The right conversation, a little more action — Forecasting using team data
The right level of conversation, especially during a pandemic, is vital. Keeping in touch with your team is key but overdoing it can be exhausting for all involved. For the Agile Delivery team, we want to let data do the talking; saving energy to deliver value and having conversations that matter! We aim to find efficient, data driven and value adding methods for planning work and delivering features.
It’s great to see the many different ways that ITV Technology teams work together to deliver value. Demand is always high but so is the urge for trying new things and experimenting. We encourage teams to be flexible and find a way of working that suits them. It’s important to protect culture and the solid foundations built.
The Agile Delivery team identified that using throughput provides a flexible, more data driven way to forecast features and new work. Throughput is defined as the amount of items passing through a system or a process. We want to use an interpretation of forecasting that maintains and improves our good culture whilst enabling continuous improvement. Our first step for forecasting was to record the amount of tickets (stories or tasks) the team had completed within a two week period. We could then split out tickets into the classification of issues listed below:
- Items of Value: new capabilities that can be used and benefit the user, no matter the size. Also, any performance enhancements to our code/services that we consider will indirectly benefit the end user. This is included in throughput.
- Bugs and BAU: unexpected and unintended behaviour that needs correcting. Also, any ongoing maintenance. This is excluded from throughput as it’s not additional value, although still crucial to our services.
The measurement of valuable tasks completed from previous weeks allows a forecast of when new features are likely to be completed. You can see from the chart above how many tickets are completed on average. This allows you to count future tickets and forecast a most likely completion date. The forecast isn’t meant as a hard due date as things can change, for example; scope, team capacity and other factors.
Some of the benefits we have seen from using our interpretation of throughput to forecast delivery are:
- More time spent on delivery: forecasting takes a limited amount of effort from the development team allowing them to focus on the task at hand. The development team can plan in tickets as normal and the work can be forecasted from that planning session.
- Flexibility and Speed: throughput forecasting caters for any change in requirements and new tickets being completed and adding on to the scope. Developers can adapt quickly to this. This means work can be planned in a more ad-hoc way rather than having a big planning session upfront.
- Improving accuracy: using this data driven way of forecasting, averages will grow more accurate over time allowing for better probability of when things will be delivered.
- Improving team performance (perhaps the most important): visualising throughput over time can lead to insight into team trends. This can help the team reflect and adapt ways of working and see what worked at what time.
Our interpretation of throughput forecasting has been used already on implementing sliders on the ITV Hub Homepage for ‘Recommendations for you’ and ‘Continue Watching’.
In the Agile Delivery team, we are looking to refine this process and improve even further. We understand that throughput forecasts should be based on ‘productivity’ and not ‘activity’ meaning we should always be reviewing what is seen as items of value. It is also important to communicate to stakeholders why we use this method of forecasting, rather than more traditional methods of estimating such as story points.
Overall, throughput forecasts help us maintain our high performing culture, while delivering value based on the business needs. Limited effort is required for forecasting from development teams allowing focus on delivering value sooner. Conversations are still essential to discuss how value can be delivered but we can save time talking about when, by letting the data do the talking. We are working with teams and stakeholders to build a mutual understanding of team stats and how this can benefit the department. So far we’ve found that using throughput has saved us lots of time in meetings, rather than estimating every task. We aim to iteratively improve our methods of throughput forecasting so we capture true productivity.
Interested in what you’ve read about — we’re hiring in Direct to Consumer Technology: https://itv.taleo.net/careersection/2/moresearch.ftl?lang=en&portal=101430233
