From Value Stream Management to Feature Stream Enablement
- The tech and tools don't matter (Use the in-place toolchain...Jira, Jenkins & Github...)
- The process and methodology that work best for you (Agile, Scrum, Kanban, Lean, XP, Waterfall...)
- Teams can finally see work with full transparency (full-stream, full-stack, full-flow...all tied to KPIs and OKRs)
2. Internal focus…Lean efficiency External focus…Business agility
Manage & Measure
Execute & Deliver
Backlog Items
Frontlog Items
Feature
Team-
Tasks
Feature
Code-Tasks
What Enterprise Persona Groups Care About?
Risk & quality aware
Threat & impact sensitive
Standards & process focused
Control-point & quick reaction supportive
Value: activity records & efficiency data
Industry & market aware
Trend & transition sensitive
Customer & competition focused
Innovation & delivery speed supportive
Value: potential-growth & value-delivery data
ITIL
ITSM
CMMI
Kanban
Scrum
DevOps agile
Customer-centric
CAB
Market-trends
Competitive landscape
3. A Transition From Mega Projects - To Streamed Products - To Feature-Streams
To Feature!
Mega-BIG
Traditional
Waterfall…
…slow siloed
teams and
hand-offs
Small collaborative teams
& faster iterative deliveries
From Project
Read: Project To Product
To Product
Small and fast, connected
teams & feature-streams
From traditional
complex branching
To trunk-driven
simplification
Typically 1 year + Typically 90 days + Typically 14 days +
1X
4X
25X
Fewer bugs, quick learning, more innovation, faster time to value…
4. Business planning
& funding
What is an Enterprise Feature-Stream (Full-Stream & Full-Stack…Value Stream)?
Backlog
needed functions Bus. Value
Hypothesis
& Planning
with Dev.
Alignment
User Story
10. Product-to-customer data capture and continuous learning (i.e. painted door)
Business
Prioritized
Backlog
Trunk-driven
feature development
Process-defined
deployment events
Feature-aware
Ops Monitoring
Customer
Events
Aligned
to value
{
Velocity
& quality
{
Automation &
predictability
{ Risk mitigation
& fast-fix
{
Sustainability
& support
{
Business goals
& OKRs
Business planning
& funding
Data-driven Enterprise Feature-Streams (Full-Stream, Full-Stack…Value Streams)
Readiness-defined
feature release
Data-driven
decision-making
Understand
customer
sentiment
Y N ?
User-Story
Hypothesis
Validation
11. Let’s just jump in…
Where do we want to start
your feature-stream discussion?
Backlog
Items
Frontlo
g
Items
Feature
Team Tasks
Feature
Code Tasks
12. Who Are The Persona Group Leaders In Your Organization…And Will They Care?
No-Risk Release practices
Governance and/or IT
Governance and/or IT
Business and/or Dev
Business and/or Dev
IT and/or Governance
Dev & Bus. / Dev & Gov.
IT and/or Governance
Dev and/or Business
13. Accelerate Development & Increase software quality and security
De-Risk Deployments & Unleash multi-variant Feature innovations
Promote continuous learning and experimentation & Improve the business and customer experience
Who said, “You can’t have your cake and eat it too?”
Editor's Notes
Software-drive organizations tend to be driven to “make changes” based on 4 primary motives.
Internal and External motives… my east / west polarizing forces…
External motives are centered on markets, customers or competitors…and decisions tend to measure on the ROI (return on investment) - I call it AGILITY
Internal motives are centered on inside the 4 walls of an organization…including stands, security, audits, best-practices…with ROE (return on effort)... LEAN
Above and Below motives… my north / south polarizing forces…
Above the line motives are centralized decision that will benefit/impact the corporate effort, company strategy, whole org… with business outcomes or OKRs
Below the line motives are coming from teams, groups, segmented specialists that make “decisions” to help their efforts or achieve siloed - KPI goals
Ironically, HR rewards these polarized motives with independent and isolated reward systems that ultimately hurt the company in the long-run (Agility/Lean/KPIs/OKRs).
Value stream discussions help build bridges and blend efforts, that allow ALL 4 of the polarizing forces to be represented in defining healthy company goals.
Let me walk through the persona groups a bit more….
Above and Below “the line” is a demarcation line between those who manage strategies that impact software and those who work with the software process…
Backlog items are the “wish list of features” that need to be worked on…
Frontlog items are the “non-negotiables or standards” that help, guide or guard the software practices of a company…
Before and After “the line” is a demarcation line between those who focus on internal activities to measure success and those that focus on external factors to that directly or indirectly deliver success
Lean efficiencies - focus on smoothing out bumpy processes, accelerate slow and error prone activities and track tasks to fix quickly
Business agility - looks opportunities to improve their external market position or customer experience or gain a competitive advantage
Ultimately the business owners will need to take a modern approach to to their software and product strategies…
Please consider reading Project To Product, but realize there is a flaw…large organizations need to manage activity at the Feature level…
Initiatives align to Portfolios, Portfolios align to Products, and Products breakdown to Features
This is the most efficient way to drive teams, mange software and grow your business… Questions????
All this talk of independent decision making by agile teams and DevOps groups should not mean that they can choose their work priorities with utter disregard for the business priorities, corporate strategies and customer needs… and yet many do…and their MBO’s reward them for their busy independent working style…
Feature Flags have been around for almost 10 years… But some organizations are only now realizing they need to explore core development level changes (i.e. trunk-driven, test-driven, CI/CD, DevOps practices…) for their teams, toolchains and delivery process to increase velocity with scaled quality and fast-learning environments.
Changes at this level can increase the volume of outputs and the speed of change, but that is very different than increasing delivering Value or meeting Outcome goals.
Topics for another time Agile vs. Agility or Leaning your innovation or Dying with your DevOps-factories
Accelerate development
Encapsulate your code in an if/then/else statement
Core requirement for Continuous Delivery - Deploy Selectively and progressively
Safe launches - keep changes dark until functionality is complete and tested
Testing in production (small scale target users and closely monitored), supported with a Kill switch
Data-driven development for observability and feedback
Instrument the application
Collect data and analyze
Deploying code is very different then Releasing a Feature in production….
With the use of Feature Flags / Feature Toggles, they can allow the “deployed” Feature to sit in an “inactive” state until it gets the green light to go live.
That means all Dev work can be deployed and the velocity of their completed work can be sent down the line without delay and debate or impact to the scheduled deployment dates and events.
But it allows the Deploy/CAB/Ops/Security/Compliance/Quality teams to evaluate EACH feature and decide to deploy it as active or inactive of any number of reasons.
Collaborative Dev and Ops
Guardrails on the front-end and back-end engineering metrics
Granular user targeting
Phased rollouts
Seperate application deployment prctices from feature release decision-making
Strategically, the risk-mitigation of a feature needs time and work wo make sure it is beneficial and justifiable to Release…
Having Progressive Release design and the realtime Toggle function reduces most of the concerns of a big-bang release.
Fast-fix, scaled use, time limits, user limits…all can be factored in as a progressive release learns through the transitional processes.
Ensure Security and Compliance with Audit Trails
Splits, Segments, Metrics
Create
Update
Kill split
Reallocate split
Add/Remove Owners
Add/Remove Tags
Includes a diff to see changes
Monitor via Console
Extract for external analysis
Feature with Ops Monitoring designs can alert and provide engineering and product teams with a fast-response to adjust environments or opt for a lesser performance designs or hit the kill switch and inactivate the feature universally. Outages should easily be avoided once features become self-monitoring, provide early alerts and are designed with optional performance and security functionality, as a just in case toggle option.
Feature Experimentation
A/B/n Testing (Learning, validating, re-risking, trend monitoring…)
Hypothesis-driven development and product management
Data-driven stream, product and portfolio decision-making
Backend, User-centric experience optimization
The whole flow of a feature stream is listening and learning.
End-point experiments can also be run to validate user interest in future features.
This data can be required before Dev’s backlog items are posted.
This ensures that objective value data is considered in the business prioritization and that the feature work is justified with clear outcome goals.
ROI (External success)
ROE (Internal success)
Terms:
Splits - our term for feature flags and toggles
Treatments - our term for variations available for any given split.
Attributes - our term for key/value pairs which can be passed to get Treatment() for targeting.
Impressions - Events - MTKs
Experimentation is not just about the effects on user behaviors…
Guardrail metrics identify issues with releases
Measurement is available at any rollout percentage
Large sample sizes can reveal smaller changes
Feature Monitoring can alert you to degradations