In this quick video, Enterprise Coaches Colleen Johnson and Andy Cleff discuss…
- The right way to use flow metrics
- Difference between leading and lagging indicators
- 4 metrics + 1 bonus to help you measure flow at the team, system, and portfolio level
If you’d like to learn more about Flow Metrics, contact us..
Flow Metrics Video Transcript
Andy Cleff
Hello, I’m Andy Cleff and I’m joined today with Colleen Johnson. Both of us are transformation coaches with Agile Velocity.
We want to talk today about organizational agility and specifically, metrics that matter when optimizing for the flow of value through your system.
Colleen Johnson
Absolutely. We’re going to look here at how you can use metrics to help manage the madness. What that means for us is using and tracking metrics that are going to drive your team to some action, not just creating dashboards or creating reports or red light, yellow light, green light type of information.
We want metrics that drive action. So what can we do to leverage the data that we can collect to help us actually manage the madness?
Andy Cleff
I love that subtitle. So we’re going to introduce four core metrics plus one bonus metric that’ll help you measure flow at the team, system, or the org level.
Colleen, let’s ping-pong back and forth.
I’ll do the first one: WIP [Work in Progress] –the number of work items started but not finished.
This is a leading indicator and we’ll unpack the difference between leading and lagging in a moment.
Colleen Johnson
The next one’s throughput: the number of work items finished for a specific unit of time, You’re probably already tracking throughput in some fashion.
This often looks like how many items are completed for a sprint or for API. This is typically a lagging indicator because it’s telling us about something for work that’s already completed.
Andy Cleff
Number three, another leading indicator, how come I’m getting all the leading indicators?
Work-item age. It’s the amount of time that’s lapsed between when a work item started and our current date.
Colleen Johnson
Our last one for our flow measures is cycle time: the amount of a lapse time between when we started work on an item and when it finished.
So how long does it take for us to complete a user story or a feature or an epic. This is also a lagging indicator because it’s on work that is already completed.
Andy Cleff
So there you have the four key flow metrics: WIP, throughput, work-item age, and cycle time. We promised you a bonus. The bonus is one of my favorite things to look at and yes, you can measure the human element in your system.
We often talk about employee engagement. There’s a rather distressing poll that comes out of Gallup every year or every couple of years, and the recent one, I think it’s 2023 data, says about 23% of employees are actively engaged at work
So do the math.
There’s a huge chunk that are not satisfied, not willing to go that extra mile, not really passionate about what they’re doing.
I think that’s probably a whole other talk in itself, but they can be measured and you can roll that into your core metrics.
Colleen Johnson
I love that. When you think about that employee element too, think about the turnover that you mentioned or how frequently people are leaving the organization.
That may vary right? Based on the economic climate a little bit, but it’s such an important indicator of the culture and the larger ecosystem. How long do people stay?
Andy Cleff
Well, that’s lagging. By the time they’ve left, it’s a little bit late. The leading indicators are employee referrals and offer acceptance, of course. You can do feedback on a regular basis, one-on-ones, team retros and larger surveys.
And we’re in creative knowledge work so team learning opportunities, team learning logs, all these things are great.
Leading indicators give you an idea of what might happen next.
Lagging indicators–they’re the rearview mirror. I don’t know about you, Colleen but I like to look through the windshield.
Colleen Johnson
As we start to think about these, I mentioned the concept of what data can we make actionable. It’s not to say that those lagging indicators aren’t still actionable, but we tend to look at lagging indicators in things like retrospectives.
Those are the things that help us figure out process changes and policy changes.
They help us drive how we work and how we’re organized around our work. As we start to think about those leading indicators, those are the things that help us figure out what action to take immediately.
If you think of a signal, an alarm, that’s our signal that something needs to change right away.
What’s awesome is we start to break down these different types of measures and how we use them so that they work at every part of the business.
So this works at the team level, this works at the system level, this works at the portfolio level–whatever you would like to call those levels.
There’s a lot of flavors of scaling out there. So it could be stories, features, epics. It could be the team, system, portfolio.
You had an interesting one too, outcomes, opportunities, and experiments, right?
Whatever level that you’re breaking up your work and organizing your value around these metrics are still going to work.
Andy Cleff
The things that you need to get started are really simple.
You need a definition of workflow: You need to know when it starts and when it is complete. That’s it. Two data points for any of these work items in your system.
Notice we haven’t mentioned size, complexity, or effort.
It’s a paradox but it works.
So, starting and finishing… is helpful. If you’ve used the definition of ready and definition of done, they’re starting points, they can help you understand what it means to pull work items into and start and when it’s done so that you can actually finish your items.
Two points: When we started and when we finished it, that’s all you need to gather.
Colleen Johnson
Super fun fact about starting date and ending date. If you are unfortunately calculating this manually or with the spreadsheet, and not in a tool, you’ll want to add one day.
If you’re subtracting the end date from your start date, you always do a +1 because if I start something today and I finish it today, my cycle time on that particular item isn’t zero. It’s one day.
If you’re calculating it manually for cycle time, at least, always, remember to add 1 day.
When we start digging into what to do with all this data, we start to think about how to make this data actionable.
We’re going to focus on cycle time and throughput.
We can use start date and end date to tell us how long something took, but we can also start to use it to figure out our throughput.
So how many items are we getting done for a given period of time?
You can pick what that interval looks like. The easiest way to get started is to probably look at how many items we complete every day.
For a lot of teams, that’s not a whole number.
It might be like we get 0.4 stories done every day because it’s going to look at how many over time.
What does that throughput rate look like?
What’s great about throughput is we can start to use that to build what’s called a Monte Carlo simulation. As Andy said, we don’t need to estimate, we don’t need to look at complexity. We don’t need to look at hours or any of the typical ways that we go about trying to figure out when something’s going to get done. We can actually leverage our past throughput data to run a simulation to help us start to see when future work will be completed.
And you don’t need a ton of data for this.
You really only need about 10 completed work items and those completed work items could be stories, your completed work items could be features.
So like we said, we can apply this kind of simulation and ability to forecast to any type of work or level of work in our organization. That throughput and the ability to run the simulation against how many work items were completed is what is going to give us the ability to forecast when future work items are going to be done.
Andy Cleff
So this might be a leap of faith for some of you, but you don’t have to measure. You can keep measuring some of these things if you want to.
But we encourage you to overlay the systems.
Don’t stop, just add. You don’t need size, complexity, hours, et cetera.
And you don’t need hundreds or thousands of data points. You just need discrete work items, when it started, and when it finished and it can be at the team level.
If it’s part of a larger workflow, because there are handoffs, you can scale your measurements. What you get from this, if you take the leap and believe us for a moment, the Monte Carlo simulations give you the probability of a variety of outcomes.
If you’ve been in your financial service provider’s office, you know, you have an 85% chance that you’ll get this sort of range of return. It’s very similar for both. Cycle time is useful for a single work item.
You know, 85% confident that this epic, feature, or story will be done in two days, 10 days, 30 days, or less. If you want to look at multiple things, you use the throughput number. It’s the same thing, you get a range and a probability.
Why is this important? Why is this different?
Many of us are used to what we call a deterministic forecast: X on Y promise. That doesn’t work that way in these complex systems, teams or teams of teams.
My favorite metaphor is that, they [teams] are not wristwatches, they are gardens.
They are, they’re not dreamliners that you can take apart and put back together and they’ll always work.
Rainforests cause and effect are loosely correlated and yet you only know about it in retrospect.
If you want to look to the future, take a small number of data points, run some simulations and give yourself some probability and insights into what your system is capable of doing.
Colleen Johnson
I love that, Andy. This goes far beyond just trying to forecast when work is going to be done, whether it’s like you said, a single-item forecast or a multiple-item forecast. This also really helps us figure out where we can improve.
One of the key things we talk about when we’re looking at our metrics from a flow metrics perspective is how stable our system is.
I was in a conversation earlier today where somebody was asking me how we measure predictability.
I said, “I think you’re asking the wrong question.”
We want to ask how stable our system is because a stable system produces a more predictable outcome.
When we look at system stability, those are things like, are we keeping a consistent amount of work in progress over time. Do we have less variability of our cycle time and less variability of our throughput data so that we know that we’re working in an optimal operating capacity all of the time?
When those things start to come together, we get more predictable. It also helps us, as I mentioned, see those opportunities to improve those lagging indicators.
We talked about throughput and cycle time where we see something that’s way outside of what we expected that helps us, go back and say, do we need to change something?
Maybe we need to change some of our workflow policies or change how work is entering our system to make sure that we’re addressing some of that risk earlier in our process.
Andy Cleff
It shines a light on so many opportunities. One of the beautiful lenses that came to us via Dominica DeGrandis, was the “thieves of time”, when you start visualizing your cycle time, your throughput, your work item age, or your WIP, you will see some amazing things.
You will notice you have too much WIP, which is a vicious feedback cycle. You have too much. So you’re waiting. But what are you gonna do? You’re gonna go start something else. It’s endless.
When you see work-item age climbing up, chances are you have external dependencies in teams of teams or even within a team.
Other things that show up are unplanned work, conflicting priorities, neglected work, all these things are gonna rear their ugly heads.
Now you have some data and some feedback loops.
When you begin to try experiments, reduce your batch size, reduce your WIP, visualize your dependencies.
Even though these are lagging indicators, you will see soon enough in a couple of cycles, changes, and then you can amplify the good, share it across your system.
Colleen Johnson
We hope we gave you a few ideas on how you can use flow metrics and employee engagement to manage the madness.
If you’re interested in more content from us, check out our YouTube channel from all of the coaches at Agile Velocity for more tips and tricks on how to leverage flow metrics and other Agile Outcomes and Agile Practices to help better your organization.