Measure And Improve Software Delivery
Organizational — from 21 books.
Mike West, Editor · Leading With Data · engineeringleader.guide/guides/engineering-analytics
Arrow keys or swipe to move · N for notes · F for full screen
Measure And Improve Software Delivery
Organizational — from 21 books.
Mike West, Editor · Leading With Data · engineeringleader.guide/guides/engineering-analytics
The problem
Organizational and business erodes when it is left to instinct instead of method.
You were taught the moves piecemeal, never the whole model.
Measure And Improve Software Delivery · Mike West, Editor · Leading With Data
The model
- 01Continuous Delivery & Deployment Automation
- 02Loosely Coupled / Modular Architecture
- 03Software Delivery Performance
- 04Maintainability & Evolvability
- 05Testability & Test Strategy
- 06Shared Domain Understanding & Model Integrity
- 07Data & Consistency Integrity
- 08Developer Cognitive Load
How it grows
- New to the role0–1 yr
- Running the function1–3 yr
- Shaping the whole3+ yr
Measure And Improve Software Delivery · Mike West, Editor · Leading With Data
Step 1 of 5 · Continuous Delivery & Deployment Automation
Deploying on demand is the capability; how often you actually deploy is a business decision layered on top.
Misconception
Teams believe continuous delivery means deploying to production constantly, so they conclude it doesn't apply to their regulated or infrequently-shipping product.
Reality
The capability is about being able to deploy on demand at any time, not about deployment cadence; a team that ships quarterly but could ship in an hour still holds the capability and reaps its safety benefits.
Measure And Improve Software Delivery · Mike West, Editor · Leading With Data
Step 2 of 5 · Loosely Coupled / Modular Architecture
The test of loose coupling is independent deployability without cross-team coordination, not the number of services.
Misconception
Splitting a monolith into microservices automatically produces loose coupling and independent deployability.
Reality
Coupling is measured by whether you can change, test, and deploy a component without coordinating with other teams — a distributed monolith of chatty services is more tightly coupled than a well-modularized monolith.
Measure And Improve Software Delivery · Mike West, Editor · Leading With Data
Step 3 of 5 · Software Delivery Performance
The four metrics are a single instrument; report throughput and stability as a pair, always.
Misconception
Speed and stability are a trade-off, so pushing deployment frequency up must drive change fail rate up too.
Reality
High performers achieve both simultaneously; the same practices that make deployment fast — small batches, automation, fast feedback — also make it safe, so tempo and stability rise together rather than trading off.
Measure And Improve Software Delivery · Mike West, Editor · Leading With Data
Step 4 of 5 · Maintainability & Evolvability
Maintainability is the speed of all future change, which is why deferring it compounds cost.
Misconception
Maintainability is something you can defer and 'clean up later' once the product proves itself.
Reality
Maintainability is a rate, not a state — it governs how fast every future change goes, so deferring it doesn't postpone the cost, it compounds it into every subsequent commit.
Measure And Improve Software Delivery · Mike West, Editor · Leading With Data
Step 5 of 5 · Testability & Test Strategy
Fast, deterministic feedback beats raw coverage; a flaky suite is worse than a smaller reliable one.
Misconception
More tests and higher coverage always mean better quality and safer delivery.
Reality
The value is in fast, deterministic feedback and a healthy pyramid; a mass of slow end-to-end tests can lower confidence because their flakiness trains teams to ignore failures.
Measure And Improve Software Delivery · Mike West, Editor · Leading With Data
A framework to use
The 24 Capabilities Model for High Performance
This framework is part of the full guide. See the guide →
Measure And Improve Software Delivery · Mike West, Editor · Leading With Data
The cheat sheet
- Continuous Delivery & Deployment AutomationDeploying on demand is the capability; how often you actually deploy is a business decision layered on top.
- Loosely Coupled / Modular ArchitectureThe test of loose coupling is independent deployability without cross-team coordination, not the number of services.
- Software Delivery PerformanceThe four metrics are a single instrument; report throughput and stability as a pair, always.
- Maintainability & EvolvabilityMaintainability is the speed of all future change, which is why deferring it compounds cost.
- Testability & Test StrategyFast, deterministic feedback beats raw coverage; a flaky suite is worse than a smaller reliable one.
- Shared Domain Understanding & Model IntegrityThe domain model lives in code and speech, not in a diagram you file away.
- Organizational & Business OutcomesDelivery performance drives outcomes only when paired with value—speed alone can amplify waste.
- System Reliability & RobustnessLoose coupling contains failures; redundancy alone doesn't prevent cascades.
Measure And Improve Software Delivery · Mike West, Editor · Leading With Data
Sources
- Accelerate The Science of DevOps — Nicole Forsgren, Jez Humble, Gene Kim
- Architecture Patterns with Python — Harry J. W. Percival & Bob Gregory
- Data Mesh Delivering Data-Driven Value at Scale (Final Release) — Dehghani, Zhamak
- Data Analysis with LLMs — Immanuel Trummer
- Domain-Driven Design: Tackling Complexity in the Heart of Software — Eric Evans
- Effective Data Science Infrastructure — Ville Tuulos
- Excel VBA Programming For Dummies — Dick Kusleika
- Graph Databases in Action — Dave Bechberger & Josh Perryman
- Showstopper the breakneckrace to create Windows NT and the next generation at Microsoft — Zachary, G. Pascal
- Prompt Engineering for LLMs: The Art and Science of Building Large Language Model-Based Applications — John Berryman & Albert Ziegler
+ 11 more sources in the full guide
Measure And Improve Software Delivery · Mike West, Editor · Leading With Data