(Nicole Forsgren, Abi Noda; Shift Key Press Nov 2025; ISBN 978-1662966378)

(As a general rule, this book is following the social media style of authorship, at a slightly larger scale: Instead of one sentence paragraphs, they're doing one- or two-page chapters. Overall, I'm not entirely impressed so far, particularly since I think a good chunk of the "seven steps" are just a retelling of Kotter's Change Management steps. Actually, as I go through the material, I begin to suspect a good chunk of it was written by an LLM. The case studies and examples probably aren't, but a lot of the prose in between feels very LLM-ish. Stuff in Step 3 feels more genuine and additive.)

PART I Understanding DevEx

Chapter 1 - What is Developer Experience?

Developer experience isn’t just about making developers happy--it’s about removing friction that slows down your entire business.

Chapter 2 - Why Developer Experience Matters

Poor developer experience isn’t just about productivity--it can hide catastrophic business risks in plain sight.

One morning in 2012, a developer at Knight Capital pushed a routine update. Nothing seemed off--until the system started hemorrhaging money. In just 45 minutes, a single change triggered $460 million in losses. The culprit? An old feature flag, long retired, accidentally reactivated by a deployment script. But the real problem wasn’t the bug--it was a development environment that made problems impossible to detect and fix quickly, turning a routine mistake into a company-ending disaster. (Dolfing, Henrico. “Case Study 4: The $440 Million Software Error at Knight Capital.” 5 June 2019, www.henricodolfing.com/2019/06/project-failure-case-study-knight-capital.html).

Fast-forward to July 2025: Jason Lemkin was using Replit’s AI coding assistant to build a database for his company SaaStr.AI. He explicitly set a “code freeze”—no changes to live data. But the AI ignored his instructions and deleted his entire production database. (https://www.pcmag.com/news/vibe-coding-fiasco-replite-ai-agent-goes-rogue-deletes-company-database)

Chapter 3 - Friction: The Value Killer

Friction can slow any progress, especially in software, which is why improving DevEx is so important.

Friction is anything that slows down or gets in the way of completing work:

Eliminating these friction points creates immediate business value.

“Productivity” sets off alarm bells--”Experience” opens conversations.

PART II The Three Essential Elements of DevEx

Chapter 4 - The DevEx Framework

Feedback loops, flow state, and cognitive load, each of which is examined further in its own chapter.

Chapter 5 - Feedback Loops

See OODA.

Chapter 6 - Flow State

See Csikszentmihalyi.

Chapter 7 - Cognitive Load

"Excessive cognitive load directly impedes developers’ ability to deliver value to customers."

"The solution is systematic reduction of unnecessary complexity."

"AI presents both opportunities and risks for cognitive load."

Chapter 8 - Use the Framework to Identify and Remove Friction

"The three elements of DevEx provide a framework for spotting and fixing friction."

PART III Making the Business Case

Chapter 9 - Translate Developer Experience into Business Value

"C-suite executives speak the language of business outcomes."

"Talk about recovering time: Convert developer hours into dollar value."

"Talk about saving money: Quantify cost savings."

"Talk about making money: Accelerating revenue through DevEx."

"Capital One's DevEx transformation" citation here.

"Talk about proven correlations: Link technical metrics to business outcomes." ... "A specific example of this approach is the developer experience index (DXI), developed by DX. The DXI score is the average of 14 survey questions that ask developers about different aspects of their development process. By correlating this score against self-reported time loss, DX has found that each one-point gain in DXI score translates to saving 13 minutes per week per developer, equivalent to 10 hours annually. (Noda, Abi, and Laura Tacho. “The One Number You Need to Increase Roi per Engineer.” DX, 2025, getdx.com/research/the-one-number-you-need-to-increase-roi-per-engineer/.) This correlation allows organizations to predict financial impact from DevEx improvements."

Chapter 10 - Connect Your Work to What Others Care About

"To gain leadership support, demonstrate how developer experience directly solves their specific business problems."

"Align your DevEx work with the problems that keep your CEO and CTO up at night."

"Tie it to your company’s biggest priority."

"Give senior leaders something to brag about."

"Reveal performance gaps."

"Always reframe developer experience from a technical nice-to-have into a strategic business enabler."

PART IV Improving DevEx: A 7-Step Process

STEP 1 Start Your DevEx Journey

Reminder: Kotter's 8 steps:

  1. Establishing a sense of urgency
  2. Forming a powerful guiding coalition
  3. Creating a vision
  4. Communicating the vision
  5. Empowering others to act on the vision
  6. Planning for and creating short-term wins
  7. Consolidating improvements and producing still more change
  8. Mantaining new approaches

Chapter 11 - Interview Developers

"To get good information, you’ll need to interview a diverse set of developers." 12-15. Consider factors such as years of experience, product area, type of development (e.g., legacy, mobile, cloud, data science). Make two planning lists: 1. Demographics that matter. For example, work location (onsite, remote, hybrid) and experience level (junior or senior devs). 2. Work types and organizational areas where devs work.

"Let developers know that talking to you is a good use of their time."

"Building strong developer experience starts with earning developers’ trust."

"Most importantly, honor developers’ reality."

"During the interview, let them know you are there to listen and learn."

"You’ll know you’ve done enough interviews when you start hearing the same problems repeatedly—this is called saturation. It usually happens at around 12-15 interviews, but you might need fewer if clear patterns emerge earlier (generally no less than 7), or more if you’re consistently getting new information."

Chapter 12 - Synthesize What You Learn

"Start identifying labels while you’re still in interview mode."

"Build a flexible framework that evolves with your understanding."

"Capture these themes and supporting evidence in a simple document."

"Take a second pass through your notes."

Chapter 13 - Visualize Your Developers’ Workflow, Tools, and Friction

"List the steps your developers have to do to write and ship code."

"List everything, even if it seems trivial—you never know where friction might be lurking."

"Don’t forget specialized developer workflows."

"Consider turning your lists into workflow diagrams."

Chapter 14 - Understand Your Stakeholders

"You might be surprised just how many stakeholders you have."

Start by listing all potential stakeholder groups. These may include:

Chapter 15 - Share What You’ve Learned

"Once you’ve synthesized what you learned, be sure to share your findings."

"Once your summaries are shared, be prepared for continued dialogue."

STEP 2 Start Small and Get a Quick Win

Chapter 16 - Pick the Right Projects

"When kicking off a DevEx improvement initiative, you need visible wins that validate your approach and transform skeptics into advocates."

"The RICE framework--Reach, Impact, Confidence, Effort--gives you a simple way to evaluate opportunities without getting bogged down in analysis paralysis. RICE can help you tease out projects that have a higher likelihood of success, targeting those with high reach, high impact, and high confidence, while keeping effort low. For early project selection, think of this as “Quick RICE” or “gut-check RICE”—you’re making rapid assessments to identify promising candidates, not conducting detailed analysis. That deeper dive comes later when you have more data.

When evaluating your candidate projects from interview synthesis:

  1. Score each project on a simple scale (High/Medium/Low) for each RICE component.
  2. Look for patterns--projects scoring High on reach and impact with Medium or Low effort are often winners.
  3. Trust your instincts--if something feels off despite good scores, dig deeper.
  4. Consider combinations--sometimes two smaller projects together create more impact than one large one.

"Turn evaluation into clarity."

"Your DevEx breakthrough might be a process, not a tool."

Chapter 17 - Share Early Wins to Gain Momentum

"Fail fast and share your learnings."

"Celebrate wins along the way."

"Bring others along on the journey."

"Share your findings thoughtfully."

"Keep the conversation going."

Chapter 18 - Avoid Common Pitfalls

Van Buren and Safferstone studied 5,400 leaders implementing new initiatives in companies. Their work reveals common pitfalls that can undermine long-term success. (Van Buren, Mark E., and Todd Safferstone. “The quick wins paradox.” Harvard Business Review 87.1 (2009): 54-61.) Here’s how successful leaders navigate these challenges:

STEP 3 Use Data to Optimize Your Own DevEx Work

Chapter 19 - Establish Your Data Foundation

There are a few kinds of data to think about, and they overlap a bit.

We will also talk about self-report and system data, each of which can be either qualitative or quantitative.

. Qualitative Data Quantitative Data
Self-report Data Interview transcripts; Open-ended survey responses; Feedback comments Survey ratings; Satisfaction scores; Self-reported productivity metrics
System Data Error messages; Log entries; Documentation content; User flow Build times; Deployment frequencies; Error rates; System performance metrics

"Surveys and system data are strategic partners, each with unique strengths."

"The most successful DevEx initiatives use both self-report and system data."

Chapter 20 - Identify and Use the Data You Already Have

"Most organizations have untapped metrics that can provide valuable insights into developer experience."

For example:

Find early insights from existing data. Once you find and document your existing data sources, here are some ways to extract valuable insights:

Chapter 21 - Use Surveys for Fast Insights

"You can get actionable insights now through surveys."

"A powerful DevEx survey doesn’t need to be long—and it shouldn’t be."

"Measure value with the Sean Ellis Test. While active user metrics and satisfaction scores provide valuable signals, occasionally you need a more direct assessment of how essential your solution—or an existing tool—has become to users. The Sean Ellis Test offers a simple and powerful method that startup companies use to determine product-market fit, and it works equally well for internal developer tools. The test centers on a single powerful question: “How would you feel if you could no longer use [your platform/tool]?” With just three possible responses: Very disappointed; Somewhat disappointed ;Not disappointed (it isn’t that useful). Through extensive work with startups, Ellis found that products with strong market fit consistently have at least 40 percent of users answering they would be “very disappointed” if they could no longer use the tool. This threshold has become a reliable benchmark for product success." ... "However, be cautious when interpreting results for mandated tools where users have no alternatives. In these cases, consider adapting the question. For example: If you had a choice of developer tools, how would you feel if you could no longer use [current tool]?” This framing helps distinguish between genuine tool value and organizational dependency. You can also pair the Ellis Test with satisfaction questions to better understand whether high disappointment scores reflect tool love or job necessity."

Chapter 22 - Invest in System Metrics Over Time

When starting out, you might rely heavily on surveys (around 80 percent) with some system data (around 20 percent), giving you quick, broad insights into developer pain points. As you progress in your initiative and you instrument more systems, this balance shifts along an S-curve pattern—system data gradually increases, then ramps up significantly as you build more automated collection capabilities. In mature DevEx programs, system data often comprises the majority of your information simply due to the volume you can collect automatically. Surveys remain crucial throughout your journey and never disappear entirely. Even in mature programs, you’ll continue using survey data for instrumentation decisions, uncovering hidden friction points, and measuring developer satisfaction. The key is that while system data provides precision and continuous monitoring, survey data provides the context and human perspective that systems can’t capture.

Beware survey fatigue.
We all know this but it’s worth repeating: Bombard your developers with surveys and they’ll stop giving you the insights you actually need. You need to be realistic about what data you can get, what you can’t get, and how often you can get it.
Developers only have so much attention, and work is their priority. Your developers are solving complex problems all day. When you ask them to stop that work to fill out a survey, you’re making a withdrawal from a limited attention bank. Make too many withdrawals, and the account closes. Here are some tips for survey success:

Do a final check. Before you hit “send” on that survey, ask yourself: “If I were in my developers’ shoes, would I stop what I’m doing to answer these questions?” and “Would I care about the results of this survey? Would I want to share the results with someone else?” If the answers aren’t obvious yeses, you need to reposition your approach. The teams that win at developer feedback aren’t the ones with the most surveys—they’re the ones who respect the context their developers operate in and position their feedback requests accordingly.

Survey data can be very useful when you don’t have good system data, but don’t fall into the trap of waiting for perfect system data before taking action. Perfect is the enemy of progress. The transition to more system data happens organically as you identify data sources with reasonable quality and coverage across your organization. For instance, if pull requests are a point of friction, you’ll want to analyze metrics like PR cycle time (how long it takes to complete a PR) and PR dwell time (a component of overall cycle time, focusing on how long a PR waits before it is acted on). These objective measures provide precise data about specific friction points.

When considering your data, beware of common fallacies that can mislead your DevEx efforts. The streetlight effect occurs when you focus exclusively on areas where you already have strong system data. Don’t limit your attention to what’s easy to measure. Even in mature DevEx improvement programs, surveys continue to provide crucial insights about emerging issues, subjective experiences, and areas that are difficult to instrument. This is why maintaining some survey component remains valuable even as system data grows to dominate your overall metrics portfolio. Equally important is avoiding the snapshot fallacy. Don’t treat a single survey or data collection as permanent truth. DevEx needs evolve as your organization changes, technologies advance, and developers gain experience. Plan for regular pulse surveys and ongoing system data analysis to maintain an accurate, current picture of developer experience.

Chapter 23 - Make Sure You Capture the Right Data

"The most effective data strategies work backward from problems to metrics. Start with the pain points your surveys reveal, then identify what specific data would help you understand and track improvements in those areas."

"We suggest you also consider the dimensions of developer experience your metrics represent. Frameworks like SPACE (satisfaction, performance, activity, communication, and efficiency and flow) can help ensure you’re capturing a complete picture rather than just what’s easy to measure."

"Don’t forget to collect data to build a strong business case. Many DevEx improvement initiatives stumble because they focus too narrowly on technical metrics like build times and test effectiveness. While these measurements are essential, they often don’t resonate with all stakeholders or clearly demonstrate business value."

"As you get more data, you’ll need to be more careful about analyzing and drawing conclusions from it."

"Get representative data to understand what’s happening. To build an accurate picture, collect data from across your organization. Your data collection strategy—whether for interviews, surveys, or system logs—directly impacts your conclusions."

"Get enough data to understand what’s happening. Your goal is to get enough data to feel confident about insights and next steps, without getting bogged down in academic-level statistical significance. Here are some good rules of thumb for when you’re just getting started.

"Get enough data from enough people to understand what’s happening. For surveys, a response rate above 30 percent is pretty good for most organizational contexts. However, 20-30 percent is fairly typical and usually gives you workable data. If you’re below 15 percent, that’s when you should start wondering if there might be issues with your distribution approach or if the survey itself might be too long or overly complicated."

"When analyzing developer feedback, focus on representation over raw numbers. A 30 percent response rate from developers across different teams, roles, and experience levels will tell you more about patterns across the organization than a 70 percent response rate that’s mostly from one group. This principle applies to both survey data and behavioral metrics."

"Building high response rates. You can achieve surprisingly high participation rates over time by building trust and demonstrating impact. For example, Amazon’s annual developer survey achieved over 90 percent response rates despite taking an hour to complete—a remarkable achievement that came from consistently showing developers how their feedback drove real changes and transparently reporting results back to them. Similarly, hundreds of companies using the DX platform see response rates over 90 percent, thanks to thoughtful survey design, transparent dashboards, and follow-up action that the companies take on the data. The key insight: Response rates aren’t just about the survey itself, but about the broader culture of feedback and action. When developers see their input driving real changes, they’re willing to invest more time in future surveys."

Chapter 24 - Turn Data Into Actionable Insights

When you’re diving into DevEx data, start with the basics.

Performance bottlenecks emerge when your statistical measures don’t align. The gap between median and average in the build time example points to infrastructure inconsistencies or problematic edge cases that are dragging down the development process.

Usage patterns show up in frequency counts and trends over time. A trend line of deployment frequencies could reveal that teams deploy less often on Mondays—sparking questions about team workflows, meeting schedules, or even weekend incident recovery patterns.

Friction points can become visible through distributions and frequency counts. When you analyze where developers spend the most time or encounter the most errors, you can spot the bottlenecks.

Visualizations are your friend--they transform abstract numbers into patterns that people can quickly grasp. A histogram of build times might immediately show you have a “long tail” of problematic builds, while a trend line of deployment frequencies could reveal that teams deploy less often on Mondays or the end of the fiscal year. These visual patterns often spark the right questions: “Why do we see this dip here?” or “What changed when this metric improved?” Even better, visualizations make it easier to share insights with stakeholders who might get lost in raw numbers.

You can do much of the analysis yourself, but sometimes it’s best to partner with a data scientist.

Privacy and security are important to consider when sharing results and visualizations. ... A good rule is the Rule of 5: Never show data for groups smaller than five developers. This prevents someone from figuring out, say, which specific developer is struggling with a particular API or deployment system. Also, be extra careful with free-text comments in surveys—they can accidentally contain sensitive information.

STEP 4 Decide Strategy and Priority

Every item on your ever-growing list is important to someone, but may not resonate with someone else.

Chapter 25 - Focus Your DevEx Efforts

Use RICE criteria to clarify decisions and create buy-in—now with real data.

An important part of prioritizing is saying no. ... RICE helps you say no systematically by forcing you to evaluate the full picture—not just the loudest complaints or the most interesting technical challenges.

Chapter 26 - Use Relevant Criteria to Set Priority for Developer Experience

RICE can help you tease out projects that have a higher likelihood of success, targeting those with high reach, high impact, and high confidence, while keeping effort low.

RICE = (Reach x Impact x Confidence) / Effort

Reach measures how many developers are affected and how often. This is your foundation for understanding true impact scale. Possible questions to ask:

Use consistent units across all calculations. For example, don’t mix “daily impacts” for some items with “number of developers affected” for others. This ensures a fair comparison across different types of improvements.

Impact assesses how significantly this change will improve developers’ daily experience. Questions to ask:

Confidence reflects how certain you are that the initiative will succeed as planned, and is typically estimated as a percentage (e.g., 80% confident). Consider:

Effort measures the total investment needed, typically in person-months. Be sure to include:

Look for business impact and strategic alignment. Start by asking, “Is it easy for me to talk about this challenge in terms of metrics that business leaders care about?” Equally important is considering, “How does this initiative align with or leverage existing work to maximize business value?”

Think about collective vs. local action.

Consider timelines, urgency, and organizational readiness. Key questions here include:

Use constraints and risks. Great questions to ask here include:

Don’t forget team and organizational factors. Ask yourself, “Which challenge allows us to strengthen our people and organization while solving technical problems?” Look for projects where team members can make meaningful contributions that showcase their expertise while creating opportunities for growth and recognition. The most successful initiatives align with teams’ professional development interests and incentives.

Don’t forget other considerations. The examples above aren’t comprehensive–you’ll want to include criteria specific to your organization.

Chapter 27 - See the Bigger Picture: Use a Quick Rubric

Avoid getting lost in the weeds by looking at the big picture.

Create your rubric. Start with the criteria that matter most for your organization. Common DevEx prioritization criteria include:

Make the most of your analysis. The combination of RICE scoring and strategic rubrics gives you perspective:

STEP 5 Sell Your Strategy

Many DevEx teams underestimate the importance of strategic communications.

Chapter 28 - Leverage Your Stakeholder Analysis for Effective Communication

The relationships you’ve built and insights you’ve gathered during your listening tour will help you craft messages that resonate with each group.

Remember your stakeholders are more than just developers.

Tailor your approach based on stakeholder characteristics.

Refine your stakeholder list with communication-specific details. "We often find it helpful to list key stakeholders and information in a table or a spreadsheet. This is handy for organization as well as identifying any information you may be missing for certain stakeholder groups. Your stakeholders—and the information you decide to collect—will differ. Here, you’ll identify major stakeholders, their motivation (why they care), what support level they need (skeptical, neutral, enthusiastic), your engagement strategy (inform, convince, collaborate), their information needs, their engagement strategy, and their preferred communication channels."

Your stakeholders need to know what the problem is and why they should care. Don’t assume everyone sees the DevEx problems faced by your organization as clearly as you do. People come to the table with diverse backgrounds, shaping their views on what’s good, bad, or even possible in your systems. Some stakeholders may also have preconceived solutions—for instance, leadership might assume that adopting AI tools will automatically solve developer productivity challenges, without understanding the limitations of AI or the specific friction points that need addressing first.

Use the EMI framework for stakeholder communications. When crafting DevEx communications, apply the EMI framework to ensure messages resonate with each stakeholder group:

Think of your DevEx initiative like a product—and “sell” it to your stakeholders. When communicating about your DevEx improvement initiative, adopt a product mindset to make your message more compelling. This approach helps you:

For “investors” (leaders who fund your work), you’ll need to be able to briefly outline the business case, competitive landscape, and expected returns. For “customers” (developers and engineering managers), you’ll need to emphasize how your solution addresses their specific pain points. This product-based framing transforms abstract concepts like “improving developer experience” into concrete solutions that stakeholders can easily understand and support. For a more detailed approach to treating your DevEx tooling and automation ecosystem as a product, see the Third Practice: Make Technology Sustainable and Effective.

Chapter 29 - Effective Messaging Takes Customization and Repetition

Psych 101.

Tell people what’s going on—and say it more than once.

Give your DevEx initiative an identity.

You’ll want more materials to help others confidently spread the word without always needing you.

Build capability, not just awareness. "Effective DevEx communication goes beyond informing people—it builds their capability to understand, engage with, and champion your initiatives. Design your materials to move stakeholders through a three-stage progression:

STEP 6 Drive Change at (Your) Scale

Chapter 30 - Adapt Your Approach to Your Scope

Chapter 31 - Drive Change With Global Scope of Control

Chapter 32 - Drive Change With Local Scope of Control

Chapter 33 - Drive Change With Middle Ground Scope of Control

STEP 7 Evaluate Your Progress and Show Value

Chapter 34 - Analyze Your Data

Chapter 35 - Prepare to Communicate Your Results

Chapter 36 - Share Progress

Chapter 37 - Navigate Conflicting Metrics

Chapter 38 - Learn and Improve

Chapter 39 - Evaluation Is Just One Step in a Virtuous Cycle

PART V Evolving and Sustaining DevEx

THE FIRST PRACTICE Resource Your DevEx Initiative

Chapter 40 - Bootstrap Your Early Efforts

Chapter 41 - Build a Dedicated Function

Chapter 42 - Developing Talent for DevEx Roles

Chapter 43 - Make Developer Experience Everyone’s Job

Chapter 44 - Secure Budget at Different Organizational Stages

Chapter 45 - Making DevEx Stick

THE SECOND PRACTICE Create Structures to Support Change

Chapter 46 - A Framework For DevEx Change

Chapter 47 - Support Teams Through Change

Chapter 48 - Guide Organizations Through Change

Chapter 49 - Support Yourself Through Challenging Work

THE THIRD PRACTICE Make Technology Sustainable and Effective

Chapter 50 - Adopt a Product Mindset

Chapter 51 - Use Product Management Principles to Build Better Solutions

Chapter 52 - Use Tooling Standardization to Drive Efficiency

Chapter 53 - Deal with Technical Debt in DevEx Initiatives

Chapter 54 - Bring It All Together: The Sustainable Technology Mindset

THE FOURTH PRACTICE Design Metrics That Evolve and Amplify Impact

Chapter 55 - Design Metrics for Your Stakeholders

Chapter 56 - Use Metrics to Solve Your Stakeholders’ Problems

Chapter 57 - Fast Is Relative to Your Competition

Chapter 58 - Start Simple and Scale Smartly

Chapter 59 - How AI Changes (and Doesn’t Change) the Metrics You Need

Chapter 60 - Maintain Your Metrics Like Any Product

Final Thoughts

Chapter 61 - What Are You Waiting For?


Tags: reading   software development   process   management  

Last modified 14 December 2025