SAFe Lean-Agile principles

1. Take an economic view

The entire chain of leadership, management and knowledge workers must understand the economic impact of the choices they’re making. Traditionally, the economic constraints on their activities are known only to the decision-makers and authorities who understand the business, marketplace, and customer finances. However, centralising such knowledge means that a worker’s everyday decisions are either made without this information or escalated to those who have it.


Deliver early and often

*Marshmallow Challenge

Deliver value incrementally

Value is higher early on

Understand trade-off parameters

  • Sequence jobs for maximum benefit
  • Do not consider money already spent
  • Make economic choices continuously
  • Empower local decision making
  • If you only quantify one thing, quantify the cost of delay


2. Apply systems thinking


Systems thinking takes a holistic approach to solution development, incorporating all aspects of a system and its environment into the design, development, deployment, and maintenance of the system itself.


3 Aspects of Systems Thinking

  • Optimising a component does not optimise the system
  • For the system to behave well as a system, a higher level of understanding of behaviour and architecture is required
  • The value of a system passes through its interconnections
  • A system can evolve no faster than its slowest integration point

Optimise the full value stream

  • Most problems will surface as delays
  • Most of the time spent getting to market is a result of these delays
  • Reducing delays is the fastest way to reduce time to market
  • Focus on the delays!


3. Assume variability; preserve options


Developers are naturally inclined to reduce variability, since it could lead to bad outcomes. However, the opposite could also be true. More so, the economics associated with the timing and type of variability is what determines the value of the outcome and a focus on eliminating variability too soon perpetuates a risk-avoidance culture where people feel they can't make mistakes and learn from experience what does/doesn't work.


Development occurs in an uncertain world

Apply a set based approach

Lean knowledge works understand that very little is actually known at the beginning of a project. Traditional design practices, however, tend to converge upon a single option and then modify it until it eventually meets the system intent.


Alternatively, a Set-Based Design (SBD) casts a wider net, considering multiple designs at the start, followed by continuously evaluating the economic  and technical trade-offs and eliminate the weaker options over time to ultimately converge on a final design, based on knowledge gained up to that point.



4. Build incrementally with fast, integrated learning cycles


Iterative learning cycle

  • Fast feedback accelerates knowledge
  • Improves learning efficiency by decreasing the time between action and effect
  • Reduces the cost of risk-taking by truncating unsuccessful paths quickly
  • Facilitated by small batch sizes
  • Requires increased investment in development environment

Apply fast learning cycles

  • Integration points accelerate learning
  • Development can proceed no faster than the slowest learning loop
  • Improvement comes through synchronisation of design loops and faster learning cycles


Integration points reduce risk

  • The more frequent the points, the faster the learning
  • Integration points create knowledge from uncertainty



5. Base milestones on objective evaluation of working systems


Problem: Phase gate milestones

  • Force too early design
  • Assume a point solution exists and can be built right first time
  • Create huge batches and long queues
  • Centralises requirements and design in program management

Problem: Phase gate milestones

  • Phase gates fix requirements and designs too early making adjustments costly and late as new facts emerge
  • Tracking progress via traditional milestones delays critical learning points until it's too late

Apply objective milestones

  • PI Demos are orchestrated to deliver objective progress, product and process metrics

Iterate to optimum solution

  • Objective milestones facilitate learning and allow for continuous, cost effective adjustments towards an optimum solution



6. Visualise and limit WIP, reduce batch sizes and manage queue lengths


Long queues are bad

Reduce queue lengths

  • Faster processing time decreases wait
  • Control wait times by controlling queue lengths

Visualise & limit work in process to control queues

  • BVIR - Big Visible Information Radiator

Importance of small batches

  • Small batches go through the system faster with lower variability
  • Large batches increase variability
  • High utilisation increase variability
  • Severe project slippage is the most likely result
  • Most import batch is the transport (handoff) batch
  • Proximity (col-location) enables small batch size
  • Good infrastructure enables small batches

Finding optimum batch size

  • Optimum batch size is an example of a U curve optimisation
  • Total costs are the sum of the holding costs and transaction costs
  • Higher holding costs shift batch size lower

Reduce Batch Size

  • Reducing transaction costs reduces total costs and shifts optimum batch size lower
  • Reducing batch size:
    • Increases predictability
    • Accelerates feedback
    • Reduces rework
    • Lowers cost
  • Batch size reduction probably saves twice what you think



7. Apply cadence, synchronise with cross domain planning


Cadence & Synchronisation

  • Cadence
    • Converts unpredictable events into predictable ones. Lowers cost. e*
    • Makes waiting times for new work predictable
    • Supports regular planning and cross-functional organisation
    • Limits batch sizes to a single interval
    • Controls injection of new work
    • Provides scheduled integration points
  • Synchronisation (PI Planning event)
    • Causes multiple events to happen at the same time
    • Facilitates cross-functional trade-offs
    • Provides routines dependency management
    • Supports full system and integration and assessment
  • Provides multiple feedback perspectives e*

Control variability with planning cadence

Synchronise with cross-domain planning

  • All stakeholders face-to-face
  • Management sets the mission, with minimum possible constraints
  • Requirements and design happen
  • Important stakeholder decisions are accelerated
  • Teams create and take responsibility for plans


8. Unlock the intrinsic motivation of knowledge workers


Drive: The puzzling puzzles of Harry Harlow




9. Decentralise decision making


Define the economic logic behind a decision; empower others to actually make them.


  • Infrequent - Not made very often and usually not urgent
    • e.g. internationalisation strategy
  • Long lasting - Once made, highly unlikely to change
    • e.g. common technology platform
  • Significant economies of scale - provides large and broad economic benefit
    • e.g. compensation strategy
Decentralise everything else

  • Frequent & common - routine, every day decisions
    • e.g. team & program backlog
  • Time critical - High cost of delay
    • e.g. point of release to customer
  • Require local information - Specific and local technology or customer context is required
    • e.g. feature criteria


The above Principles should be used (as a guide) in the Retrospectives

(idea: laminated copies of each of the principles available / on walls)