3D Kanban
An XSCALE 3D Kanban.

3D Kanban

Despite the fuss that's been made of its Japanese name and Lean origins, a Kanban is just a card wall where you pull cards toward done from the right rather than pushing them into progress on the left. Pulling toward done minimises work in progress, which is good for ROI and related measures.

But what's a 3D Kanban? The obvious 3 Ds are Design, Delivery and DevOps laid out as if they were stacked on top of each other. Looks a bit complicated, especially if you're used to a 3-column Scrum board. Turning something simple into something complicated doesn't seem like an achievement ... unless there's another benefit?

Business Agility requires Design, Delivery and DevOps to work continuously, hand-in-glove. Self-directing portfolios of self-managing streams of self-organising teams. Relations between many different kanbans make this harder to do and time spent keeping all the boards lined up with each other is kind of painful. So in this respect a 3D Kanban could be simpler. But there's a bigger benefit ...

The Beauty of Breadth-First

On a 3D Kanban we only split work items into smaller pieces when we actually need the pieces. Boxing work up breadth-first rather than detailing long-ass product backlogs makes the context for prioritisation decisions explicit. You can't just walk up the nearest foothill in a highly dimensional design space and expect to discover Everest. Breadth-first targets the global maximum of ROI rather than just some local maximum.

Whether it's XSCALE XPM, Devops 3rd Way, Google's Design Sprints, or Lean's Set-Based Concurrent Engineering, breadth-first avoids exploring any part of the design space in depth until there's a sketchy view of all of it.

Here we're using XSCALE concepts to chunk from Products down to Stories. But you can use any breakdown structure you prefer so long as you do it just in time to track a card's flow. Because less cards on the board means less complexity and more YAGNI.

No alt text provided for this image

XSCALE's definitions of these things are minimalist:

  • Products are units of marketing and customer experience for some value streams.
  • Epics are business objectives relevant to bringing the Product to market.
  • Features are the smallest parts of an Epic with appreciable business value.
  • Scenarios are sets of BDD acceptance criteria for a Feature.
  • Stories are segmentations of Scenarios that share INVEST properties.

Of course you're free to bring your own distinctions to the cards on your own Kanban. The point here is representing the structure breadth-first. Think of the colours here as just a way to see how the boxes on the right are always pulled directly out of the boxes on the left. Let's follow the wall left to right and see how that works.

No alt text provided for this image

Walking The Wall

Walking the wall is a kind of standup meeting you use to propel your workflow. You can make your 3D kanban have a very different workflow to this one, which is based on the XPM, XAP and XBA patterns:

  • Canvas means boiling business constraints and metrics down to an Epic landscape.
  • Analyze means figuring out what releasable Feature-sets are relevant to each Epic.
  • Estimate means figuring out priority & budget per Feature to make a release plan..
  • Design/Test/Build/CI is the classic agile SDLC.
  • FIT is Feature Integration Testing in a Production-Like Environment.
  • SIT is Systems Integration Testing in Production using a Dark/Canary/A-B flow.
  • Science is where we deploy, experiment and gather data to inform the Canvas work.

So long as BDD tests are written in Functional style their analytics can be collected longitudinally to identify the bottleneck market constraint. Models combining these analytics into metrics can be combined with cumulative flow through a 3D Kanban to inform Throughput Diagrams similar to this:

See Also:

Exponential Business Agility, Corey Ladas. Business Bingo

Agustín Villena

Alineamos operacionalmente la cultura, los procesos y la tecnología con la estrategia, Acelerando la toma de decisiones de los equipos gerenciales y proyectos tecnológicos para un mundo cambiante

8y

Suspiciously similar to the kanban board proposed by Corey Ladas in his article http://leansoftwareengineering.com/2009/07/01/a-swimlane-for-ad-hoc-workflow/. A link to the original source would be appreciated...

  • No alternative text description for this image

I've had some questions about how we go about the estimation / prioritisation on a 3D Kanban and wrote it up at https://www.garudax.id/pulse/business-bingo-peter-merel

A great approach indeed, I am only saddened by two things, though: 1. The addition of terminology (though I can adapt...) 2. a physical board can't really serve the purpose, But perhaps the age of digital tools has come (2500 years after agility started ;)

Like
Reply

Harpreet Singh & Peter Merel, We visualize any & every actionable work originating from some Market Need. So we do StoryMapping like... MarketNeed to UserStory to StorySlice. These 3 levels help us to make a RoadMap, Release Plan & see our progress in a holistic way, w/o getting lost in truckloads of tasks. Secondly, this visualization helps us to talk to the market facing teams (PO, Sales etc) in the language of "what impact is created in last month" rather than "how many story-points work is done in last month" As far as display wall is concerned, we use an electronic board, mounted in a corridor, which keeps flicking the Sprints-wall & Release-wall. Our idea to mount the electronic wall in the corridor was to Radiate Information, bring in Transparency and have Single Version of Truth. The wall for StorySlices is heavily used in StandUp meetings. We built our own electric board called islandBRIDGE and we have been using it for our 2 products. Just in case if you want to know, islandBRIDGE is for bridging the islands (silos) in software engineering.

To view or add a comment, sign in

More articles by Peter Merel

  • Claude & Gemini Say Apple Can't Reason

    Apple's "Illusion of Intelligence" claim that AI can't reason is, according to Claude Opus 4 and Gemini 2.5 Pro…

  • Superhuman Prompting

    It's well known that prompting formulas dramatically improve the quality of LLM outputs. In many cases a well-prompted…

    8 Comments
  • #AI and the #ClimateCrisis

    AI used recreationally is no more likely to solve the climate crisis than a hammer cracking walnuts will build a house.…

  • Prompting AI-Driven Development ... in Anger

    TL;DR: AI codes quicker, better, and cheaper than humans, but also does things almost but not quite entirely unlike…

  • Merel's Wager & Test-First AGI

    Sam Altman says it's only a model's behavior that we can check, not its source code. Even if AI model weights were…

  • AI–Driven Development

    AI writes software faster, cheaper, and better than humans, but it also hallucinates and misinterprets us. Agile teams…

    11 Comments
  • AI & Agile Alignment: How-To

    The Intelligence Revolution We've been paying people to automate work we used to pay people to do since agriculture…

  • AI & Agile Alignment 101

    There is no agility without alignment; anyone with back pain will tell you that. The Agile movement began in the 90s as…

    8 Comments
  • Camelot 2.0

    TL;DR: A simpler method of Autonomy in Alignment. This version of Camelot is much easier to explain and to do across…

  • Autonomy in Alignment

    Autonomy Without Alignment In a change of direction, SAFe's Dean Leffingwell quotes Jim Collins: Autonomy without…

    11 Comments

Others also viewed

Explore content categories