Metrics and Prediction: When Will It Get Done?

22 Mar 2022

When registrations for an online event hit multiples of hundreds, you know it’s a hot topic! Jahanger Hussain (Jags), Lean Agile Trainer at Radtac, addressed many screens of virtual attendees when he spoke on this topic on 24th February.

‘When will it get done?’ is possibly the most critical question for stakeholders and customers across all industries and disciplines. Jags approached it from a software development perspective, but the challenge is the same for teams, leaders, HR, procurement, marketing, sales – when will the work be completed? 
 
As Jags pointed out, answering this appears at first to involve a bit of crystal ball gazing. As human beings, we can’t predict the future and when we try we’re often far off the mark. History shows we are traditionally poor at forecasting; our cognitive biases are just too strong.  

In planning poker, agilists estimate according to effort and complexity. Using relative sizing, we compare work items and assign storypoints. Whilst Jags acknowledged the value of this, he highlighted this doesn’t consider the ‘other things going on’, such as completing tasks, context switching around work in progress, bottlenecks, blockers and new work demands.  

In some contexts, we need greater predictability – which Jags defined as ‘knowing in advance what to expect’. 

One of the many benefits of using digital workflow visualisation boards is that they collect data from which we start to predict with greater confidence. In today’s world, where we accept ever-increasing levels of change and uncertainty, we don’t want to dismiss this opportunity too quickly. Making the most of the data we have, helps reduce the risk for the business and increases learning for the team.  

Enter technology! 

The digital workflow boards used by most Agile teams can process data from the past to predict what’s likely to happen in the future. For Jags, you don’t need a massive amount of data points to achieve a reasonably high level of accuracy. He suggested just 20 may be sufficient. Measures that are relatively easy to capture would include: 

Cycle Time – the time taken to complete a task from start to finish 

Throughput – the number of tasks completed in a period of time 

Work Item Age – the elapsed time since the task was started  

From this data we can calculate the probability of a task being finished in a time frame. Where the risk of work items not being completed is high, then the team will know they need to swarm to ensure delivery is on time. Data helps us to understand the patterns within our workflow, so the team can get an indication of where to focus their effort.  

Jags ended with an exploration of Monte Carlo Analysis, a mathematical technique that enables data-driven decision making. Monte Carlo forecasts are created by running many random simulations of historical data to project the probable outcome.  

By embracing data analysis, and making best use of what we have, ‘When will it be done?’ is a question that can be answered with reasonable accuracy.  

Jahanger Hussain (Jags) is Lean Agile Trainer at cprime Radtac, an Agile Business Consortium Accredited Delivery Partner 

Do you want to go watch the event in full?

Head over to https://www.agilebusiness.org/page/resource-video-bitesize-kanban-metrics


Please note blogs reflect the opinions of their authors and do not necessarily reflect the recommendations or guidance of the Agile Business Consortium.