Dean Leffingwell Answers the Top 7 Questions About SAFe Value Streams

January 5, 2017 Lee Cunnigham

Do you know the difference between a value stream and an Agile Release Train (ART)? Do you know how DevOps fits into the Scaled Agile Framework® (SAFe®) value stream concept? Recently, we held a webinar with Dean Leffingwell to answer these top seven questions, and more. You can see the full webinar here, but for your convenience, we’ve provided a summary in this post:

1- What is a value stream?


“A value stream is the fundamental organizing construct in SAFe. It’s where we start.  Each value stream is a sequence of steps used to deliver value to the customer. There’s a trigger, something that causes it, and there’s some form of monetization or value delivered at the end. The steps in the middle are the development steps.

Value streams are typically cross-functional, cross-organizational, and cross-geographical. When we develop systems to support value streams or deliver value directly, we do that with an Agile Release Train (ART). The ART contains all the right people, and all the skills needed, to deliver some or all of a value stream.”

2- Why organize around value streams?

“The reason we organize around value streams is very simple. We want to accelerate time to market. We do that by optimizing the flow through the system as a whole.

Organizing Around SAFe Value Streams

If we’re organized that way, we have fewer hand offs, and we can work with smaller batch sizes, which go through the system more quickly. It is far easier to build in quality because everyone is working together. You get built-in alignment between business and development. Business is part of the value stream, development is part of the value stream, DevOps is part of the value stream. It’s just one flow.”

3- What is value stream mapping?

“Once we understand how value flows, we can apply an analysis tool called value stream mapping. Value stream mapping is an analytic process that asks: What are the steps? What’s the time between the steps? What can we do to minimize delays between steps?

SAFe Value Stream Mapping

When we conduct value stream mapping, we discover interesting things. We discover that the time it takes to implement a feature is a very small percentage of the total delivery time. By small, I mean typically between three to seven percent. So something that takes a month to build, might take a year to deploy. But then you can start to see what you can do about it.”

4- What is an Agile Release Train (ART)?









“An ART is a virtual organization that includes all of the people, with all of the requisite skills, to deliver value. After all, people do all the work, and we realize value with people in ARTs.

There are also ARTs that are physical organizations. What I mean by that, is that they are actual reporting structures. Even then, not everyone on the train necessarily reports into one individual. More generally, you start to see a more amorphous organizational model take shape, which is what we think of as the ‘buy a t-shirt mode.’

A person may report into operations and maintenance or the business, but we like to ‘buy them a t-shirt’ to welcome them to the ART. Everyone on the ART has the same t-shirt, so to speak; that’s the metaphor for being part of a larger team-of-teams. They’ll come to PI planning, and participate in the major events like the system demo, so logically they’re simply part of the ART, no matter who they report to.

We don’t typically suggest changing an enterprise’s organizational structure going in, because that’s a big deal and ARTs will tend to evolve over time anyway. With the virtual structure of the ART, all we need is permission to plan together, to commit together, to execute together, to demo together, and inspect and adapt together. It’s pretty hard to say no to that. To achieve this, ARTs run a sequence of events that we call iterations and Program Increments (PI) where everyone works together to deliver value.”

5- What does the customer have to do with the value stream?

Customers & SAFe

“The customer has just about everything to do with a value stream. We don’t have models in agile, or lean, or SAFe that don’t have the customer as integral to the development process. They are not developers, but they play a critical role. They’re central to providing input on ideation and development feedback. And of course, they’re involved in ultimate value delivery. There are no models where we get to say ‘Oh, my customer is remote and they’re not part of development, so we’ll just hope for the best when we deliver’. It may be true that they are remote and part of a different organization, but if we’re headed down a lean and agile path, the customer is absolutely an integral part of the value stream. They get the same t-shirt, and have specific responsibilities, just like everyone else.”

6- How do you organize ARTs in the value stream?

Organizing SAFe ARTs in Value Streams

“That’s the art and science of understanding value streams, agile at scale, and SAFe. ARTs are designed in such a way that each ART can deliver a set of capabilities, and in some cases, release independently of other ARTs.  For example, let’s think about a geophysical data collection satellite. That system doesn’t just have the obvious satellite with cameras, there’s also a ground station, and a web farm that feeds geophysical data to the end user. Technically, these are very different systems, but they have to work together to deliver the data. And their release cycles are not the same. You could easily consider the satellite as its own value stream, the ground station as a value stream, and the web farm is a value stream. Inside those, there could easily be multiple ARTs. You might think about that business as three large value streams and some number of ARTs in each. But it depends heavily on scope.”

7- What does DevOps have to do with the value stream?

SAFe & DevOps

“Value occurs only when the end user is operating the solution. The value stream can’t have a model that involves ideation and development, but excludes deployment. The DevOps pipeline is simply part of the value stream.

Take our example of the satellite. The company has terabytes and terabytes of data that they feed from the web farm to the end user. All of that data is obviously fed via deployed systems.

DevOps is a simply a capability of an ART. It’s just part of what ARTs do. If you disconnect the customer or DevOps, that’s not a value stream. That’s only a software development process. That isn’t taking a systems view.”


I hope these seven questions and answers from Dean Leffingwell help you better understand SAFe value streams. These seven questions are just a handful of the questions Dean answered about SAFe in our recent webinar. Check out the full webinar on-demand, How to Use SAFe to Deliver Value at Enterprise Scale.

Scaled Agile Framework and SAFe are registered trademarks of Scaled Agile, Inc. Figures 1,2,4,5,6 and 7 are reproduced here with permission from © 2011-2017 Scaled Agile, Inc.  All rights reserved.

The post Dean Leffingwell Answers the Top 7 Questions About SAFe Value Streams appeared first on VersionOne Blog.

About the Author

Lee Cunnigham

Lee Cunningham is the senior director, enterprise agile strategy at CollabNet VersionOne. He is focused on advising the executive sponsors and senior leaders of large enterprises, providing insight and practical guidance to meet the challenges of large-scale agile. He works with organizations around the world, and his experience, pragmatic approach, and customer focus keep him in demand as a trusted advisor.

More Content by Lee Cunnigham
Previous Article
Application Development Industry Analysts Weigh in on Greatest Gains Made in DevOps/ALM in 2016
Application Development Industry Analysts Weigh in on Greatest Gains Made in DevOps/ALM in 2016

Last year was a monumental one for CollabNet. Our customers, partners, and staff helped us to: Make signifi...

Next Article
Top 10 Agile & DevOps Blog Posts of 2016
Top 10 Agile & DevOps Blog Posts of 2016

To kick off 2017, we’ve rounded up the most popular agile & DevOps posts of 2016