planner), it firstly needs to understand context. In order to answer the question, “What is the weather going to be tomorrow?”, we need to know the location, time of year, and the weather trend for the past few days. With these data points, we can then use experience, based on historical data, to predict and model what is essentially a future outcome – i.e., tomorrow’s weather forecast.
In the case of building a project plan, knowing the size, scope, and type of project is a first step in giving the software some guidance as to where to focus its search when coming up with suggestions. When planning, you are predicting. You are trying to predict as accurately as possible a future outcome. This prediction is based on context and historical analogies.
Convergence of Technology Advancement and the Planning Discipline
In recent years, computing technologies have enabled the likes of neural networks and expert systems to return more accurate and sensible predictions. This approach today is coined under the already overused and ridiculously broad term: Artificial Intelligence. I believe our approach to planning should really be more along the lines of augmented intelligence. We shouldn’t kid ourselves into thinking we can totally replace the expertise of a planner. However, we should be bold enough to embrace the fact that computers are smart enough now to assist during the planning process, make smart suggestions, and give guidance.
Mistakes Make Us Smarter
So, if a planning tool were able to store knowledge and subsequently provide suggestions based on that knowledge during the planning process, how would it get smarter over time? Well, that’s easy. As the planning tool offers its suggestion, the planner either accepts or pushes back on the suggestion. This push-back or acceptance is how the tool learns – it can calibrate how often its suggestions are correct and adjust accordingly. The more interaction between the planner and the planning tool, the smarter the planning tool becomes.
Is This a Reality?
At InEight Basis, we have spent the past two years bringing this vision to reality. Today, when building a project plan, InEight Basis proactively makes suggestions on durations as you are building out your plan. It offers guidance on standard rates and costs based on the scope of work. It highlights commonly experienced risks, issues, and opportunities based on the type of project, location of the project, and even the contractor working on the project.
At the same time, ensuring a bit of a reality check, InEight Basis also drives plan realism through human consensus. Capturing team members’ suggestions, as well as the software’s suggestions, ensures you continue to leverage your team’s expertise.
Did I incorporate suggestions from the computer drawn from my historical knowledge? Did my team’s input and feedback make it into the plan? Were those inputs (both the computer and my team’s) valid? The answers tell me if I have built a validated plan that my team members and project stakeholders buy into. Measuring the level of consensus amongst my team is an indicator that the plan is achievable, and more achievable than a plan that has significant disparity/differences in opinion.
I believe we are in the early stages of a much-needed revamp and leap forward in the science of project planning and. Beyond the predictive planning and benchmarking capabilities that we now offer in the InEight Basis software, we are already working on additional augmented intelligence capabilities that will further enable a planner to focus on building an achievable plan. Working “on the plan” not “in the plan” is a much better use of a planner’s valuable time.