Can this simple number become the key metric for project governance professionals?
Assurance and governance across project portfolios is a difficult art. It’s further complicated by inaccurate information or delayed reporting from individual projects. Here at Sharktower, our experience working across large change portfolios with financial services clients has informed building our project delivery platform. We’ve prioritised making real-time project data available to all who need it as a core principle of our software, and we’re now using advanced machine learning techniques to help you process and assess this information too. This is reflected in our flagship machine learning model, the Project Health Score.
Project Health Score is a deceptively simple concept, reporting a single, unified score between 1–100 to reflect the current status of a project. It’s a natural successor to things like RAG reporting in providing a high level view of a project which can be supplemented with more detail when needed.
The health score is calculated only using the information you enter into Sharktower. This means data such as how many stories and tasks were in progress, how many were delayed or blocked and who the team members were that created these tasks are all used as inputs to the machine learning model. The model uses this information to find hidden patterns that match its knowledge of previous projects with differing levels of success. Each project managed within Sharktower builds on this repository of information, and thereby progressively improves the predictions.
The project health score also offers a historical view, allowing you to assess trends over time on a week by week basis, mapped against key project milestones. You can compare with the portfolio average as well to understand performance in a peer group context.
For governance professionals, being able to track this score across a portfolio of projects gives you a quick view of where to focus time, either in complement to or as a replacement for individual project managers’ RAG reports. It could even become the one metric that matters — although as a composite of many different factors, it’s a much more balanced metric than a single number would usually be.
It will take a serious cultural shift to trust machine learning models as a key part of project governance though, and we know we’re only at the start of this journey. If you’re involved in project governance, assurance or delivery, please do get in touch and tell us your thoughts: are you excited about the possibilities machine learning offers, or do you think human experience and judgement will always be more valuable?