Fogg × Octalysis: The 8 Motivations Behind B=MAP’s One

A behavior refuses to happen even when your user wants it badly. The prompt fired, the button was one tap away, the person nodded when asked whether they cared. Nothing moved. If you have shipped a product, you have watched this. Every framework has an answer for it. Most of the answers are half right.

The framework practitioners reach for first is BJ Fogg’s. Behavior equals Motivation, Ability, and Prompt, converging at the same moment. It is elegant, it is memorable, and it has survived nearly two decades of product review because it is true. The trouble is not the equation. The trouble is that one of its three variables is doing the work of eight.

I have taught the Octalysis Framework since 2012, and every time I audit a product with a Fogg practitioner beside me the same thing happens. We agree on Ability. We agree on Prompt. Then we sit and stare at Motivation. They see one dial. I see eight. And the reason the behavior did not happen almost always lives in one of the seven dials the shared vocabulary cannot name.

This piece takes Fogg’s equation seriously and asks what it looks like when the M is decomposed into its actual dimensions. The result is Fogg with a proper vector inside the term he compressed to keep the model teachable. He was right to compress. It is now time to expand.

Speed Run Notes

• Fogg’s B=MAP is correct as a structural equation. Behavior needs all three, and without a prompt no motivation converts. That part stands.
• The word “Motivation” hides eight distinct psychological drivers. A user with high CD8 (Loss & Avoidance) and low CD2 (Development & Accomplishment) will act differently from a user with the reverse profile, even if their aggregate motivation scores the same.
• Each of the 8 Core Drives requires its own Prompt type and its own Ability path. Sending a Facilitator prompt at a CD5 (Social Influence) motive fails the same way sending a Signal at a CD8 loss motive over-fires it.
• The reformulated equation is Behavior = (CD1 · CD2 · … · CD8) · Ability · Prompt, where the specific active drives determine which Prompt type and Ability path fits.
• An audit tool at the end names the drives you are firing, the ability path each requires, and the prompt that lands each one. Use it before your next product review.
• AI does two things to this system at once. It makes Prompts effectively free, and it can flex Ability per user. Both changes concentrate the design work on which Core Drives you are actually recruiting, which is the layer generative systems cannot invent for you.

Author Credibility: Yu-kai Chou

Yu-kai Chou created the Octalysis Framework after studying gamification since 2003, years before the term entered mainstream vocabulary. As a Human-Systems Architect & Behavioral Designer, his framework has been applied by LEGO, Microsoft, Porsche, Coca-Cola, Salesforce, and MrBeast, impacting over 1.5 Billion Users.

A Product Where the Motivation Was High and the Behavior Died Anyway

A health-tech client came to me in 2024 with a stubborn problem. Their app had a daily journaling feature that users said in every survey they loved and wanted to keep. Engagement on the feature was collapsing. The team had done the Fogg audit properly. Motivation was scored high, straight from the survey data. Ability was low because the form took eight fields, so they trimmed it to three. Prompt was well timed, a morning notification at the moment users had said was best for them. All three variables green. Behavior still red.

We ran the Octalysis scan and the picture rewrote itself. The users’ stated motivation was not one motive. It was two, and they were fighting. A slice of the audience was motivated by Core Drive 1 (CD1): Epic Meaning & Calling, the belief that daily reflection made them a better person. A larger slice was motivated by Core Drive 8 (CD8): Loss & Avoidance, the fear of losing a streak. The Prompt design assumed CD2 (Development & Accomplishment), a friendly nudge toward progress. It sparked neither drive. The CD1 users needed a reason that dignified the ritual. The CD8 users needed the streak itself to be visible in the notification.

Rewriting the prompts to fork by inferred motive lifted daily completion by 38 percent inside a month. Same feature, same ability, same delivery time. The variable we had touched was the one Fogg’s equation compresses to a scalar and never asks you to look at.

§1. What Fogg Gets Right

Before decomposing the M I want to plant a flag. Fogg’s equation is one of the most useful things behavioral design has produced this century, and my synthesis rests on it holding true.

Its structural claim is that behavior requires the convergence of three ingredients at a single moment. Take any one away and the behavior does not happen. That claim survives every real-world audit I have ever run. I have never seen a behavior that occurred without a prompt of some kind, even an implicit one, and I have never seen a behavior that happened when ability was outside the user’s current budget of time, money, cognitive load, physical effort, social permission, or routine fit. Those six dimensions of ability, first articulated in Fogg’s 2009 paper A Behavior Model for Persuasive Design and refined in his 2019 book Tiny Habits, are still the cleanest checklist in the field.

The equation also carries a design instruction that I quote in workshops constantly. When a behavior does not happen, do not push motivation harder first. Reduce the ability cost first. Motivation is expensive, unstable, and mostly outside your control as a designer. Ability lives in the product and you can move it today. This priority is one of the durable gifts of the Stanford Behavior Design Lab’s body of work and it aligns with everything the actions-vs-motivations analysis teaches. Behavior is the unit. Feelings are the fuel. Ability is the pipeline. Prompt is the ignition. The equation is right at every level of that reading.

Fogg is also right that the model works across cultures, ages, and domains. That is what makes it worth building on. A model with local truth is a curiosity. A model with universal truth is a foundation. B=MAP is a foundation. Everything below is an argument for what belongs inside its M term, not a case for a different foundation.

§2. What Fogg’s M Actually Contains

Octalysis started because I refused to accept that motivation was one thing. I had spent thousands of hours inside games and could feel that the drive to grind for gear was not the drive to help my guild, which was not the drive to see what the next chest contained, which was not the drive to protect a streak I had built. Self-Determination Theory named autonomy, competence, and relatedness. Maslow named a stack of needs. Cialdini named a set of influence levers. None of them named the whole territory, and none of them named the black-hat half where behavior gets produced by scarcity, fear, and unpredictability. I mapped what I could feel in the games onto an octagon, and eight Core Drives emerged as the psychological engines behind all human motivation. That is the layer Fogg’s M compresses. Each drive requires its own kind of Prompt, its own Ability path, and its own behavior arc; treating them as interchangeable is what turns a well-designed model into a mediocre product.

CD1: Epic Meaning & Calling

Core Drive 1 moves people when they believe they are part of something bigger than themselves. The Prompt this drive wants is a mission reminder, a message that reconnects the small next action to the larger narrative. Facilitator prompts that emphasize speed feel insulting to CD1 users because speed is not why they are here. The Ability path CD1 tolerates the most friction because the meaning subsidizes effort; Wikipedia editors will teach themselves markup because the mission is worth the learning curve. The behavior arc is long, slow, and resistant to short-term optimization. Waze designed CD1 into the product by naming users “heroes of the road,” and the naming does the recruiting.

CD2: Development & Accomplishment

Core Drive 2 is the drive of visible progress and earned mastery. Its Prompt is a Signal, a clean nudge that says the next step is here and the progress bar will move if you take it. Ability has to feel challenging enough to be earned; a CD2 experience that is trivially easy stops feeling like development at all. The arc is stepwise, milestone-driven, and hungry for feedback. Duolingo’s XP counter, Fitbit’s steps target, and every leveling system in every RPG live here. This is also the drive most product teams over-fund because it is the most measurable, which is a separate problem I keep flagging.

CD3: Empowerment of Creativity & Feedback

Core Drive 3 is the drive to explore and combine, with the system responding meaningfully to what you make. Its Prompt is an invitation to play, phrased as a possibility rather than an instruction. Ability needs enough building blocks to combine and enough feedback to learn the rules of the space, which is why Minecraft’s early hours have almost no explicit guidance and still hook creative players. The arc is exploratory and non-linear, with long plateaus punctuated by breakthroughs. Prompting a CD3 user with a rigid to-do checklist reads as a demotion.

CD4: Ownership & Possession

Core Drive 4 activates once a user starts to feel that something inside the product is theirs. The Prompt this drive wants is a reminder about the asset itself, not about a task. “Your garden is thirsty” lands where “It is time to water” does not. The Ability path must protect what the user has built, because any friction that jeopardizes accumulated ownership feels like a betrayal. The behavior arc is patient and long, and it compounds; the endowment effect means the value of the possession grows with time held even without any objective change. This is why Substack writers stay on Substack past the point where the numbers argue for leaving.

CD5: Social Influence & Relatedness

Core Drive 5 moves people through what other humans think, do, and expect of them. Its Prompt is a social object: a friend just did X, your teammate is waiting, someone acknowledged you. Ability is often best delivered as a group task, because CD5 users draw effort from the group and stall alone. Strava’s 2025 numbers make this concrete: the platform crossed 14 billion kudos given for the year and 50 million monthly active users, roughly double its nearest competitor, and runners with larger Strava networks reported higher self-efficacy and exercised more consistently. The arc synchronizes with peer rhythms rather than personal ones, which is why individualist onboarding flows drop CD5 users the fastest.

CD6: Scarcity & Impatience

Core Drive 6 makes people want what they cannot easily have, and want it more the harder it is to get. Its Prompt is a scarcity signal delivered at exactly the right moment: a rationed slot, a fading window, a queue you can see. Ability has to be reachable but not effortless, because total ease neutralizes the whole drive. Repetition wears CD6 out fast, so the arc is bursty and fatigue-sensitive. Booking.com’s countdown timers work because hotel bookings are infrequent; the same countdown on a daily-use product corrodes trust within weeks.

CD7: Unpredictability & Curiosity

Core Drive 7 is engaged when a user does not know what will happen next and wants to find out. The Prompt CD7 wants is a tease, not a description. “Something is waiting” outperforms “Your daily report is ready” every time. The Ability path is usually low, because the drive itself supplies the pull; the button just needs to be there. The behavior arc is highly repeatable but heavily subject to habituation, which is why every CD7 mechanic decays and needs refreshing. TikTok’s entire product loop is CD7 delivery at a cadence engineered to postpone the decay. That decay curve is why Hook Model loops need white-hat pairing when CD7 is running the show. Unpaired CD7 delivers hits without the CD1 or CD2 payoff that keeps the user engaged for reasons they will thank the product for later.

CD8: Loss & Avoidance

Core Drive 8 moves people through the fear of losing something already held or already accumulated. The Prompt CD8 wants is a loss framing, made specific to the user’s asset. “Your 217-day streak ends in three hours” is CD8 done well; “Come back to us” is CD8 done as generic re-engagement, which fails. The Ability path here is a design trap. Making the recovery too easy defuses the loss; making it too hard turns the drive into resentment. The behavior arc is short, sharp, and expensive to repeat, which is what Duolingo’s own research on binge-versus-pace attrition (and its Weekend Amulet mitigation) has captured. Streak Freezes and Weekend Amulets exist because unmodulated CD8 kills the very users it converts.

Eight drives, eight Prompt shapes, eight Ability paths, eight arcs. A Fogg audit that scores M as one number is averaging across all of this. That is not wrong, exactly; it is just underspecified enough that the design instructions it produces do not know which drive they are talking to.

§3. Why the Scalar-Motivation Assumption Fails

The scalar assumption breaks in three specific ways I have watched over the last two years of client work. Each mode has a signature; once you know the signature you can spot the failure inside the first hour of an audit.

Mode one: the Prompt was designed for the wrong drive. A leading corporate learning platform in 2025 A/B tested two versions of its weekly reminder. Version A was “Continue your streak of 12 weeks with a 5-minute lesson,” a CD8 loss frame. Version B was “Your team lead just finished this module, join them,” a CD5 social frame. The social frame won by 46 percent among employees under 30 and lost by 22 percent among employees over 50. The M was neither wrong nor absent in either audience. It was differently constituted, and the Prompt design was implicitly betting on one composition. Fogg’s equation, as usually applied, cannot see this split.

Mode two: high stated motivation and a mismatched Ability path. The health-tech story I opened with is one instance. Another: a personal-finance client in early 2025 built an aggressive budgeting feature after surveys showed 78 percent of users “strongly wanted” better spending discipline. The M read high. The Ability path was individualist, a solo tracker with a solo dashboard. Adoption stayed flat until they added a family-share option. Half of the highly motivated users had a CD5 motivation for the behavior. They wanted the discipline with a partner, and the solo Ability path made the behavior effectively impossible for them, no matter how much they wanted it. This is the same reason group weight-loss programs outperform individual ones even when the individual programs are objectively easier.

Mode three: motivation flip during use. A meditation app I audited in 2024 had strong onboarding, weak week-two retention, and worse week-eight retention. The Fogg reading was that motivation faded and the prompts stopped landing. The Octalysis reading was that the drive shifted drives. Users came in on CD1, the meaning of a calmer mind. By week two the meaning had become abstract and CD2 progress cues were needed. By week eight the users who stayed had migrated to CD4, feeling that their practice was theirs, and were now stalling because the app treated them like beginners. Same user, three different Ms across the timeline. A single scalar hides all three transitions.

Versions of each mode appear in older literature. The COM-B Behavior Change Wheel tries to open up capability and opportunity, and the ARCS Model expands motivation into attention, relevance, confidence, and satisfaction. Both improve on the Fogg scalar, but neither carries the sharp black-hat half that Octalysis carries in CD6, CD7, and CD8. A product that only recruits white-hat drives feels wholesome and fails to convert; a product that only recruits black-hat drives converts and burns users out. A scalar M cannot tell you which side of the octagon you have staffed.

§4. The Fogg × Octalysis Synthesis

The move is straightforward: keep B=MAP as the structural equation, and require every audit to specify what is inside M. The reformulated form I now teach is:

Behavior = ( CD₁ · CD₂ · CD₃ · CD₄ · CD₅ · CD₆ · CD₇ · CD₈ )_i · Ability · Prompt

The subscript i matters. It refers to the specific activation pattern for user segment i at time i. Two products with the same aggregate M can have very different CD profiles, and two identical CD profiles can shift over time inside the same user. The equation now carries that specificity in its notation instead of hiding it inside one variable.

The multiplication is deliberate, borrowed from the Fogg convention that the three MAP terms combine multiplicatively. If any drive that a design assumes is present is actually zero, the design assumption breaks, in the same way that a zero on Ability or Prompt zeros out Behavior. This is why partial-fit designs feel “almost right” and stubbornly fail. The missing drive is absent, not merely weakened, and no amount of pushing the other terms compensates.

My Octalysis Score, defined as the sum of the squares of the 8 CDs, still applies at the aggregate level and correctly rewards concentration over spread, because one drive at 10 (100) beats eight drives at 2 (32). The reformulated equation adds the directional information the aggregate score cannot carry. Two products can score the same 100 and be entirely different products; one might be all CD8 pressure and one might be all CD1 meaning. Fogg’s equation now has to name which.

The synthesis is not adversarial. Nothing in B=MAP is deleted. Every ability and prompt heuristic Fogg has published still applies inside the reformulated model. The claim is only that Motivation is a vector rather than a scalar, and that the vector determines which specific Ability and Prompt design lands. Fogg’s own three Core Motivators (Sensation, Anticipation, Belonging) already gesture at multidimensionality; Octalysis carries the decomposition to a granularity where design instructions differentiate.

§5. Design Implications

If you used B=MAP yesterday and you use the reformulated version tomorrow, here is what changes in your workday.

You start every audit by naming the top two Core Drives you believe are active for the target behavior, not just scoring M. Two, not one; the interaction between the top two is where most of the design leverage lives, and where the failure modes hide. On a task like “log a meal in the calorie app,” the top two are usually CD8 (fear of overshooting goals) and CD2 (progress on the streak). On a task like “post a photo,” the top two are usually CD5 (social feedback) and CD4 (my curated feed).

You pick your Prompt type per drive, not per motivation level. Facilitator, Signal, and Spark, the three types Fogg named, still apply. What changes is that the mapping is now specific: Facilitator for low-friction CD2 or CD7 nudges, Signal for CD5 social objects and CD4 ownership reminders, Spark for CD1 mission reconnection and CD8 loss framing. The mismatched-prompt failure mode goes away when the Prompt design has to declare which drive it is addressing.

You audit your Ability path against the drive, not against a universal simplicity target. A CD1 experience can carry more friction because meaning subsidizes effort, which is why Wikipedia editors tolerate markup. CD7 has to be effortless at the moment of the tease, which is why TikTok’s open-app-and-scroll is a single gesture. Ownership drives like CD4 need protection above all, so any redesign that reshuffles a user’s saved artifacts is dangerous even when it “improves” the flow.

You separate White Hat and Black Hat drives inside the M term explicitly. White-hat drives (CD1, CD2, CD3, CD4, CD5) make users feel powerful and generative; they build long-term retention and low churn but rarely convert on a specific moment. Black-hat drives (CD6, CD7, CD8) make users feel urgent or anxious; they convert on the moment but burn users out if run continuously. The rhythm between them, described in more detail in the Core Drive interaction guide, is what turns a competent design into a durable one.

You add a frequency check per drive. A CD8 prompt used once a quarter feels caring; the same prompt used weekly feels like coercion, and daily feels like abuse. This is the same rule I have taught for years about Booking.com’s countdowns versus Amazon’s comfort. The reformulated equation makes frequency a per-drive setting.

You time your Sparks to the phase. Discovery and onboarding usually need CD1 sparks, not CD8. Endgame benefits from CD4 and CD5 sparks, because your veterans are motivated by ownership and community more than by beginner progress. A one-size prompt schedule keeps missing precisely because it ignores this.

You stop treating “motivation is low, so we need a stronger prompt” as a reflex. In the reformulated model, low aggregate motivation is a hint that a specific drive you assumed was present is not, and the design fix is usually to recruit a different drive. The 2024 gamification and AI field data shows what happens when teams double down on prompt intensity instead: notification fatigue rises and total engagement falls.

Finally, you build metrics per drive. A single retention number aggregates users across all eight drives and hides the migration story. Splitting week-two churn by inferred drive class is what exposed the CD1 to CD2 handoff that was breaking the meditation app’s funnel.

§6. The One-Page Audit Tool

Here is the audit I run before I write anything else on a client engagement. It takes about twenty minutes for a single target behavior. Print it, use it, mark it up. It is the operational form of the synthesis.

Step 1. Name the target behavior in one sentence. One user, one moment, one action. If you cannot write it as “When [user] [context], they will [action],” the behavior is not yet a target. This is Fogg’s discipline and I keep it.

Step 2. Score the 8 Core Drives for the target user at this moment. Not the ideal user, not the founder, not the power user. The realistic median user for the target behavior. Score each drive 0 to 10 based on how strongly it is likely to be active. Use the Octagon Builder if you want a visual output. The two highest drives are your primary M vector.

Step 3. For each of the top two drives, name the Prompt type that fits. Use this cheat sheet as a starting point, and override with observation:

• CD1: Spark, mission reconnection framing
• CD2: Signal, progress-visible nudge
• CD3: Invitation, phrased as possibility not instruction
• CD4: Signal, asset-focused (“your X needs Y”)
• CD5: Signal, social object (“friend did X”)
• CD6: Spark or Signal, timed scarcity window
• CD7: Facilitator or Spark, tease not description
• CD8: Spark, loss-specific to user asset

Step 4. Name the Ability path per drive. Not per user. Per drive. Ask what specific dimension of ability the drive tolerates most and least. CD1 tolerates cognitive friction; CD7 does not. CD5 tolerates social effort with a partner; CD5 stalls in solo tasks. CD8 requires an ability path that lets the loss be recovered but not too easily.

Step 5. Predict the behavior arc per drive. One line each. CD1 is slow and compounds. CD2 is stepwise and milestone-hungry. CD3 is exploratory. CD4 is patient and compounds. CD5 is peer-synchronized. CD6 is bursty and fatigue-sensitive. CD7 is repeatable but habituation-sensitive. CD8 is sharp, short, and expensive to repeat.

Step 6. Run the frequency check. How often will the same user encounter the same prompt shape? A daily CD6 or CD8 answer means you are over-firing a black-hat drive and will burn users. A monthly CD2 or CD5 answer means you are under-firing a white-hat drive and will lose momentum.

Step 7. Run the phase check. Which phase is the user in: Discovery, Onboarding, Scaffolding, or Endgame? Do the drives, prompts, and ability paths fit the phase? Endgame veterans do not want beginner prompts. Discovery users do not yet own anything, so CD4 recruitment is premature. This is the step I see teams skip most.

Step 8. Ship the smallest test that isolates the top drive. Not the top three; the single most active drive, with the matching Prompt type and Ability path, held for two weeks against a control. This is where Lean Startup discipline and Octalysis discipline reinforce each other cleanly.

Every audit I run in 2026 follows this eight-step sheet. Behavioral design is a small number of moves practiced until they are instincts. This is one of the moves.

§7. Where Fogg Is Still Right About the Equation

I want to be candid about what this synthesis does not change. The Fogg equation’s structure survives every criticism I have ever raised about the M term.

Ability still has to be present. If the user does not have the time, money, cognitive room, physical capacity, social permission, or routine slot, the behavior does not occur regardless of what the M vector looks like. The eight-drive decomposition changes which ability path you invest in, not whether you need one at all.

Prompt still has to fire at the right moment. A perfectly aimed CD5 social object delivered three hours after the social moment has passed is inert. Fogg’s point that well-timed prompts often matter more than raw motivational intensity holds equally for scalar and vector Motivation.

The convergence rule holds. All three factors must be present at the same time for behavior to occur. What the Octalysis decomposition does is make the M term auditable at design time, and specify per-drive Ability and Prompt requirements. The single moment of convergence is still the single moment of convergence.

The action-line logic still applies. Behaviors above the action line happen; below it, they do not. My reformulation just recognizes that the action line lives in a different place for each Core Drive. CD7 users cross the line at almost no effort because the curiosity supplies the pull. CD1 users cross the line at high effort because the meaning subsidizes it. CD8 users cross the line under pressure because the loss frame demands it. The line is still real; there is not one line, there are eight.

This is why I refuse to describe the synthesis as an overthrow. It is a specification. Fogg pointed at the phenomenon and gave designers a memorable formula that told the truth at its own resolution. Octalysis increases the resolution of one term, and the practical work of design gets easier once the M becomes something you can point at with a finger instead of a hand.

§8. The AI Era

Three years into the generative era it is clear that AI hits B=MAP unevenly, and the reformulated equation makes the unevenness legible.

AI makes Prompts nearly free. A system that can generate a personalized notification every hour, tuned to the user’s language, mood, and history, has removed the cost side of prompt design. This is the Prompt inflation problem. When prompts are cheap, users defend themselves by tuning them all out, and total behavior falls even as the number of prompts rises. Airship’s 2025 mobile-engagement benchmarks show the same pattern: push opt-in rates have slid across categories, and notification-driven session lift is weaker where teams have leaned hardest on generative prompts. The design fix is fewer prompts, each aimed at a specific active drive, which is the layer AI still cannot invent for you.

AI can flex Ability per user. A tutoring system can generate exactly the right difficulty for this student; a coding assistant can prefill enough of the file that the human only has to review; a travel planner can compress five hours of research into three clicks. This is enormous. The question the reformulated equation forces is which drives you are preserving friction on and which you are dissolving. Flatten ability on a CD2 experience and you kill the progress arc; users complete the task but do not feel they earned it, and CD2 requires the feeling of earning to build. Flatten ability on a CD7 experience and you may unlock new engagement; the drive never needed the friction. Per-drive ability tuning is the new discipline.

You can see the same tension inside the workplace. The AI-fluent employees quit first pattern I documented earlier this year is this failure at organizational scale. AI flattened the ability path on the very tasks that used to carry CD2 and CD3 for high performers. When the earned-mastery feeling and the exploratory-craft feeling both vanished, the M vector collapsed on the two drives that used to hold the best people, and they left. A Fogg-scalar reading would have said motivation dropped. The vector reading tells you exactly which drives to rebuild the job around and which to leave to AI.

AI can amplify or manufacture different drives per user. Recommender systems now infer which Core Drives a user is most responsive to and shape prompts, ability paths, and even feature presentation accordingly. A CD5 user gets social-forward framing; a CD8 user gets loss-forward framing. Done well, this is the most personalized behavioral design ever built. Done poorly, it becomes the segmenting of users into whichever drive is easiest to monetize, which is usually CD6, CD7, or CD8. The brain-hijacking discourse gets one thing right and one thing wrong. Dark patterns do exist. Users are also not passive victims; agency is still real, and the design responsibility is to fund white-hat drives at least as heavily as black-hat ones. That is a design decision, not a technology inevitability.

A good AI-era behavior design pass looks like this. Score the CD vector for the target user segment. Have the AI generate a small number of Prompt candidates per drive. Personalize the Ability path by drive-tolerance rather than demographic. Cap black-hat frequency at levels the pre-AI design would have accepted. Measure retention by drive class. Keep the human designer at the layer where drives are chosen, because that layer requires an ethical read AI cannot manufacture on its own.

The old joke about points, badges, and leaderboards being “only the outer shell” of gamification is the same joke now about AI-generated prompts. Prompts are the shell. The M vector is the engine. Building the shell without knowing which drives it is talking to is where the current AI-era work is spending most of its money and getting most of its disappointment.

Closing

I owe Fogg a debt. Before B=MAP the field talked about behavior as though it were a single messy variable. His equation made it structural, and it made behavioral design a discipline product teams could adopt without a psychology degree. The compression of Motivation to a scalar was the price of that adoption. Without it the model would not have spread.

Two decades later the field can afford the vector. Product teams have the tools, the data, and the interpretive capacity to specify which of the 8 Core Drives their design is recruiting. The reformulated equation asks only that the M term become a checklist rather than a slider. Once that shift lands, the design instructions get sharper, the failure modes stop hiding, and the AI-era temptation to inflate Prompts and dissolve Ability meets a disciplined counter-argument at the drive level.

If you take one thing from this piece, take this. The next time a Fogg audit tells you a behavior failed because “motivation was low,” do not raise the motivation. Ask which of the 8 Core Drives was actually being recruited, and whether you have designed a Prompt and an Ability path that fit that specific drive at this specific moment for this specific user. Nine times out of ten the drive you were counting on was not the drive that showed up, and the fix is a redesign at the vector layer, not a louder shout at the scalar. That is where behavior lives now, and where the good work is.

I keep a running library of every framework I meet in the wild, mapped onto Octalysis, in the Behavioral Framework Library. The full Octalysis treatment lives in Actionable Gamification and the newer 10,000 Hours of Play. If you want to score your own design against the 8 drives before your next standup, the Octagon Builder takes about eight minutes. Bring your audit back to me on Discord.

Related Reading

• Hook Model vs Octalysis: Why Hooks Create Addicts, Not Habits. The related BJ Fogg × Nir Eyal lineage read through the same vector lens.
• Actions vs Motivations in Behavioral Frameworks. The predecessor essay to this one on why single-motivation models keep missing.
• COM-B Behavior Change Wheel. Michie’s extension of Fogg that expands capability and opportunity but keeps motivation flat.
• ARCS Model of Motivation. Keller’s four-part motivation split, mapped onto the 8 Core Drives.
• MINDSPACE Framework. The UK government’s behavior-change checklist, benchmarked against the Octalysis vector.
• Gamification and AI: Motivation Design for the AI Age. Where the AI-era section of this piece is grounded.

WOULD YOU LIKE YU-KAI CHOU TO WORK WITH YOUR ORGANIZATION?