Better Sprint Estimates
What do you do when you consistently find yourself in a situation where your estimates don’t meet reality.
Another 2-week sprint ended without the team completing most of the tasks on the Sprint Backlog. Exasperated, the manager said, “This team isn’t good at finishing sprints.” The team needed to be better at meeting their estimates. They needed a plan, but first they needed to understand the problem.
Why this Matters?
Estimation is imperfect, and planning has a lot of uncertainty. It’s OK when Teams sometimes miss their targets. Never missing could mean that they aren’t being ambitious enough. Consistently missing by a lot can indicate a problem with your planning process. There is value in predictability; knowing what you can get done in a planning interval (Sprint, Iteration, or whatever you want to call it) is useful for helping business stakeholders make commitments to their stakeholders, and for the team to feel a sense of control and accomplishment.
Concepts and Words
This discussion will use some words that are mostly Scrum terminology, but they have analogous concepts in other incremental planning approaches
Sprint: A planning interval
Forecast: What the team thinks can be done in a Sprint.
Work Item: A “thing: you plan to do. This could be a User Story, or a task that’s part of a user story or other work.
Estimate: a value assigned to a work item to say how big it is. Teams use time, points, and even “count of items” or slicing, after normalizing all work items to be the same size.
Things worth noting:
The unit of estimation does not matter as much as the idea that the estimation process helps a team understand the work.
Estimates are often based on an “average” team member, so a newer/less skilled team member might get less done than someone who knows the code better. You want to avoid creating skill and knowledge silos by having people work on things outside of the core areas of expertise; these should average out, and also be considered when planning the work.
A common estimation approach is to use a variation of “Planning Poker,” asking each team member to provide an estimate, and (quickly) reconciling the differences.
Common Causes for Missed Goals
Since every team has different dynamics, it’s important not to make assumptions about causes and solutions without considering the possibilities. Three common reasons for missed goals are:
Errors in estimation for work items: something that is 3 units really is 5 to 7. This can happen for many reasons, ranging from a misunderstanding of the work to optimism
A misunderstanding of capacity: You may think a team of 5 has “20 units” of capacity, but you neglected to consider predictable things like vacations, routine meetings, and routine support overhead. The skill level of team members also matters—estimates should not assume a particular person is working on a work item, as people work at different rates.
Distractions: You might have had to support a critical production issue, or there might have been a number of small interruptions that took away from team time. In some cases, there are unavoidable and hard to plan for.
A Retrospective can reveal some problems and solutions more effectively than having a manager decide on issues and solutions.
Where to Start
A useful approach is to consider the factors in this order:
Capacity
Distraction
Estimation
Capacity
Look at this sprint and prior ones, and compare the forecast and completed work. Even if the estimates may not be accurate or precise, some patterns may emerge. A team consistently forecasts 20 points and completes 6-10, should probably start by saying that they will plan for 8-10 points of work.
If a sprint was an outlier or if there is no pattern from the last few sprints, the issue may be more subtle. Check in to see if any predictable events happened during this sprint that you forgot to identify during planning, and consider a scheme for taking predictable events into account, such as subtracting person-days for vacations, holidays, and team events.
You might have unexpected capacity gaps as well (illness, or family emergencies).
Distraction
If the current sprint is an outlier, distraction is a good place to look. Were there any unusual demands placed on the team? A production outage? More support requests? Was some key system not working reliably, causing some steps to take longer than expected? (And was that extra time unusual, or did we just not account for it?)
If your team is consistently missing its forecast, you can consider what “off the board” work is happening, and whether any of that should be accounted for in capacity planning. You can also identify typical “overhead” work, and for that, opportunities to streamline that work.
In a collaborative organization, some incidental work is to be expected, but if it gets beyond a manageable amount, you want to account for it in your planning, or identify ways to avoid it.
Estimation
Two of the more common ways estimates go wrong are:
Unclear work definition: The work item was more complex than it seemed at first read.
Misunderstanding of effort: The team agreed on what to do, but it took longer to implement.
The main value I’ve seen in estimation is checking in to see if the team has a shared understanding of the work, either the requirements or the approach. When team members give a wide range of estimates, it could be an indicator that the team should discuss what the story really is. Estimation will help clarify the work definition and identify a range of approaches that will satisfy the core goal of a work item.
It’s also possible that additional requirements issues will be discovered during the implementation of a work item. You don’t want to spend too much time on specification, especially when something may be simple. In this case, you generally want to see if any added complexity can be deferred to another story. If not, and this is a rare occurrence, you have identified a reason for the work being left undone, and you can discuss ways to minimize the chance of this happening without over-investing in requirement definition.
The Path Forward
Planning doesn’t create value, but it’s an important part of communicating with stakeholders and collaborators about what value you can deliver and when. There is little value in having perfect forecasts; spending more time planning to get a “clean Kanban Board” doesn’t add value, and occasional misses are expected. If you have prioritized the backlog and miss one thing at the end, nothing bad should happen. And if you manage to get more things done than you expected, no one should complain. But if you miss your goals consistently, it can have downstream effects on the parts of the business that were planning based on your forecasts.
While one way to meet your goals consistently is to plan for a conservative workload, that doesn’t serve business needs. Nor does being too optimistic or aggressive. By examining how well you meet your goals, you can develop plans that are predictive, with tolerable errors that help the business meet its goals.