How to Right-Size Your Stories for Better Predictability

Do your senior leaders want more predictability about when your team can finish its work? Perfect prediction is impossible, and sometimes, even reasonable prediction is quite difficult (with apologies to Yogi Berra). However, agile teams have one specific “tool” to create better predictability: right-sizing their stories. When a team right-sizes their work, they can create better predictions based on past experience. Even then, a team's prediction won't be perfect—but it will be better.

Let's start with what right-sizing is.

Define Right-Sizing

Right-sizing is a technique that says, “Most of the time, we will finish our work in approximately x number of days.”

I happen to like one- or two-day stories. When I say that, teams roll their eyes at me—until they learn to make product minimums and split stories.

However, many managers want teams to forecast work. I'm not big on relative estimation anymore. I much prefer cycle time. That's because I often discover that teams are good at estimating the work (what we used to call ideal time). However, due to delays and bottlenecks, teams are very bad at assessing the actual time, what we used to call wall time.

Forecasts demand “wall time,” not “ideal time.” That's how right-sizing can help you forecast work.

We all right-size in our lives and work. That's the percentage confidence you need to forecast future work. I'll first start with car drives. (I could have used a bus or train, but I tend to drive when I need to go somewhere.)

Example 1: Personal Right-Sizing: Drive into a Large City

I live near Boston, and most of the time (50%) it takes me 45 minutes to get to the Financial District to see a specific client. However, sometimes it takes me an hour, and the worst case was 75 minutes. Because I don't want to be late, I don't use the 50% confidence level of 45 minutes. I use the 80% confidence level of 60 minutes.

Notice I used a percentage confidence for my driving. That's the same idea for a relatively high-confidence forecast to predict when teams will finish and then start new work. You might use those forecasts for roadmaps. You can definitely use those forecasts for team backlogs.

But right-sizing works even better when we need to forecast product development.

Example 2: Team Understands Each Other and the Work

When teams know how to work together and they understand the work, they can measure their historical cycle time to forecast. Let's say that 50% of the time, they complete a story in three days. Since they measured their outliers, they know they can use six days for 80% confidence. Given their measurements, if they want a 90% confidence, they need to use twelve days as their cycle time.

In general, the longer your regular cycle times, the longer your outliers will be. The longer the cycle time, the less accurate your long-term planning can be. Insert any Yogi Berra quote about predicting the future. This is why I much prefer shorter time horizons and rolling wave planning. See the roadmap series.

The percentage confidence is what allows management to make decisions and act, given your forecast.

Example 2: New Team, New Work

What if a new team comes together on a brand new product? As I said in Predicting the Unpredictable, they need experience with each other and with the new work. How can they get that experience? Try this:

1. Ask each team member the relative cycle time they used before. What was the duration of their 50% confidence stories?
2. Listen to the answers. If most people used 3 days, that's how most people think about “typical” story size. If most people used 6 days, that's their “typical.” The team creates its right-sizing norm for now.
3. Ask the team to agree on a cycle time they can eyeball.
4. As the team creates stories, ask the team to split stories greater than their agreed-upon cycle time. Otherwise, leave the story alone.
5. The team now has their initial right-size.

As soon as the team starts to finish work, they measure their actual cycle time. As they finish more work, they will see if their cycle time increases, decreases, or stays the same. How long does a team need? I would say the longer the cycle time, the more time the team needs to continue to measure to know how to right-size.

This team can forecast a little after only three stories. If management wants a 90% forecast for quarters of work, they will either need to guess about their outlying cycle times, or they will need to finish more stories.

Your team might be somewhere between these two teams. Measure your historical cycle time, learn what the various confidence levels are for your cycle times, and then decide which confidence level you want to use for which purpose.

What if your cycle time is “too long?” You can still “right-size” but it might not look precisely like the definition.

Your Cycle Time is “Too Long” or Unpredictable to Forecast

You've measured your cycle time for the last ten stories. Your team is all over the map for cycle time. Sometimes, your cycle time is as low as three days, and sometimes it's four months. You can't possibly forecast with your cycle time.

Can you still right-size? Yes, and it's not going to look the same.

Instead of right-sizing future work, you right-size “effort” on this work. Create some working agreements around what the work is and how to approach it. Start by:

• Split existing stories into smaller stories. You might use the product minimums to decide what an MVE, MVP, MMF all are. You might ask, “Is this the smallest increment of value?” That will help you stop looking at compound “epics” or “themes” and start to right-size stories.
• Limit the team's WIP (Work in Progress). Because of Little's Law, we know the team can work on fewer items “at the same time,” as in reduce or eliminate multitasking.
• The team can collaborate, as pair, swarm, or a mob/ensemble, especially if the “smallest” increment of value is still very large.

The more your team collaborates, and reduces its WIP, the lower your cycle time will be. Also, the more predictable your cycle time will be.

Calculate Your Current Right Size

To start, measure your cycle time and create a time series chart. Sometimes, as teams work together, they naturally reduce their cycle time. Their “right-size” tends to be smaller. While I encourage teams to aim for one- or two-day stories, I am a realist. The longer your cycle time now, the more one- or two-day stories looks like a pipe dream.

But whatever your cycle time, measure it. Then, decide what percentage you need for a reasonable prediction: 50%? (Probably not.) You probably need a percentage closer to 80, 85, or 90%?

Determine how you can collaborate and reduce WIP to start reducing your cycle time overall. Then, if you want to continue to work that way, you might be done right-sizing.

Use your right-sizing first on your backlog and see how you do. Then, once you know you can right-size your backlog, move to roadmaps. You'll predict better and estimate less.

Like this: