Engineering leaders are often asked to determine what can be realistically delivered within the next month, quarter, etc. Capacity planning is ultimately a process about aligning people and projects, and helps organizations answer questions like:


• About how large are the work items we want to get done?
• How much work can each team take on?
• Should we add more engineers to our team, and what will we get if we do?

In practice, the process is fragmented and spreadsheet-driven, and teams struggle with setting realistic goals and avoiding overcommitment. Overly optimistic plans lead to downstream delays and the erosion of trust between engineering and other departments.

This work could be considered the sequel to Scenario Planner, which provided engineering leaders with a planning “calculator” to estimate project completion dates. Customers frequently requested the ability to plan for multiple deliverables at once. We determined that Scenario Planner would become overly complex if we tried to solve the problem there, and instead extended our scenario planning model to the capacity planning use case, which had already been identified as strategically important.

After an initial round of design, we validated our approach through concept testing with customers before moving into more detailed design. All testers expressed the desire to try it out and found the concept valuable despite a number of clear opportunities for improvement. Our key findings were:

• Overall, our concept resonated well and matched our users’ mental models for capacity planning
• It was useful to automatically represent carryover work, as long as it was easy to remove work items as needed
• Customers needed to select the granularity at which they did capacity planning. Our prototype showed Jira Epics, but some companies used initiatives or a custom work item for this type of planning.
• Customers wanted to plan over flexible timeframes, such as an upcoming month of an arbitrary number of weeks. Our testing prototype was very quarter-centric by comparison.

In collaboration with my team, I designed a capacity planning tool that enables engineering leaders to explore delivery tradeoffs for future periods. Building on our approach to scenario planning, Capacity Planner lets users create plans based on:

• Team size – number of developers on the team.
• Planned allocation – percentage focus on this bucket of work vs. others (e.g., bugs, tech debt).
• Planned work – work items for the time period, with automatic inclusion of ongoing work and manual work addition/estimation.

The tool provides a simple summary of the team’s planned capacity versus total capacity, allowing users to freely explore a variety of delivery scenarios and understand the impact of different options on their capacity. Capacity Planner estimates the amount of effort required for each work item based on historical velocity data.

Initial user feedback drove key enhancements, including support for vacation/ holiday time, breakdown of engineer skillsets (frontend, backend, and full-stack), and selection of specific developers for each work item.

Since GA release, Capacity Planner has seen increasing numbers of users week over week and has been used to create over 500 plans. Weekly active users spike during quarter transitions, confirming its role in quarterly planning processes.

Alongside Scenario Planner, Capacity Planner became a core part of an experimental “Delivery-only SKU” which was successfully sold to an enterprise customer, further proving the value of the Capacity Planning use case and of planning tools more generally. A dedicated Jellyfish product team continues to support and enhance the tool.

The capacity card on the top right is updated as inputs to team size, planned allocation, and planned work are modified. This layout matches the capacity “equation” of team size x focus = total capacity.

The initial state loads historical information about team size, work allocation, and projected completion dates as a starting point for the plan. This ensures planners don’t have the cognitive load of a “blank slate,” and also reflects the reality that planning never happens in a vacuum. There is almost always carryover work that is important to account for.

Skillset estimation helps users align people and projects. Users capture the skillset breakdown for the team, as well as a rough estimate of skillset requirements for each work item. These requirements are summed across the set of work items to allow comparison with the team's skills. Because capacity planning is typically a coarser style of planning, this “fuzzy” estimation felt more appropriate than item-by-item frontend/backend designations.

Plan settings include the ability to capture holidays and vacation time, which increases plan accuracy. This was one of the most frequent asks from early testers, as vacation and holidays can account for meaningful chunks over time at quarter scale.