Start at the End: How to Design for Behavior Change

In Start at the End, behavioral scientist and entrepreneur Matt Wallaert challenges the conventional wisdom of product design and business strategy, arguing that most creative endeavors fail because they begin with a focus on ideas, products, or vision statements rather than the ultimate behavioral outcome. Wallaert, drawing on his extensive experience at Microsoft and numerous startups, posits that humans are inherently behavioral scientists, constantly shaping and being shaped by external pressures. He introduces the Intervention Design Process (IDP), a systematic, science-backed framework designed to prioritize behavior change by starting with a clearly defined behavioral goal and working backward to identify and manipulate the “pressures” that influence human action. This summary will comprehensively break down every important idea, example, and insight from Wallaert’s book, ensuring nothing significant is left out, all presented in clear, accessible language.

Quick Orientation

Matt Wallaert, a seasoned behavioral scientist and entrepreneur, contends that most businesses operate on outdated “Mad Men” principles, where decisions are based on charisma and “sexy” ideas rather than measurable outcomes. This leads to a massive waste of resources, particularly in advertising, as companies attempt to create motivation where none exists for products not designed with explicit behavioral goals. Start at the End offers a revolutionary alternative: the Intervention Design Process (IDP). Wallaert promises to demystify behavioral science, making it accessible and applicable to anyone, from corporate executives to individuals trying to change personal habits. He emphasizes that true innovation stems from understanding why people do what they do and consciously influencing the “promoting” and “inhibiting” pressures that drive behavior, ultimately accelerating collective advancement toward a better world.

1. The Intervention Design Process

This chapter introduces the Intervention Design Process (IDP), a foundational, seven-step framework for consciously designing for behavior change. Wallaert illustrates the IDP through his work at Microsoft on Bing in the Classroom, demonstrating how starting with a clear behavioral goal and systematically addressing pressures leads to measurable results.

Overview of the IDP Steps

The IDP begins with a potential insight – an observation about the gap between the current world and a desired “counterfactual” one. This insight must then be validated to ensure it’s not mere speculation. Once validated, it is articulated into a precise behavioral statement defining the target behavior. The next crucial step is pressure mapping, where all forces encouraging (promoting pressures) and discouraging (inhibiting pressures) the behavior are identified. These pressures are then validated through further research. Based on the validated pressures, interventions are designed and selected to modify these forces. Before implementation, an ethical check is crucial to ensure the interventions are responsible. Finally, selected interventions are put through pilots and pilot validation, then tests and test validation, before a scale decision is made, followed by continuous measurement. The IDP is an iterative process, designed to be repeated to achieve ongoing behavior change.

Bing in the Classroom: An IDP Case Study

Wallaert recounts his experience at Microsoft with Bing in the Classroom to exemplify the IDP. The journey began with a potential insight: “Kids don’t search in school nearly as much as you’d think they would.” This was validated quantitatively by analyzing queries per student (QPS) and qualitatively through classroom observations, revealing low search engagement despite ample computer access. The target behavioral statement became: “When students have a curiosity question, and they are in school and near a computer with internet connectivity, they’ll use Bing to answer it (as measured by QPS).”

Pressure Mapping and Validation in Action

Initial assumptions about promoting pressures (e.g., lack of curiosity) were debunked through further validation, which showed that children were curious. Instead, the real barriers were inhibiting pressures related to teachers’ concerns: online safety (adult leakage), advertising in schools, privacy concerns about data collection, and the difficulty of integrating search into the curriculum without chaos. This critical validation saved a multimillion-dollar marketing campaign focused on curiosity, redirecting efforts to address the true pain points.

Designing, Selecting, and Scaling Interventions

Addressing the validated inhibiting pressures, the team designed interventions including a specialized Bing version with SafeSearch locked on, no search ads, and reduced data collection. They also created structured lesson plans centered on Bing’s homepage images. These interventions were subjected to a rigorous ethical check, consulting internal and external experts. A small, “operationally dirty” pilot in three local schools showed positive qualitative and quantitative signals (increased QPS). This success led to a larger test in bigger districts, with professional curriculum design and robust tech solutions. Despite challenges (e.g., IT resistance to desktop clients), the simplified web-based rollout proved effective, leading to a scale decision. Bing in the Classroom debuted to seven million kids covered on launch day, resulting in 40 percent more searches in school and 15 percent more at home. This success even forced Google to turn off advertising in schools, demonstrating the power of a process-driven approach over brute-force advertising. The core process from insight to pilot took roughly eight weeks, highlighting the IDP’s efficiency.

2. Potential Insights and Insight Validation

This chapter delves into the origin of behavior change: potential insights. Wallaert explains that a potential insight is the recognition of a possible “split” in the multiverse, offering an opportunity to move closer to a more optimal reality. The rigorous validation of these insights is crucial to avoid chasing fleeting observations and to ensure that interventions are built on solid ground.

Types of Potential Insights

Wallaert outlines four primary types of potential insights:

• Quantitative Insights: These arise from data, often from recognizing unexpected patterns, unexplained correlations, or studying outliers (both positive and negative). Wallaert stresses that not all data exploration should be hypothesis-driven; allowing data to guide discovery can reveal things previously unnoticed.
• Qualitative Insights: Derived from subjective experience, these come from observing and talking to diverse groups of people. Wallaert suggests activities like “people watching” or organized volunteer trips to gain firsthand understanding of target populations.
• Apocryphal Insights: These are ideas that are “common knowledge” within an organization but may lack formal proof. Wallaert advises paying close attention to these, especially when new to an organization, as they often hint at underlying truths that need validation.
• External Insights: Originating from outside the organization, these can come from academic research papers (Google Scholar is a friend) or cross-pollination with other industries and disciplines. Wallaert advocates for buying lunch for grad students to tap into their specialized knowledge.

The Importance of Insight Validation

Regardless of its origin, a potential insight must always undergo validation. Wallaert emphasizes the scientific principle of assuming things are false until proven true. The goal is to achieve convergent validity: evidence from diverse, independent sources that supports the same conclusion. For instance, if data suggests people aren’t going to optimal pharmacies, validation would involve talking to members, analyzing call transcripts, asking pharmacists, and reviewing external research. This multi-pronged approach is likened to building a table with multiple legs as far apart as possible, preventing confirmation bias—the brain’s tendency to seek out information that supports existing beliefs. Wallaert suggests assigning different validation types to specialized researchers and having them work independently before comparing notes, further reducing groupthink and bias. He advocates for rewarding invalidating an insight as much as validating one, fostering a culture of true scientific inquiry.

Town Halls and Diversity in Insight Generation

Wallaert highlights the case of Flamin’ Hot Cheetos, whose creation stemmed from a janitor, Richard Montañez, who recognized a market gap for Latinx consumers. This story exemplifies the power of generating insights horizontally (agnostic of hierarchy). Companies like Frito-Lay and Bonobos demonstrate how listening to employees, providing accessible resources for validation (like open data warehouses), and encouraging direct communication (e.g., Frito-Lay’s CEO’s direct line) can unlock a “big, wide funnel of opportunities” for behavior change. Diversity in the room during insight generation, encompassing gender, ethnic, cultural, and cognitive variations, significantly reduces blind spots and improves the quality of potential insights. Wallaert concludes by urging readers to shift focus from loving solutions to loving the problem, emphasizing that effective behavior change starts by deeply understanding the unmet needs that insights reveal.

3. Behavioral Statement

The behavioral statement is the anchor of the Intervention Design Process, clearly articulating the precise behavioral outcome intended. This chapter explains its critical structure, the common pitfalls to avoid, and how it lays the foundation for all subsequent design efforts.

Defining the Behavioral Statement

A behavioral statement is a concise, binary articulation of the desired counterfactual world, framed from an explicitly behavioral perspective. It serves as the clear goal, preventing the common mistake of focusing on process or vague visions. Wallaert provides a five-variable template:

When [population] wants to [motivation], and they [limitations], they will [behavior] (as measured by [data]).

Each variable must be binary (satisfied or not) and measurable:

• Population: The specific group whose behavior is being changed. (e.g., “people,” “Latinx,” “K–12 students”)
• Motivation: The core underlying desire driving the potential behavior. (e.g., “get from Point A to Point B”)
• Limitations: Binary preconditions necessary for the behavior to happen, explicitly outside your control. (e.g., “have a smartphone with connectivity,” “live in San Francisco”) This is crucial: don’t list inhibiting pressures here, as those can be modified by interventions.
• Behavior: The measurable action the population will consistently perform given the motivation and limitations. (e.g., “take an Uber”)
• Data: The quantifiable metric used to measure the behavior’s occurrence. (e.g., “rides”)

Wallaert uses Uber’s initial behavioral statement as a prime example: “When people want to get from Point A to Point B, and they have a smartphone with connectivity and an electronic form of payment and live in San Francisco, they will take an Uber (as measured by rides).” He notes Uber’s broad population and general motivation were unusual, reflecting a “go-big-or-go-home” strategy. The limitations were initially daunting but targeted the right early adopters. The behavior/data linkage was exceptionally strong, as rides were automatically measured.

Common Mistakes in Writing Behavioral Statements

Wallaert highlights several common errors:

• Choosing the Wrong Behavior: Often, organizations confuse a vision statement with a behavioral one. Microsoft’s old vision, “A computer on every desk and in every home running Microsoft software,” failed because it focused on the existence of an object, not its usage. This led Microsoft’s Office teams to prioritize sales (features for niche corporate clients) over actual user engagement, resulting in products like Google Docs gaining ground. Similarly, the focus on netbook sales ignored the poor user experience, linking slow performance to Windows itself. The key is to focus on the measurable action users take.
• Choosing No Behavior: This is worse than choosing the wrong one. Vague goals like “Our job is to make the customers love our product” are immeasurable and short-circuit the entire IDP, leading to inconsistent efforts across departments. “Love” is not a behavior.
• Timid Behavioral Statements: Wallaert argues against equivocation. An audacious, absolute goal (e.g., Uber’s “always use Uber”) sets a higher anchor for intervention design, leading to more transformative solutions rather than incremental improvements. The Statue of Liberty example illustrates how an initial anchor impacts perception and ambition.
• Clinging to Your First Behavioral Statement: Markets and businesses evolve. Uber’s statement adapted from “moving people” to “moving something” to include deliveries, and its limitations changed (e.g., accepting cash in some markets). Similarly, watch companies pivoted from “telling time” to “status” when cell phones became ubiquitous. Behavioral statements must be willing to pivot when market forces shift or initial goals are achieved.

Behavioral Statements in Planning

Wallaert advocates for integrating behavioral statements into organizational planning:

• Transparency: Post them prominently, discuss them in meetings, and orient planning processes around them. They enable strong decision-making by providing a clear metric for comparing options.
• Cascading Statements: While the CEO owns the overarching statement, smaller, more specific behavioral statements can be created for departments or individuals. This creates a clear, direct line between individual accountability and the organization’s ultimate behavioral goal, fostering autonomy and clear hierarchies based on behavioral accountability rather than titles. This approach aligns with Objectives and Key Results (OKRs), where the “data” becomes the Key Result and the rest is the Objective.

4. Pressure Mapping and Pressure Validation

This chapter unveils the “secret sauce” of behavior change: pressure mapping. It introduces the fundamental concept of competing pressures – promoting (up arrow) and inhibiting (down arrow) – that determine whether a behavior occurs. Understanding and validating these forces is crucial for designing effective interventions.

The Competing Pressures Model

Wallaert presents a simple visual: ↑ ↓. These arrows represent the dynamic balance of forces that create behavior.

• Promoting Pressures (↑): Make a behavior more likely.
• Inhibiting Pressures (↓): Make a behavior less likely.

The actual behavior is the net product of these forces. If promoting pressures are stronger, the behavior happens; if inhibiting pressures dominate, it doesn’t. Both sides are equally responsible for the outcome. Wallaert uses the analogy of a Mylar balloon that “wants” to float (desired behavior, Point B). Pushing it up is a promoting pressure. Rain, gravity, or someone pushing it down are inhibiting pressures. To make it float, you can add more upward force (more promoting pressure) or reduce downward forces (less inhibiting pressure). The goal of pressure mapping is to understand all these forces to inform intervention design.

Why We Eat M&M’s: Exploring Promoting Pressures

Wallaert uses M&M’s consumption as a prime example. Why do we eat them?

• Taste: A powerful and obvious promoting pressure.
• Beauty/Color: M&M’s are visually appealing. Studies show people eat more M&M’s from a bowl with more colors. The public’s strong reaction to the color change in 1995 (tan to blue) demonstrates the often-unconscious power of color, which is tied to identity.
• Counterrational Pressures (Calories/Hunger): While intuition might suggest calories are an inhibiting pressure due to health concerns, in the context of midday blood sugar crashes, calories (as “hunger-fighting” fuel) become a promoting pressure. Snickers, for example, successfully leverages this by advertising its “satisfying” and “energizing” qualities.
• Cultural Associations: M&M’s benefit from being iconic, nostalgic, lighthearted, and ubiquitous, all contributing to promoting consumption.

Wallaert emphasizes that pressure mapping requires letting go of natural assumptions and recognizing the influence of irrational and counterrational forces. The “flipping the scenario” trick (imagining the opposite condition, e.g., puke-green M&M’s) helps reveal hidden pressures.

Why We Don’t Eat M&M’s: Exploring Inhibiting Pressures

Despite the strong promoting pressures, people don’t eat M&M’s constantly due to inhibiting pressures:

• Physical Availability: This is a primary inhibitor. People eat more candy when it’s on their desk versus across the room.
• Psychological Availability: Google’s practice of putting candy in opaque containers and fruit in clear ones significantly reduced calorie consumption, demonstrating the power of “out of sight, out of mind.”
• Counterrational Pressures (Context/Branding): While lighthearted branding promotes M&M’s for kids, it becomes an inhibiting pressure for a romantic dinner (where Lindt or Ferrero Rocher are preferred). Calories, too, can be an inhibiting pressure when health concerns are salient (e.g., “dad bod”).
• Cost: A dollar is a strong inhibiting pressure for a child’s allowance, or in a world where many live on less than $2.50 a day. However, a high price can also be a counterrational promoting pressure if it signals quality or luxury.

Wallaert stresses that all pressures are context-dependent.

Avoiding Predictable Problems

A common bias is to focus disproportionately on promoting pressures when trying to increase a behavior (e.g., adding rewards) and inhibiting pressures when trying to decrease one (e.g., punishments). This leads to untapped upside in the opposite direction. Uber’s success, for example, largely stems from reducing inhibiting pressures (e.g., automatic payment, reducing wait times) rather than just adding promoting pressures. The IDP forces consideration of both sides of the equation.

Generating and Validating Pressures

Pressures are generated and validated through the same rigorous methods as insights: research and convergent validity (interviews, data science, observations). Wallaert recommends involving the research team directly in pressure mapping and intervention design sessions. Techniques to ensure completeness include focusing on one type of pressure at a time, reversing the behavioral statement’s polarity, and ensuring a diverse group for mapping sessions to minimize blind spots. The very act of using the arrows and listing pressures is an intervention to counteract the natural bias towards promoting pressures and “sounding good.”

5. Intervention Design and Intervention Selection

This chapter transitions from understanding pressures to actively shaping them through intervention design and selection. Wallaert cautions against simply listing interventions for each pressure, emphasizing the importance of creativity in combining and scaling solutions, while also stressing the need for disciplined selection based on potential efficacy rather than intuition or perceived “sexiness.”

Intervention Design: Translating Pressures into Action

Intervention design is the process of translating validated pressures into actionable solutions that can be created. Wallaert likens it to pulling levers, where pressures are the levers and interventions are how they are pulled. He highlights a crucial point: a single pressure can inspire many interventions, and one intervention can address multiple pressures. The goal is not just quantity of ideas, but diversity of ideas, allowing for novel approaches.

Wallaert illustrates this with Clover Health’s flu shot initiative for the black population.

• Problem: Low flu shot rates among black members, leading to higher hospitalizations and deaths, despite shots being free and accessible.
• Validated Pressures:
  • Lack of promoting pressure: “Why do I need one? I’m healthy.”
  • Significant inhibiting pressures (magnified in the black community due to medical racism):
    • Distrust due to changing formulas, perceived “hidden experimentation” (Tuskegee Syphilis Study context), and side effects (mild arm pain).
    • Inconvenience: Having to get it at a specific time/place from a medical professional.
    • Lack of guaranteed efficacy: Unlike other vaccines, it doesn’t prevent flu entirely, only reduces severity.

In the design phase, the team avoided “Mad Men” brainstorming (e.g., “Beyoncé tweets!”). Instead, they focused on grouping pressures to find single interventions with broad impact (e.g., trust being key). For the strong inhibiting pressure of lack of trust in the medical establishment, particularly among older black community members, insights pointed to church leaders as trusted figures. This led to a range of faith-based interventions: drafting letters from leaders, sermons on importance, church fundraiser drives, or even on-site flu clinics within churches. Another intervention idea leveraged Personal Health Motivation data (members’ own words on why health is important) to create personalized promoting messages for flu shots (e.g., “to be healthy for grandkids”). The process aimed for volume of ideas, with a reminder that practicality is secondary at this stage; interventions can be scaled back later.

Intervention Selection: Making Smart Bets

This is described as the “worst and most subjective part” of the IDP, as it requires judgment calls on which interventions to pilot (since not all can be pursued).

• Core Principle: Don’t just pick one; select a range of pilots that maximize the chances of creating behavior change. This involves embracing the assumption that interventions won’t work initially, fighting confirmation bias.
• Optimum Distinctiveness: The goal is to choose a set of options that cover as much of the spectrum as possible with relatively little overlap (e.g., strawberry, marmalade, and kiwi jams, not just variations of strawberry). This ensures broad learning.
• Strategies for Reduction:
  • Combine Interventions: For the flu shot, a “flu clinic in a historically black church on a Sunday with a reminder about protecting the congregation” combined faith-based support, physical availability, convenience, and a promoting pressure (community protection). Wallaert notes that knowing which specific part of a combined intervention works is less important than if the overall intervention is scalable and effective.
  • Scale Down Interventions: An intervention like driving to every member’s house for a flu shot might be too impractical but can inform what’s valuable (convenience, personalization) for larger, more feasible combined interventions.
  • Look at Coverage: Consider which interventions apply to which parts of the population, under what circumstances, and at what times, aiming for the cheapest, easiest-to-create, most broadly reaching options.

Wallaert concludes with the Meetup spam problem example. CEO Scott Heiferman’s counterintuitive suggestion to add a required checkbox, “I pledge to create real, face-to-face community,” was initially met with skepticism (as it added an inhibiting pressure to registration). However, it ultimately reduced spammers and increased successfully created meet-ups by 16%. This occurred because the reminder of the company’s mission strengthened the promoting pressure for passionate organizers, overcoming the minor inhibiting pressure. This story underscores the importance of being open to being proven wrong and piloting interventions based directly on validated pressures, even if they seem counterintuitive.

6. Ethical Check

Wallaert dedicates an entire chapter to the ethical check because of the inherent power and responsibility involved in changing human behavior. He tackles the societal stigma around explicit behavior change and provides a clear framework to ensure interventions are responsible and aligned with the target population’s motivations.

The Paradox of Behavior Change and Free Will

Wallaert acknowledges the societal discomfort with consciously changing behavior, often associating it with manipulation. This stems from a self-serving bias: “when I do good things, it is because I’m a good person, but when I do bad things, it is because I was affected by my environment.” Conversely, others’ bad deeds are attributed to their character. This bias creates a need for a “behavior-changing villain” to explain one’s own perceived shortcomings. Wallaert argues that changing behavior is not inherently ethical or unethical; it is the intent and method that determine its morality.

The Two Ethical Problems: What and How

The ethical check addresses two fundamental behavioral gaps:

1. The Intention-Action Gap: People intend to do something but don’t (e.g., want to go to the gym but don’t go). This is the focus of most behavioral science. Wallaert argues that for this gap, the “what” ethical problem is solved by including a motivation in the behavioral statement. If people want to do something, forcing it isn’t the issue; the ethics lie in how we help them achieve it.
2. The Intention-Goal Gap: People have a goal but no intention of doing the associated behavior (e.g., want six-pack abs but won’t do crunches). This is more ethically fraught.

The Ethics Rule for “What” and “How”

Wallaert presents a comprehensive ethical rule, built in clauses:

If:

• Your outcome behavior is not the result of any of the population’s motivations (addressing the intention-goal gap).
  • This means if the population doesn’t want to do the behavior, even after providing information linking it to their goal, it’s unethical to force it. For flu shots for black members, if members wanted to be healthy but didn’t intend to get a flu shot and knew the link, the original behavioral statement was unethical. The solution was to find an alternative motivation they did possess, like “keeping others healthy” (e.g., grandkids, church congregation). This honored a different motivation.
• Or the benefit of your outcome behavior or an intervention to produce it does not outweigh the cost to an alternative motivation.
  • This addresses whether the positive outcome or the intervention itself creates undue negative consequences. For instance, a flu shot letter with extreme, false language (“Your not getting a flu shot will kill your grandchildren”) would be unethical because the emotional cost (sadness, anger) outweighs the benefit, violating a motivation to feel positive emotions. Similarly, smoking might make someone “look cool” (a motivation) but the “cost of dying” far outweighs that benefit, making it unethical.
• Or you are unwilling to publicly describe and take responsibility for the outcome behavior or intervention.
  • This introduces transparency as a crucial safeguard. Since cost/benefit is subjective, public scrutiny by those without personal biases (like a cigarette marketer’s vested interest) helps ensure ethical judgment. Wallaert stresses pushing this boundary as much as possible, through internal review committees, external experts, and public disclosure of interventions (even failures).

Case Studies in Ethical Failure

Wallaert critiques major tech companies for their ethical missteps:

• Uber’s Driver Incentives: Uber incentivized drivers to stay on the road past safe limits, arguing “any driver can stop work literally at the tap of a button.” Wallaert condemns this “But they have free will!” defense, arguing that by deliberately designing for behavior change, you accept responsibility for its results. The benefit of staying on the road did not outweigh the cost of potential accidents or stress.
• Facebook’s Emotional Contagion Study: In 2014, Facebook manipulated users’ newsfeeds to show more positive or negative content, intentionally affecting their emotional states. Wallaert highlights:
  • Lack of Pilot: A large-scale intervention (700,000 users) was run without a pilot, preventing early detection of harm.
  • Avoidance of Review: Facebook deliberately avoided external Institutional Review Board (IRB) review.
  • Obfuscation: Researchers apologized for the anxiety publishing the study caused, not the negative affect itself. Facebook’s CTO’s blog post deliberately omitted promoting negative content.
  • Wallaert notes Facebook’s subsequent creation of ethical guidelines and internal review committees (similar to his own suggestions), but questions why this wasn’t done earlier. He points out that Facebook lost billions in market value following the outcry, underscoring the link between ethics and business viability.

Wallaert concludes by reminding readers that the IDP provides guardrails, but individual responsibility for the behavioral outcomes created is paramount.

7. Pilot and Pilot Validation, Test and Test Validation, Scale Decision and Continuous Measurement

This chapter details the final, crucial stages of the IDP: piloting, testing, and scaling interventions, emphasizing their distinct purposes and the continuous need for validation. Wallaert clarifies the iterative nature of these phases and the statistical concepts that underpin confident decision-making.

Pilot: Learning Quickly, Minimizing Risk

Pilots are tightly scoped interventions designed for speed and resource efficiency.

• Purpose: To quickly determine if an intervention can at all create the desired behavior change. Wallaert stresses that we expect pilots not to work, as this mindset helps combat confirmation bias.
• Population: Small, targeted groups.
• Methodology: “Operationally dirty” – minimal investment in durable processes, as iteration and refinement are expected. This reduces employee abrasion (people invest themselves in their creations, and large, polished failures are more demoralizing).
• Speed: A rule of thumb is no longer than two weeks to get into the field. If it takes longer, scale it back.
• Validation: Pilot validation involves both qualitative and quantitative confirmation that the intervention is “headed in the right direction.” Statistical significance isn’t required due to the small sample size (N), but it helps refine measurement instruments for later stages.
• Decision Points: If a pilot shows a null or negative result, the options are to revise and rerun or kill it and revisit the pressure map. Wallaert advises considering patterns across multiple pilots: if multiple interventions targeting the same pressure fail, that pressure might be weaker than initially thought. Ultimately, the goal is behavior change, not validating individual interventions or pressures.

Statistics in Validation: Effect Size and P-Value

Wallaert briefly explains key statistical concepts:

• Why Statistics?: People are neither perfectly predictable nor unpredictable. Statistics helps determine how much of a behavior change is due to the intervention versus random variation.
• Sampling: Since interventions can’t be run on everyone, a sample is used to generalize. Larger samples yield more confidence.
• Between-Subjects Test: Compares a treatment group (received intervention) to a control group (did not).
• Within-Subjects Test: Compares behavior before and after the intervention in the same group.
• Effect Size: Quantifies how much the intervention changed behavior (e.g., “20 percent more people will get a flu shot”). Higher number means greater effectiveness.
• P-Value: Indicates how confident you can be that the effect was real and not random chance. A lower p-value means higher confidence. (e.g., p=0.2 means 20% chance of being wrong).
• Challenging Convention: Wallaert argues that the academic convention of p < 0.05 (less than 5% chance of error) is too strict for business. For pilots, a p=0.2 might be acceptable if the intervention has low cost and high potential benefit, because the goal is to decide whether to proceed to a larger test, not to prove a universal truth. The primary advantage of pilots is the low effort, which makes it easier to admit failure and terminate ineffective interventions, combating confirmation bias.

Test: “Is the Juice Worth the Squeeze?”

Tests are like pilots but on a larger scale with greater operational diligence.

• Purpose: To determine if the intervention is worth scaling, considering both its impact on behavior and the resources required.
• Population: Larger sample size, aiming for statistical significance.
• Methodology: More operationalized and refined than a pilot, to see if it can become part of standard operating procedure.
• Refinement: Interventions are often refined based on pilot learnings, and these changes are validated to ensure they don’t negate the desired outcome.
• Risk: More people are exposed, increasing the impact of failure. Leaders must have the fortitude to kill interventions even if there’s emotional investment, especially given that a p=0.2 pilot means 20% of such interventions might not work in larger tests.
• Outcome: The primary outcome of a test is a scale decision, captured in a “juice/squeeze” statement: “We are [confidence] that [intervention] will [direction] [behavior] (as measured by [data]). Scaling this requires [effort] and will result in [change].” This standardized statement prioritizes data-driven decisions over political pitches.

Scale Decision and Continuous Monitoring

Once a test proves worth, a scale decision is made. However, the process doesn’t end there.

• Continuous Monitoring: Even scaled interventions eventually stop working or become less effective. Continuous monitoring is ongoing validation to track the health of the entire behavior-change portfolio.
• Piranha Effect: Multiple interventions targeting the same behavior can cannibalize each other’s gains. Continuous monitoring helps assess the incremental effect of each additional intervention and optimize the “blend of squeeze.”
• Responding to Change: Pressures shift over time (e.g., changing ad effectiveness). Continuous monitoring allows detection of these shifts, prompting modification or termination of interventions that are no longer effective, with documented history informing future efforts.
• Interruptive Alerts: Dashboards alone aren’t enough; continuous monitoring requires alerts to notify stakeholders when interventions deviate from expected performance.

Wallaert closes by reiterating the IDP’s power to transform decision-making from a “tautology masquerading as a meritocracy” (where ideas succeed based on who pitches them) to a balanced, data-driven process focused on measurable behavior change.

8. The End of the Beginning

Wallaert acknowledges the messy reality of implementing the IDP but encourages readers to embrace it. He emphasizes that the book is a call to action: to become a behavioral scientist, formally or informally, and to shift organizational culture toward validation and outcome-focused work.

Embrace the Messiness and the Mission

Wallaert admits that even with his resources, implementing the IDP is rarely as smooth as presented. Organizations and people are complex. However, this messiness is not a reason to avoid the process but to embrace it. He reiterates Teddy Roosevelt’s quote, “Far and away the best prize that life has to offer is the chance to work hard at work worth doing,” connecting it to the meaning found in working towards behavior change. He encourages readers to:

• Go forth and multiply: Spread the ideas, create their own IDP versions.
• Democratize behavioral science: Remind others that if “science is their method and behavior their outcome, they too are behavioral scientists.”
• Seek help: Wallaert offers his own time freely, embodying the zealot’s passion for conversion.

Beyond the Basics: Advanced Behavior Change

Wallaert notes that the remaining chapters in Part 2 are “case studies in behavior change, deep dives on specific pressures, and ruminations on some of the thorny bits.” They can be read nonlinearly and are intended for those ready for “more” – those who have absorbed Part 1 and are ready to tackle the nuances of human behavior.

Applying the IDP Internally: Changing Organizational Behavior

Wallaert anticipates a common objection: “This won’t work in my organization due to lack of executive support, resources, or the right people.” He flips the script: the IDP is a universal process for designing behavior change, applicable not just to external customers but also to internal organizational behavior.

• Challenge: If you feel stumped by internal resistance, identify the specific organizational behavior that blocks IDP implementation.
• Action: Write a behavioral statement for that internal behavior, find and validate insights, map pressures, design interventions, and then pilot, pilot, pilot.
• Benefits: While initial implementation might be slower, systematic validation at every step leads to continuous learning and an organized process where “everything you do builds on itself.” This contrasts sharply with “just ship it” cultures that constantly pivot without accumulating knowledge.

The Power of Systematic Change

Wallaert concludes with a powerful statement of purpose: “Life invites science. Everything is subject to deliberate change.” Having read the book, readers no longer have the excuse that a behavior “cannot be changed.” They now possess a systematic process to create change. While choosing which battles are “worth your time” is personal, the message is clear: be undaunted. When the “right thing is the easy thing, everyone will do the right thing.” He believes that even starting with changing the behavior of just one person, or just oneself, contributes to a larger challenge to the human race.

9. Priming, Moderation, and Mediation

This chapter delves into identity as the “One Ring of pressures,” arguing it’s the most powerful force shaping human behavior and thus a crucial leverage point for interventions. It explains how to incorporate identity into the IDP through priming, moderation, and mediation.

Identity as a Powerful Pressure

Wallaert argues that identity drives enormous resource allocation, from consumption choices to social media expression. It’s why advertising focuses heavily on connecting products to our self-concept. He defines identity not just by “roles” (dad, behavioral scientist) but as a hierarchy of values associated with specific behaviors: “I’m the kind of person who [value/behavior].”

• In-groups: Roles and associated values/behaviors that affirm who we are (e.g., country boys listen to Johnny Cash). These are promoting pressures.
• Out-groups: Roles against which we define ourselves (e.g., not liking the “old boys’ club”). These can be inhibiting pressures (avoiding what the out-group does) or even promoting pressures (doing the opposite of what the out-group likes).
• Values: Act like “colored glasses,” modifying how other pressures (e.g., cost, which becomes “luxury”) are perceived.
• Flexibility: Identities are flexible and context-dependent (e.g., religious identity matters more at church than work), influencing the strength of pressures.

Interventions to Leverage Identity

Wallaert explains three techniques for using identity in interventions:

• Priming: Activating an existing strong connection between an identity and a behavior. This can be done consciously (asking questions about identity) or unconsciously (objects in the environment). He cites a study where Asian women performed better on math tests when primed with their “Asian ethnicity” but worse when primed with their “female gender.” However, he notes that priming works best when the association is strong and clear, and can weaken over time as societal stereotypes evolve.
• Moderation: Modifying the strength of an existing connection between an identity and a behavior. Always’s #LikeAGirl campaign is a prime example. It aimed to weaken the association between “girl” and “terrible running/throwing” by showing actual girls running confidently. The intervention (the conversation in the ad) changed the perception of the association, allowing women to run more confidently.
• Mediation: Creating an entirely new motivational pathway by introducing a “waypoint” between a role/value and a behavior where no strong connection previously existed. Wallaert shares his mother’s journey from being computer-averse to becoming a nursing informaticist. His own attempts to force computing on her failed because his “promoting pressures” (customization, power) didn’t align with her identity. Her hospital successfully mediated by linking “computing” to “care” (a value central to her identity as a nurse). This new association (computing = care) created a motivational pathway that transformed her behavior. Mediation creates opportunities for moderation, which then creates opportunities for priming.

Wallaert concludes that using identity in interventions requires care due to ethical responsibilities but is powerful because it addresses how people see themselves and others.

10. Optimum Cognition

This chapter focuses on cognitive attention as a universal, limited resource and a powerful pressure shaping behavior. Wallaert argues that effective behavior change design seeks to achieve “optimum cognition” – ensuring people spend their mental resources on what they care most about, and as little as possible on what they don’t.

Cognitive Attention as a Limited Resource

Our brains are cognitive misers, relying on biases and heuristics to conserve mental resources, especially under stress. This means that if interventions solely focus on promoting pressures (like advertising), it becomes a “nuclear arms race” for attention. Wallaert points out that while Facebook initially maximized “time on site” (attention), a better approach might be to make the same level of connection less cognitively expensive. The goal isn’t zero cognitive spend for all behaviors but a Goldilocks brain state: “just enough cognition to keep us happy, not so much that we end our day exhausted for the wrong reasons.”

Strategic Allocation of Cognitive Resources

• Uber’s Success: Uber is effective because it makes getting from Point A to Point B cognitively light, removing the burden of planning and worry. This is desirable for people who don’t want to spend mental resources on transport.
• Cars as a Counter-Example: Millions of cars are sold because some people enjoy and want to spend cognitive resources on driving, choosing, and maintaining them. Behavioral scientists must create products that allow for this spectrum of desired cognitive spend.

Gaining Insights into Cognitive Preferences

To design for optimum cognition, insights must be specific:

• Quantitative: Analyze how people allocate time, money, and other resources, and which behaviors are dropped first under resource constraints.
• Qualitative: Ask about cognitive trade-offs and where people wish to spend more or less time, but take answers with a grain of salt due to introspection biases.
• Blue Apron Problem: Blue Apron aimed to make cooking “easy” (reducing cognitive spend) but actually made ingredient and recipe selection easy, while cooking itself remained effortful. This targeted a smaller population who wanted to spend mental resources on the act of cooking, not sourcing. Wallaert notes that removing cognitive spend indiscriminately can reduce enjoyment if the effort was part of the pleasure (e.g., the “joy of stopping running” after a run).

Cognitive Habits and Interventions

Wallaert highlights two key cognitive habits and their implications for interventions:

• Automaticity vs. Curation: Some people (like Wallaert with clothes) want to minimize cognitive spend and prefer automation. Others (like Wallaert with computers) want to maximize spend and seek curation (deep, thoughtful content, configurators). Interventions can be designed to cater to these preferences, often through interface differences, potentially allowing a single product infrastructure to serve different market segments (e.g., a “Green Spatula” brand for Blue Apron that simplifies recipes).
• Satisficing vs. Maximizing:
  • Satisficing: Looking for “good enough.” Highly cognitively efficient but doesn’t yield the absolute best.
  • Maximizing: Seeking the absolute best. High cognitive investment.
    Interventions can target these: limiting choice sets impacts maximizers (makes finding “best” easier), while abundant choice might overwhelm satisficers.

Cognitive Environment and Salient Cues

The cognitive environment in which a behavior occurs (crowded bar vs. quiet office, tired vs. alert, drunk vs. sober) profoundly impacts available cognitive resources and thus behavior.

• MacDonald Paper Revisited: Alcohol impairs cognition, leading people to focus on salient cues. This opens opportunities for behavior change using defaults or strong, obvious cues when cognitive resources are low.
• Defaulting: The more overloaded the environment, the more likely people accept defaults (e.g., organ donation rates in Germany vs. Austria). An intervention leveraging a busy environment and a strong default can drive significant change.
• Reducing Cognitive Burden: Can encourage more reasoned approaches when the desired behavior results from careful consideration (e.g., lesser punishment for crimes of passion vs. premeditated murder).
• Novelty: Our brains attend to novel stimuli and ignore repetitive ones. Manipulating the novelty of a behavioral cue can increase attention and strengthen an intervention.

Wallaert’s advice: be specific about where your population wants to spend resources, consider cognitive habits and environments, and direct brainpower wisely.

11. Uniqueness and Belonging

This chapter explores the fundamental human paradox of simultaneously desiring uniqueness (to stand out) and belonging (to fit in). Wallaert frames this “snowflake-in-a-blizzard problem” as a constant, resource-consuming balancing act that offers fertile ground for behavioral interventions.

The Paradox in Action

• Website Log-ins: Websites subtly cater to both needs. Displaying your name and profile picture (e.g., “Hi, Matt!”) taps into uniqueness. Showing how many friends liked a book or song (e.g., via Facebook Connect) leverages the need for belonging and homophily (tendency of like things to group). These seemingly small interventions reliably increase engagement.
• Coke’s Name Cans: An easily scalable intervention that honors uniqueness through customization.
• Integrating Pressures: Wallaert advises consistently asking how uniqueness and belonging can be integrated into almost any intervention.

Cultural and Socioeconomic Influences

The balance between uniqueness and belonging varies:

• Western vs. Eastern Cultures: Western cultures, emphasizing uniqueness, perceive more individual “choices.” Eastern cultures, focusing on belonging, see actions more as part of a group.
• Socioeconomic Status (SES): High-SES individuals, secure in their belonging, often prioritize uniqueness (e.g., buying a unique car). Low-SES individuals, who may already “stand out” due to circumstance, often seek belonging (e.g., forming a car club if a friend buys the same unique car). Wallaert emphasizes using these general tendencies at a population level for intervention design.
• Political Rallies: Trump rallies appeal to belonging (uniformity, shared chants). Clinton rallies, while celebrating diversity, needed to make the expression of uniqueness easier (e.g., tools for branding uniqueness on social media) to effectively leverage that inherent promoting pressure in her diverse base.

The Uniqueness/Belonging Matrix: Stable vs. Unstable Liking/Disliking

Wallaert introduces a 2×2 matrix based on a person’s valence of reaction (liker/disliker) and the stability of that reaction:

• Stable Liker (Snowflake): Deeply engaged, preference is core to identity, impervious to popular opinion. (e.g., Wallaert’s love for Johnny Cash). They offer deep insights and are valuable beta testers (e.g., Microsoft Insiders programs). They want direct engagement with the product/brand itself, not just social signaling.
• Stable Disliker (Snowflake): Also deeply engaged, but with a permanent negative opinion, often rooted in identity. (e.g., Wallaert’s dislike of thin-crust pizza). They are valuable for identifying inhibiting pressures and also seek direct engagement (e.g., conversation about differences).
• Unstable Liker (Blizzard): Preferences are influenced by current trends, seeking to “fit in.” (e.g., liking Johnny Cash after Walk the Line). Their primary drive is belonging. They are natural recruiters and respond well to referral programs, social sharing buttons, and factoid-style content that supports group affiliation without highlighting their impermanent preferences. Caution: They are highly sensitive to herd mentality and will drop preferences if the group shifts. Continuous monitoring is crucial for interventions targeting this group.
• Unstable Disliker (Hipster/Blizzard): Defines self by rejecting popular trends, seeking belonging through shared antagonism. (e.g., disliking Johnny Cash after Walk the Line). They actively recruit others to dislike.
  • Intervention: One effective strategy is to silence them by showing their behavior is unacceptable to the majority, depriving them of the community they seek. Wallaert cites the “Ask ForHelp” fake Target Facebook account that mocked anti-gender-norm commenters, effectively shutting down their “community of haters.” Silence is a decent outcome for this group.

Operationalizing Uniqueness and Belonging

Wallaert suggests two practical tips:

• Researcher Focus: Have qualitative and quantitative researchers focus on these four identity types (stable/unstable likers/dislikers) during the insight phase to avoid blind spots.
• Team Role-Playing: During pressure mapping and intervention design, encourage team members to embody these identity roles and discuss how they would react, providing valuable feedback for modifying interventions.

12. Special Factors of Inhibiting Pressures

Wallaert confesses his bias: he loves inhibiting pressures and argues that they possess special properties that make them disproportionately effective in behavior change, often outperforming promoting pressures. This chapter details these unique characteristics.

Why Inhibiting Pressures Are Special

1. Likely Ignored: Because of the natural human bias towards promoting pressures (rewards, gains), inhibiting pressures are often overlooked. This makes them fertile ground for new, impactful interventions and low-hanging fruit.
2. Higher Effectiveness/Efficiency: Promoting pressure interventions are a “crowded space” (e.g., $220 billion in advertising) where effectiveness is reduced due to intense competition. Inhibiting pressures offer opportunities to be distinctive and stand out, achieving significant change with smaller efforts. Uber’s success with electronic payment (an inhibiting pressure reduction) is a prime example; it was a simple change but had an outsized impact because no one else was addressing it.
3. Homogeneity: Unlike promoting pressures, which can vary widely across individuals (e.g., different reasons for eating M&M’s), inhibiting pressures tend to apply universally. Cost, availability, and convenience affect everyone, making interventions addressing them disproportionately efficient.
4. Longevity: Inhibiting pressures “don’t date themselves as quickly.” While brand messages and promoting pressures need constant refreshing to stay relevant, fundamental inhibitors like cost or basic availability remain constant over long periods, making their reduction a more durable intervention.
5. Predictability: Inhibiting pressures often come with quantifiable units (distance, dollars). While not perfectly linear in human perception, these units provide greater control and understanding for powering interventions, especially at scale.
6. Prospect Theory (Nobel Prize Worthy): Daniel Kahneman and Amos Tversky’s prospect theory states that “equivalent losses hurt worse than equivalent gains feel good.” This implies that reducing an inhibiting pressure will generally be more effective than adding an equivalent promoting pressure, especially when an inhibitor is eliminated entirely (e.g., the “penny gap” – getting a free T-shirt versus paying a penny for it changes behavior dramatically).

Wallaert concludes by emphasizing that while not all inhibiting pressures are superior, their collective special properties make focusing on them a highly strategic and often neglected approach to behavior change.

13. Competing Behaviors

This advanced chapter introduces the concept of competing behaviors, acknowledging that human actions are interconnected. It explains how to identify and address alternative behaviors that vie with your desired outcome, whether through direct competition or co-optation. Wallaert cautions that this is a complex strategy for “Behavior Change 201.”

Interconnectedness of Behavior

The IDP typically examines a single behavior in isolation. However, Wallaert notes that “at some level, everything competes with everything else.” Changing one pressure or behavior inevitably affects others. For example, lowering the price of bottled water affects soda sales. The core idea is that reducing alternative behaviors is a valid strategy for increasing your outcome behavior.

The Competing Behaviors Model

This model requires at least two sets of arrows (two IDPs) instead of one.

• You run a standard IDP for your outcome behavior (e.g., increasing bottled water sales).
• You then run an entirely new IDP focused on eliminating a specific alternative behavior (e.g., reducing soda sales).
• Crucially, when measuring the second IDP, you still measure the impact on your true behavior of interest (bottled water sales), not just the reduction in soda sales.

Risks and Opportunities

• Risk of Shrinking the Market: Trying to eliminate alternative behaviors that are too close to your target behavior can inadvertently shrink the overall market for related products, leading to less market share even if your product gains a higher percentage.
• Goldilocks Principle: Identifying the right alternative behavior to target is key—it must be “close enough that it is a viable alternative, far enough away that reducing it doesn’t inadvertently tank your outcome behavior.” Trying to eliminate wildly unrelated behaviors (e.g., skydiving instead of reading) is a waste of resources.

Co-opting Competing Behaviors (Synergistic Bundles)

Instead of extinguishing alternative behaviors, they can be co-opted. Wallaert uses the Uber vs. Netflix “war” as a prime example:

• Uber’s Goal: Get people to go out (and get drunk) on a Friday night.
• Netflix’s Goal: Get people to stay in on a Friday night.
• Mutually Exclusive: These were competing behaviors, though neither company might have explicitly recognized the “war.”
• Resolution through Product Development:
  • Uber began delivering food: Now it’s fine if you stay in.
  • Netflix started mobile streaming: Now it’s fine if you go out (you can watch Netflix in an Uber).
• Synergistic Bundles: The ideal (though not always realized) solution would be explicit partnerships, like free Netflix streaming in Ubers. These bundles are powerful alternatives to outright competition.

Wallaert advises smaller organizations to focus on their own behavioral goals rather than creating competition where it doesn’t naturally exist, emphasizing that this is a “right time, right place” strategy for larger entities.

14. Eliminating and Replacing Behavior

This chapter addresses the specific challenge of eliminating a behavior, which is crucial not only for prosocial goals (e.g., reducing spending) but also for capitalist aims (e.g., Uber replacing car ownership). Wallaert highlights that while the IDP still applies, a key quirk is the need to replace the eliminated behavior to avoid unintended, worse outcomes.

The IDP for Elimination

Eliminating behavior fundamentally uses the same IDP, but with a reversed focus:

• Increase Inhibiting Pressures.
• Reduce Promoting Pressures.
• Wallaert reminds us of the bias: just as we default to promoting pressures when increasing behavior, we tend to default to inhibiting pressures (punishments!) when eliminating behavior. This means there’s untapped upside in removing the promoting pressures that cause the behavior in the first place.

The “Doowy” Quirk: Don’t Leave a Motivational Vacuum

The critical difference in eliminating behavior is the need to address the motivation in the behavioral statement. If a behavior is eliminated without providing an alternative way to meet the underlying motivation, a behavioral gap is created. “Nature abhors a vacuum,” and people will find new (often worse) behaviors to fill that gap. Cady from Mean Girls is used as an example: stopping Regina George’s “mean girl” behavior without providing a new pathway for teenage identity expression led to even worse group dynamics.

The Case of Smoking Cessation

Wallaert analyzes the public health victory of reducing smoking rates:

• Early Focus: Inhibiting Pressures:
  • Extreme Warnings: Giant warnings on packages leveraged the fear of death.
  • Public Bans: Smoking moved from a public to a private behavior.
  • High Taxes & Regulations: Made cigarettes expensive and harder to sell.
  • These worked to an extent.
• Later Focus: Promoting Pressures: The biggest win was attacking the perception of smoking as “cool” (a powerful promoting pressure for teens driven by uniqueness and belonging).
  • Counter-Advertising: Ads showing negative consequences (e.g., attractive woman smoking causing man to veer away, implying no sex).
  • Advertising Bans: Banning tobacco ads removed companies’ ability to promote “coolness” or glamorize smoking.
  • Killing Joe Camel: The “Joe Camel” cartoon character was hugely effective at promoting smoking to teens. Its elimination saved lives.

The Unmet Motivation and Its Consequences

Despite the success, killing cigarettes left a massive, gaping hole for the underlying motivation: the teen need to look cool, have ritual, foster belonging, and initiate social connection. This vacuum was filled by e-cigarettes (Juul), which offer similar ritual, social prompts, and “cool” factors, but in a “marginally improved” (less harmful) form. Juul’s $13 billion investment from Altria (a cigarette company) highlights the natural synergy when the motivation remains the same but the behavior shifts. To truly “end smoking,” a replacement behavior that honored these intrinsic motivations was needed.

Counterexample: Heart Disease in Africa

Wallaert provides a successful counterexample:

• Problem: Africans adding excessive sodium to food (behavior) to combat a bland diet (motivation).
• Intervention: Instead of just saying “stop salting,” they introduced an array of non-salt spices.
• Outcome: By honoring the motivation for flavorful food and providing a viable replacement, they successfully eliminated excessive salt use without causing other unhealthy additions (like sugar or oil) to fill the void.

The key lesson is that motivations are what matter most. To eliminate a behavior effectively, you must identify its underlying motivation and provide an alternative, preferable behavior to satisfy that motivation.

15. Mini Case Studies

This “lightning round” chapter offers several concise case studies, demonstrating the IDP’s application in diverse, real-world scenarios, particularly highlighting the power of inhibiting pressures and counterintuitive interventions.

The Good Samaritan: Time as an Inhibiting Pressure

• Problem: Why do people fail to help others in distress?
• Study: Darley and Batson (1973) had seminary students walk past an actor feigning distress.
• Hypotheses Tested (Promoting Pressures):
  • Religiosity/Identity: Did personal beliefs about helping predict action? No.
  • Priming: Did giving a sermon on the Good Samaritan parable prime helping behavior? No.
• Key Finding (Inhibiting Pressure): The single most powerful determinant was timeliness. Students who were told they were running late were significantly less likely to stop (only 1 out of 10), even when rushing to give a sermon about helping.
• Takeaway: Perceptions of lateness (a cognitive/environmental inhibiting pressure) can powerfully override strong promoting pressures like moral values or identity.

The Waiting Subway Rider: Ambiguity as an Inhibiting Pressure

• Problem: People hate waiting for public transit, perceiving delays even when service is good. Expensive infrastructure changes to speed up trains are often not feasible.
• Inhibiting Pressure: Ambiguity – not knowing when the next train will arrive. This causes anxiety and paralysis.
• Intervention: Installing countdown clocks (e.g., in NYC subways).
• Outcome: Even without speeding up trains, countdown clocks made people feel like transit was 30% faster because they eliminated ambiguity, allowing riders to make informed decisions.
• Takeaway: Perception is reality in behavior change. Eliminating the inhibiting pressure of ambiguity can be a powerful, often low-cost, intervention. Disney similarly overstates wait times to create positive surprises.

The Frequent Traveler: Identity as a Promoting Pressure for Efficiency

• Problem: Airlines struggled to get business travelers to check bags, despite eliminating fees, guaranteeing quick delivery, and minimizing loss. Overhead space issues led to delays.
• Existing Inhibiting Pressures Removed: Cost, baggage loss risk, wait time at carousel.
• Underlying Promoting Pressure: Business travelers strongly link their identity to efficiency (“I’m the kind of person who is efficient”). Carrying on a bag, while seemingly efficient for the individual, contributes to overall delays, conflicting with this identity.
• Intervention (Proposed): Connect “I check my bag” to “I am efficient” through messaging.
• Takeaway: When other inhibitors are removed, an identity incongruence can become the dominant barrier. Linking the desired behavior to a core identity value can be a powerful promoting pressure.

The Absent Flight Attendant: Unmet Motivation/Limitation

• Problem: Airlines struggled to reduce last-minute sick calls from flight attendants, which cause costly operational ripples.
• Initial Interventions (Proposed): Public shaming for last-minute calls (inhibiting pressure), easier forms for early calls (reducing inhibiting pressure), sick time rebates for early calls (promoting pressure). All valid for piloting.
• Key Insight (Unmet Motivation/Limitation): Digging deeper revealed that the primary reason for last-minute calls was sick kids and lack of child care. Attendants wanted to call out early.
• Winning Intervention: Providing on-demand child care.
• Takeaway: People often want to do the right thing. The real problem is often an unmet motivation or an unaddressed limitation (like lack of child care) that prevents the desired behavior. Make the “right thing” easy.

The Yammering Product Manager: The Perils of Unclear Behavioral Statements

• Problem: Microsoft’s Yammer team (acquired for over $1 billion) faced internal disagreement about their core behavioral goal: was it engagement (how often people used Yammer) or business value (how often they created something helpful for the company)?
• Wallaert’s Intervention: A thought experiment: which person is preferred – someone highly engaged but creates no business value, or someone who logs in once but creates massive business value?
• Outcome: The room erupted in unresolved chaos.
• Takeaway: A clear, single behavioral statement is crucial for alignment, accountability, and avoiding wasted effort. If you can’t focus on one primary behavior, you can’t effectively design for it. This highlights the IDP’s function as a “forcing function” for early conflict resolution.

The Underpaid Woman: Systemic Bias and Inhibiting Pressures

• Problem: Women are significantly underpaid and underpromoted. Traditional feminist interventions often focus on promoting pressures (e.g., “lean in,” confidence coaching).
• Insight (from Wallaert’s work at Thrive): Women are excellent at saving as a function of income, but the raw wage gap makes up for any savings prowess. The issue is a lack of dollars coming in.
• GetRaised.com (Intervention): Wallaert built a tool focusing solely on inhibiting pressures related to asking for raises:
  • Reduces Effort: It uses Bureau of Labor Statistics data to tell women how much they’re underpaid.
  • Reduces Ambiguity: Recommends an optimal raise percentage.
  • Automates Process: Generates a letter to print and hand in.
  • Promoting Pressure (Mild): Follow-up emails track progress.
• Outcome: Over 80% of women who use it get a raise, averaging over $7,000.
• Takeaway: Systemic issues often have their roots in overlooked inhibiting pressures. Removing barriers (effort, ambiguity) can be more effective than trying to “motivate” someone to overcome them. This “wizard format” is repeatable (e.g., SalaryOrEquity.com to help women choose equity over salary).
• Expanding the Principle: Wallaert also applies this to increasing male feminists. His research identified “blind spot” men who acknowledge sexism but don’t act because they don’t see it proximally.
  • IAskedHer.com (Promoting Pressure): Encourages men to talk to women they’re bonded with about sexism, exposing them to proximal examples.
  • WhyMenAttend.com (Inhibiting Pressure Reduction): Found that men who attended gender-focused events did so because they were invited, saw male speakers, and the event was explicitly open to men (reducing inhibiting pressure of feeling unwelcome). Invitations from women were more effective.
• General Takeaway: Simple, open-data interventions that focus on pressures can be highly effective, even without a billion-dollar war chest.

The Snacking GI: Inhibiting Pressures and Market Domination

• Problem: In 1941, during wartime rationing, chocolate was scarce and melted easily, limiting consumption.
• Intervention (M&M’s): Forrest Mars Sr. developed a chocolate that “won’t melt” for military rations, addressing a major inhibiting pressure for the military.
• Post-War Impact: After the war, M&M’s dominated the candy market by unwittingly reducing inhibiting pressures for the general public:
  • Social Acceptability: Post-war, chocolate was less scarce, reducing social pressure against “fun” consumption.
  • Melting Problem: “Melt in your mouth, not in your hand” eliminated the inhibiting pressure of chocolate melting in warm weather, enabling consumption at picnics and ball games.
  • Cost/Distribution: Not melting allowed for wider, cheaper distribution, reducing the cost pressure of needing local production or climate-controlled stores.
• Takeaway: A world war forced the discovery of critical inhibiting pressures. Companies that are first to recognize and address previously unseen pressures (like Uber with payment, Netflix with streaming, Tinder with swiping, Juul with vaping) can lead industries. Wallaert challenges the reader to find the next one.

Key Takeaways

Start at the End is a powerful manifesto for anyone seeking to create meaningful change, from product managers to concerned citizens. Its core lesson is a radical shift in perspective: start with the behavior you want to see, not the product you want to build. By systematically applying the Intervention Design Process (IDP), focusing on validated promoting and inhibiting pressures, and embracing a culture of skepticism and continuous learning, you can dramatically increase the likelihood of achieving your desired outcomes. This approach empowers individuals and organizations to move beyond intuition and “sexy ideas,” instead building interventions that are effective, ethical, and truly transform the world.

To begin, define your behavioral goal with a precise, binary behavioral statement that includes population, motivation, limitations, behavior, and data. Then, meticulously map and validate the competing pressures (both promoting and inhibiting) that influence this behavior, remembering that often, reducing inhibiting pressures offers untapped and highly efficient leverage. Design a diverse set of interventions, rigorously pilot and test them (even if they fail, as this fuels learning), and always ensure they pass a robust ethical check based on transparency and alignment with intrinsic motivations. Finally, implement continuous monitoring to ensure sustained impact and adapt to changing pressures. This iterative, scientific approach allows for precise, measurable change, transforming aspirations into tangible reality.

What specific behavior would you most want to change in your life or organization, and which single inhibiting pressure would you prioritize removing first?