If you are using Kanban to manage your agile projects, you might wonder how you can estimate how long it will take to deliver your work items. Kanban metrics can help you measure and improve your flow efficiency, cycle time, throughput, and delivery rate. In this article, you will learn how to use these metrics to predict delivery times and plan your work accordingly.
Kanban metrics are quantitative indicators that measure how your Kanban system is functioning. They can be used to monitor and improve your workflow, identify and remove bottlenecks, and optimize value delivery. Flow efficiency, cycle time, throughput, and delivery rate are some of the most common Kanban metrics. Flow efficiency looks at the ratio of active work time to total cycle time to show how much time is spent on value-adding activities versus waiting or being blocked. Cycle time reflects how quickly you can deliver value to customers by tracking the amount of time it takes for a work item to move from start to finish. Throughput measures the number of work items completed in a given period of time, indicating how productive your team is and how much work can be handled. Lastly, delivery rate is the percentage of work items that meet quality standards and customer expectations, demonstrating the reliability and consistency of your delivery process.
Kanban metrics provide valuable insights into workflow efficiency, enabling more accurate delivery time predictions. By analyzing metrics such as cycle time, lead time, and throughput, you gain a clearer understanding of your team's performance. Monitoring WIP (Work in Progress) limits and identifying bottlenecks helps streamline processes. Regularly tracking these metrics allows for data-driven forecasting, making it possible to predict delivery times more reliably and adapt processes for continuous improvement.
Any metric that can help you in finding out the interruptions in the flow of value, are potential kanban metrics. My favourite is knowing the Load in a flow. Load implies the amount of work in process i.e. the work has been pulled and somebody started working on it but they haven't finished it yet. Now, magnify this to a Kanban system where 100 people are contributing. The throughput is minimal but the load is getting bigger and bigger and bigger. Many things could be done to fix it. Likewise, the other metrics will let you know the efficiency of the flow (lead & cycle time), the kind of work that we are pulling through Load Distribution like features more and less enablers could end up affecting the performance of the system.
I think you might want to consider looking into pipeline analytics / DORA metrics instead. Predicting delivery times is highly dependent on number of teams and deploys. It also tends to be seasonal and can be a poor indication of application quality.
Use Kanban metrics to predict delivery times by analyzing cycle time and flow efficiency. Track the time it takes for a work item to move from start to finish (cycle time) and identify areas of inefficiency. Utilize cumulative flow diagrams to visualize the flow of work and identify bottlenecks. By understanding historical performance and addressing process improvements, teams can make more accurate predictions about delivery times in Kanban projects.
KANBAN is best suited for L1/L2/L3 teams. We can slice and dice the data to measure the velocity of the ticket resolution. Drilling down with proper analysis also helps to identify the bottlenecks and action to improve efficiency. The metrics data also helps in defining SLAa and measure compliance
Collecting Kanban metrics requires tracking and recording data of work items as they move through your Kanban board. Spreadsheets, online dashboards, and specialized software can be used to automate this process and visualize your metrics. Necessary data to collect includes the date and time when a work item enters and exits each stage of your workflow, its status (active, waiting, blocked, done, etc.), type and priority, assignee and owner, as well as its quality and customer satisfaction.
If you are using Jira I recommend Actionable Agile plugin to collect Kanban metrics. It also allows to make accurate predictions based on cycle time of tickets. Monte Karlo simulations empower forecasts you can use to answer the toughest question from the business: when will it be done?
Track the start and end dates of work items. This can be done manually using a spreadsheet or whiteboard, or automatically using a Kanban software tool. Count the number of work items in progress. This can be done manually or automatically. Use a cumulative flow diagram to track the flow of work through the system. This is a visual representation of the number of work items in each stage of the process over time. Use a control chart to track the variability of metrics over time. This can help to identify trends and potential problems.
Data of metrics are supported with most tools like Azure Devops, Jira, Trello ... you can know how many unit of work delivered per period of time (Throughputs) . How mostly long to complete a unit of work. (Cycle Time)
To collect Kanban metrics effectively: 1. Kanban Board: Use a board (physical or tools like Jira, Trello) for task tracking. 2. Define Stages: Label workflow stages (Backlog, In Progress, Review, Done). 3. Lead & Cycle Time: Record task addition/completion dates for Lead Time; note work start/end dates for Cycle Time. 4. WIP Monitoring: Observe active tasks, and enforce WIP limits. 5. Throughput Calculation: Count tasks completed in set timeframes. 6. CFD Generation: Automate with digital tools or manually chart task counts over time. 7. Metric Review: Regularly analyze metrics for trends and improvements. 8. Team Involvement: Ensure team understanding and engagement in tracking. This aids in workflow efficiency and areas for enhancement.
Once you have collected your Kanban metrics, you can use them to forecast how long it will take to complete your current and future work items. To do this, you should calculate your average cycle time and throughput for a recent time period (e.g., a week, a month, a quarter, etc.) and use these metrics to estimate how many work items you can complete in the same time period. Additionally, compare your estimated output with your actual demand and backlog size. To ensure balance between supply and demand, it is necessary to adjust your work in progress (WIP) limits, prioritization, and allocation. Lastly, it is important to monitor your flow efficiency, cycle time, throughput, and delivery rate regularly and update your predictions accordingly.
Kanban metrics can be used to predict delivery times by providing insights into the team's historical performance and current workflow. By tracking metrics such as lead time, cycle time, and throughput, teams can identify trends and patterns that can help them to estimate how long it will take to complete future work items.
In a recent software development project, we applied Kanban metrics to enhance delivery predictions. By analyzing the average cycle time and throughput over a month, we estimated our team's capacity for completing work items. Adjusting WIP limits and prioritization based on this data helped us maintain a balanced flow. Regular monitoring of flow efficiency, cycle time, throughput, and delivery rate allowed us to adapt to changing demands, ensuring accurate and efficient delivery predictions. This approach significantly improved our project planning and resource allocation.
I used suggest using throughputs to predict. it can be accurate from -15% to +15%. Make sure that your unit of work is small enough or similar size of work. If your work is more complex with big/small/unknow, the accurate will be low. 20-30% deviate
The above piece of article makes no sense to be honest. Average cycle time will not help, because forecasts based on averages are inaccurate by default. If you want to deeply understand why - read the book "when will it be done" by Dan Vacante. If you want to make accurate predictions, use scatterplot diagrams for single item forecasts and Monte Carlo simulations for multiple item forecasts.
Kanban metrics like Lead Time, Cycle Time, Throughput, WIP, and Cumulative Flow Diagrams (CFD) are crucial for predicting delivery times. We analyze historical data for baseline lead and cycle times, monitor current throughput and WIP to gauge capacity, and use CFDs to identify workflow trends and bottlenecks. We recognize task complexity and size variability and continually improve processes based on these insights. For more accurate predictions, provide delivery estimates with confidence intervals (e.g., 70% confidence in completing by X date) rather than fixed dates. Consistently tracking and understanding team capabilities are key to effective forecasting.
Predicting delivery times is not enough. To deliver more value faster and better, you need to improve your Kanban metrics and delivery times. Start by analyzing your Kanban metrics to identify sources of waste, delay, and variation in your workflow. Then, implement improvement actions to eliminate or reduce inefficiencies, such as removing blockers, resolving dependencies, reducing batch sizes, and simplifying processes. Experiment with different changes and measure their impact on your Kanban metrics and delivery times. Keep the changes that improve your performance and discard the ones that don't. Finally, repeat the cycle of analysis, improvement, and measurement continuously for optimal results.
To improve your Kanban metrics and delivery times, you need to analyze the workflow and implement actions to optimize it. Here are some insights. - Identify and eliminate sources of waste, delay, and variation in your workflow, such as blockers, dependencies, batch sizes, and complex processes. - Experiment with different changes and measure their impact on your Kanban metrics and delivery times. - Keep the ones that improve your performance and discard the ones that don't. - Repeat the cycle of analysis, improvement, and measurement continuously for optimal results.
The most common reason of failure to implement Kanban in my experience is lack of discipline. And Visualisation is one of core principles. The goal of a team leader should be to teach the team to frequently and regularly update Kanban boards to always keep them up-to-date. For example, there is no point in calculating cycle time of tickets on the Kanban board if half of the work team does is not reflected on the board. There is also no point to limit work in progress if your teammates don't respect these limits and move tickets going above the column limit just for their convenience.
Communicating your Kanban metrics and delivery times is essential to align your team, stakeholders, and customers on your progress and expectations. You should share these metrics and times regularly and transparently, using visual aids like charts, graphs, and reports to illustrate them. Explain the meaning and relevance of your Kanban metrics and delivery times in relation to your goals and value proposition. Additionally, solicit feedback and suggestions from your team, stakeholders, and customers on how to improve them. Finally, don't forget to celebrate your achievements and recognize your team's efforts.
In my software project, we communicated Kanban metrics and delivery times effectively. Regular transparent updates, accompanied by visual aids, such as burn-up charts and throughput graphs, kept the team, stakeholders, and customers well-informed. We emphasized the significance of these metrics in achieving project goals and delivering value. Actively seeking feedback fostered collaboration and continuous improvement. Celebrating milestones and acknowledging the team's contributions reinforced a positive and motivated work environment. This approach ensured everyone was aligned, informed, and engaged throughout the project.
Though forecasting, managing perceptions and expectations, and delivering value within reasonable timeframes are necessary, I don’t believe that communicating Kanban metrics outside of a team is in any way necessary, unless evaluating teams against each other in order to tune teams, evaluate issues, or rebalance work across a portfolio. If you’re running high-performing teams that are regularly, effectively, and efficiently delivering value, taking the time to communicate metrics outwardly is a waste of time. It’s when you’re in transition from one type of approach to another or you need buy-in from external parties to support, influence, or authorize effort that metrics may be relevant to prove value or identify issues.
I suggest to use ONE tool for Kanban metrics for working within the agile team and for reporting to stakeholders. This will ensure consistent of data, simplify stakeholder communication, and enhances efficiency (time and cost!)
Very good article. Kanban is however in my opinion a methodology for much experienced teams. Project managers/product manager need to be very agile and their backlogs need to be very well defined so they can actually make use of all the metrics and improve on the go. Collaboration of the team members is key for accomplishments in time with quality and value.
Lead Time Analysis - A software development team used Kanban's lead time metric to predict delivery times. By tracking the time taken from task initiation to completion, they could forecast future project timelines more accurately. This approach helped in setting realistic expectations with clients, improving client satisfaction and trust. Cumulative Flow Diagrams (CFD) - Another team utilised CFDs to visualise work in progress, completed tasks, and backlog. By analysing the flow and identifying bottlenecks, they could predict potential delays and adjust resources accordingly. This proactive approach allowed them to manage delivery times effectively, even in complex projects with multiple dependencies.