Precaution Adoption: S-Tier Behavioral Designer’s Guide

In 1988, a Rutgers psychologist named Neil Weinstein published a paper that quietly broke one of the most expensive assumptions in public health: that everyone who has not yet acted on a health warning is in the same psychological state. Three decades later, governments running vaccination drives and Fortune 500 marketers running multi-million-dollar campaigns are still making that same mistake, and the brochures, push notifications, and ad creatives they ship to “non-adopters” are still failing for the reasons Weinstein named.

The fix is not bigger budgets or louder copy. The fix is admitting that “hasn’t done it yet” is six different psychological states, that each one needs a different message, and that aiming the wrong message at the wrong stage doesn’t just waste impressions — it can push someone backward.

That insight became the Precaution Adoption Process Model. It’s the most under-rated stage-based behavior-change framework in the field, partly because it sits in the shadow of Prochaska’s better-known Transtheoretical Model, and partly because its most important contribution — Stage 4, “Decided NOT to Act” — is the one most campaigners would rather pretend doesn’t exist.

This pillar takes you through all seven stages, the stage-matched intervention thesis Weinstein built around them, the brain science underneath each transition, the places PAPM falls apart, and the design protocol I run on every behavior-change brief where the audience isn’t a monolith. By the end you’ll have an audit you can run on any campaign you’re shipping next quarter, and a Stage 4 re-engagement playbook you almost certainly don’t have today.

Speed Run Notes

• Most behavior-change campaigns fail because they broadcast one message to an audience that is actually in six different psychological stages. PAPM is the stage map.
• The seven stages: Unaware, Unengaged, Deciding, Decided NOT to Act, Decided to Act, Acting, Maintenance. Stage 4 is a terminal off-ramp, not a pause.
• Stage-matched intervention is the thesis: the variable that moves someone from Stage 2 to Stage 3 is qualitatively different from the variable that moves Stage 5 to Stage 6. Same message, wrong stage, wasted reach.
• PAPM separates “doesn’t know” from “knows but isn’t engaged.” TTM lumps them into Precontemplation, which is why TTM-tailored radon and vaccination campaigns underperform PAPM-tailored ones.
• Octalysis crosswalk: Stage 1-2 runs on Curiosity. Stage 2-3 on personalized Loss + Ownership. Stage 3-5 on Social Influence. Stage 5-6 on Scarcity plus friction-removal. Wrong Drive, wrong stage, no movement.
• The Stage 4 playbook nobody runs: people who decided NOT to act need a different message than the undecided. Treating them as “still deciding” is the most common Stage 4 mistake in field campaigns.

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.

Chou has taught the Octalysis methodology at Harvard, Stanford, Yale, Tesla, Google, BCG, and IDEO.

His work has been cited by Harvard, Stanford, MIT, Forbes, Wall Street Journal, Wired, US Department of Energy, NIST, NSF, NCBI, US Department of Education, ClinicalTrials.gov, and Google Scholar — with 3,700+ more academic publications. Explore his books here.

What Is the Precaution Adoption Process Model?

The Precaution Adoption Process Model (PAPM) is a seven-stage map of how a person moves from total ignorance about a health risk or behavior to fully integrated, sustained action on it. It was developed by Neil Weinstein and colleagues at Rutgers in the late 1980s and refined across a series of empirical studies through the 1990s, with the canonical extended treatment appearing in Weinstein, Sandman & Blalock’s 2008 chapter in Health Behavior and Health Education.

The core claim is structural, not motivational. PAPM does not tell you what makes someone want to do something. It tells you what stage they’re in right now, what the next stage looks like, and what kind of input will actually move them. The model assumes that the psychological barrier between any two adjacent stages is qualitatively different from the barrier between any other two stages, which means that the intervention that worked at one transition can backfire at another.

That single assumption is the whole reason the model exists. Most behavior-change campaigns, from public-health vaccination drives to SaaS onboarding emails, treat the people they’re trying to convert as one homogenous “non-adopter” segment and broadcast one persuasion message. PAPM says that approach is hitting Stage 1 readers with Stage 5 cues, Stage 4 readers with Stage 3 cues, and Stage 6 readers with Stage 7 cues — and the conversion data should look exactly like that mismatch predicts.

The original empirical anchor was radon. In the late 1980s the United States Environmental Protection Agency (EPA) was running a national radon-testing campaign that was, on paper, well-funded and competently designed. It was also failing. Weinstein and his colleague Peter Sandman ran a series of studies on New Jersey homeowners and found something the EPA’s planners had missed: the homeowners who hadn’t tested their homes weren’t all “not yet convinced.” Some hadn’t heard of radon at all. Some had heard of it but had never connected it to their own house. Some had thought about testing and concluded it wasn’t worth the trouble. Some were going to test next month. Treating those four groups with the same brochure was the proximate cause of the campaign’s underperformance.

When Weinstein and Sandman ran the experiment that became the field’s load-bearing citation — the 1992 Health Psychology paper that gave the model its name — they tested whether stage-matched brochures outperformed generic ones. Brochures tailored to each stage roughly doubled the rate of radon test-kit purchase relative to the EPA’s one-size-fits-all version. That result became the template every PAPM application has copied since.

The Seven Stages, In Order

The model’s structure is a directed graph with seven nodes and one terminal branch. Most accounts present the stages as a linear sequence, but the actual model has an off-ramp at Stage 4 that linear diagrams hide. Reading the stages in order, the underlying psychology of each one looks like this.

Stage 1: Unaware of the Issue

The person has not heard of the risk or the behavior. The marketer’s instinct is to skip past this stage, because surely everyone has heard of X by now. The data almost never supports that instinct. In the original radon studies, roughly 30% of New Jersey homeowners hadn’t heard of radon at all, and in modern audience-segmentation work the “unaware” share is rarely below 10% for any behavior more obscure than seatbelts and handwashing.

The intervention at this stage is awareness. Not persuasion, not benefits, not call-to-action language. Awareness. The person at Stage 1 cannot evaluate benefits because they don’t yet know the category exists, so any message that assumes prior knowledge (“Have you tested your home for radon yet?”) simply reads as background noise.

Stage 2: Unengaged by the Issue

The person has heard of the issue but has not personally engaged with it. They could define radon if asked, or they could state that their HPV vaccination is overdue, but the issue lives in the abstract — it’s something other people deal with. This is the stage where the EPA’s well-meaning brochures landed in 1989 and where most disease-awareness campaigns still land today.

The intervention at this stage is personalization. The person needs to move the issue from the category of “things that exist” to the category of “things that exist for me.” That move is psychological, not informational. Adding more facts does not work; what works is a personal-risk anchor — a neighbor’s positive test result, a zip-code-specific risk map, a family-history calculator that returns a number with their name attached. The shift from Stage 2 to Stage 3 is the shift from “this is real” to “this is real for me.”

Stage 3: Deciding About Acting

The person is actively weighing the decision. They are engaged. They are doing the cost-benefit math. They are reading reviews, asking friends, looking at their schedule. This is the stage where the Health Belief Model variables (perceived susceptibility, perceived severity, perceived benefits, perceived barriers, cues to action) finally start to do their predictive work. Before Stage 3, those variables are inert because the person isn’t running the calculation yet.

The intervention at this stage is decision-support. The Stage 3 person doesn’t need more awareness or more personalization; they need help with the deliberation. Comparison tables, removed barriers, social proof from peers who have already decided, and a clear sense of what the next action would actually involve. This is also the most fragile stage, because every Stage 3 deliberation has two possible exits — Stage 4 (decided not to act) or Stage 5 (decided to act). Stage 4 is permanent in PAPM’s strict reading.

Stage 4: Decided Not to Act

This is the stage every other behavior-change model under-theorizes. The person has weighed the decision and concluded that the precaution is not for them. Maybe they decided the risk isn’t real. Maybe they decided it’s real but not large enough to act on. Maybe they decided the cost is too high. Maybe they ran into a barrier they couldn’t get past, accepted the barrier, and stopped thinking about it.

Critically, the Stage 4 person is not the same as the Stage 2 person. The Stage 2 person hasn’t deliberated yet. The Stage 4 person has, and has finished the deliberation in the “no” direction. The psychological consequence is that the Stage 4 person now has a sunk-cost defense around the decision — they’ve already publicly or privately committed to not acting, and re-opening the decision feels like admitting they were wrong before. Treating a Stage 4 audience the same way you treat a Stage 2 audience is one of the single most common mistakes in vaccination outreach, where “vaccine-hesitant” populations are routinely re-targeted with the same awareness-and-personalization content that already failed once.

The intervention at Stage 4 is qualitatively different from any earlier stage. The Stage 4 person needs a face-saving on-ramp back to deliberation — usually a change in the underlying facts (“the new variant changes the calculation”), a change in their personal circumstances (“now that you’ve had a child”), or a change in the cost-side of the decision (“the test is now free”). Pushing a Stage 4 person without one of these external reframes typically deepens the Stage 4 commitment via reactance.

Stage 5: Decided to Act

The person has formed the intention. They are going to test for radon, schedule the vaccination, sign up for the program. They have not done it yet. The gap between Stage 5 and Stage 6 is the famous intention-behavior gap, and Sheeran’s 2002 meta-analysis put the average size of that gap at roughly 47% — that is, only about half of people who form a clear behavioral intention actually carry it out, even when they sincerely mean to.

The intervention at this stage is friction-reduction and timing. The person doesn’t need more persuasion; the persuasion already worked. They need the action to be easy, scheduled, and unavoidable. Implementation intentions (Gollwitzer’s “if-then” planning), default scheduling, default appointment booking, calendar invites, “do this within 24 hours and we’ll waive the fee” deadlines — these are the Stage 5 levers. Persuasion content here is wasted; it converts a Stage 5 person back into a Stage 3 person by inviting them to re-deliberate a decision they’ve already made.

Stage 6: Acting

The person has performed the behavior. This is the stage everyone celebrates as if it’s the finish line, and for one-time precautions (radon test, single-shot vaccination) it largely is. But for any behavior that requires repetition (annual mammograms, daily medication, recurring screenings, ongoing dietary change), Stage 6 is the temporary state in which the behavior has been done once and is at risk of not being done again.

The intervention at this stage is reinforcement and habit-scaffolding. The person needs evidence that the behavior worked, social acknowledgment that they’re now in the in-group of doers, and structural supports that make the second performance easier than the first. This is also the stage where the original PAPM model is least specified, because Weinstein’s primary cases were one-time precautions where Stage 6 was effectively terminal. For repeated behaviors, the model needs supplementation from the habit literature.

Stage 7: Maintenance

The behavior has been integrated into the person’s life. They have done it enough times that the cognitive effort of deciding to do it has dropped to near zero. The behavior is now a habit, and the relevant brain machinery has shifted from the deliberative prefrontal-cortex circuits to the dorsolateral striatum’s habit network (Graybiel 2008; Wood & Neal 2007).

The intervention at this stage is protection of the existing habit. The Stage 7 person doesn’t need persuasion or scheduling; they need their existing routine not to be disrupted. The classic Stage 7 mistake is sending the same “have you tested for radon yet?” reminder to someone who has tested twice already, which signals that the campaign doesn’t know who its own users are and produces a Core Drive 8 (CD8): Loss & Avoidance reaction — the person becomes more likely to disengage from future communications.

The Stage-Matched Intervention Thesis

If the seven stages are the model’s anatomy, the stage-matched intervention thesis is its physiology. Weinstein’s claim is that the relationship between intervention content and stage transition is not linear or additive. A given intervention is uniquely effective at one specific transition and approximately useless — or counterproductive — at all the others.

The cleanest articulation of this came in Weinstein, Lyon, Sandman & Cuite’s 1998 paper in Health Psychology, which ran an experimental test on home radon testing. Participants were sorted by PAPM stage and then randomly assigned to receive one of three brochure types: a Stage 1→2 brochure (general awareness), a Stage 2→3 brochure (personalized risk), or a Stage 3→5 brochure (decision support). The result was that each brochure significantly outperformed the others at moving people through its targeted transition, and underperformed at the other two transitions. The interaction was not subtle.

This is the result that grounds the entire PAPM apparatus. Without stage-matched advantage, the seven stages would be a useful taxonomy but not a usable model. With stage-matched advantage, every behavior-change campaign has a hard ceiling determined by how accurately it can identify which stage its audience is in, and how cleanly it can route stage-specific messages to stage-specific recipients.

The implication for practitioners is that the right question is not “is our message persuasive?” The right question is “what stage is each segment of our audience in, and is the message they’re receiving the one that moves their stage?” That question almost no campaign answers, which is why almost no campaign captures the full ceiling.

What Weinstein Got Right

The first thing PAPM gets right that competing models do not is the cleavage between unaware and unengaged. The Transtheoretical Model (TTM), which is older, more cited, and dominant in the smoking-cessation and exercise literatures, lumps Weinstein’s Stage 1 and Stage 2 into a single “Precontemplation” stage. That collapse is harmless if your audience already knows the issue exists, which is roughly true for smoking and exercise in the populations TTM was built for. It is catastrophic if your audience contains a meaningful share of people who haven’t heard of the issue at all, which is true for nearly every novel-precaution campaign ever run — new vaccines, new screenings, new climate-mitigation behaviors, new financial products, new platforms.

The second thing PAPM gets right is Stage 4. Every other major behavior-change model treats “not yet acting” as a single state that can be moved through persuasion. PAPM is the only model that takes seriously the possibility that some non-actors have decided, deliberately and stably, that the precaution is not for them, and that this decision changes the psychological substrate the next intervention has to work with. The Stage 4 distinction is the basis for the “vaccine-hesitant rather than vaccine-uninformed” reframe that came to dominate post-2021 immunization research, and it’s the basis for almost every successful Stage 4 reactivation case study in the field.

The third thing PAPM gets right is the qualitative-rather-than-continuous nature of stage transitions. The Health Belief Model treats perceived susceptibility, severity, benefits, and barriers as continuous variables that combine additively to predict behavior. PAPM doesn’t deny that those variables matter, but it claims they don’t matter equally at every stage. Susceptibility moves Stage 2 to Stage 3. It does almost nothing at Stage 5 to Stage 6, where the relevant variable is friction. The shift from continuous-variable models to stage-conditional-variable models is the single most important advance in the precaution literature, and PAPM was the model that made it cleanly.

Where PAPM Falls Apart

PAPM is also wrong in interesting and specific ways. A pillar that doesn’t say so is a brochure, not a guide.

1. It Was Built for Single-Decision Precautions, Not Cycling Behaviors

Weinstein’s original cases — radon testing, mammography, osteoporosis prevention, hepatitis B vaccination — are all behaviors where one decision largely determines the outcome. You test once, get vaccinated once, or start a single screening regimen. The model handles those behaviors crisply. But many of the behaviors public-health practitioners actually need to change are oscillating: smoking, dieting, exercise, alcohol consumption. Those behaviors don’t have a clean “decided to act / now acting” line because the person decides, acts, lapses, decides again, acts, lapses, and so on across years.

The Transtheoretical Model handles this loop more naturally by including an explicit relapse mechanism that returns the person to an earlier stage. PAPM’s Stage 4 is terminal in its strict reading, which means PAPM can’t natively represent the smoker who decided not to quit in 2015, decided to quit in 2019, quit for a year, and started again in 2022. The honest answer is that PAPM is the better model for one-shot precautions and TTM is the better model for cycling behaviors, and a campaign that needs to handle both should not commit to either model alone.

2. Stage Assessment Is Harder Than the Model Pretends

To stage-match an intervention, you have to know what stage each person is in. The PAPM literature includes a standard stage-classification algorithm that uses self-report items, and the reliability of that algorithm is fine at the extremes (Stage 1 versus Stage 7 is easy) and weak at the middles. The hardest distinction is Stage 2 versus Stage 3, both of which can be self-described as “thinking about it,” and the second-hardest is Stage 3 versus Stage 5, since people in active deliberation often describe themselves as “going to” do something they haven’t yet committed to.

This isn’t just a measurement nuisance. Stage-misclassification rates above roughly 20% are enough to wipe out the stage-matched advantage in field studies, because the messages get routed to the wrong people. Sniehotta and colleagues’ work on stage-validity has made this point repeatedly: the model’s predictions are sharper than its instruments. If you’re building a stage-targeted intervention, your first design decision is the classifier, not the message.

3. The Meta-Analytic Evidence Is Modest, Not Decisive

The Weinstein-Lyon-Sandman-Cuite 1998 radon study is one of the cleanest demonstrations of stage-matched advantage in the literature, but it’s also one of a small number. Lippke and colleagues’ 2009 work on stage-matched physical activity interventions showed a modest stage-matched effect; Bridle and colleagues’ 2005 meta-analysis on TTM (which makes analogous claims) was famously tepid. The honest summary is that stage-matched interventions reliably outperform single-message broadcasts when the classifier is good and the stage differences are large, and they do not reliably outperform when the classifier is noisy or the stages are close together.

Weinstein himself, in the 2008 book chapter, is careful to describe PAPM as a “process description” rather than a prescriptive intervention theory. The seven stages are well-supported as a typology of where people are. The claim that you can move them by stage-matched messages is well-supported in controlled conditions and only partially supported in messy field conditions.

What’s Really Happening Inside the Brain

The seven stages map cleanly onto a handful of well-characterized neural circuits, and the map makes the model harder to dismiss as a tidy psychological just-so story.

The Stage 1 to Stage 2 transition is, neurally, a novelty-detection event. The brain’s ventral attention network (centered on the temporo-parietal junction and the ventral frontal cortex) tags the new information as worth orienting toward, and the locus coeruleus’s noradrenergic burst marks the moment of “wait, what’s that?” Without that orienting response, the information slides past in a habituated background state and the person stays at Stage 1.

The Stage 2 to Stage 3 transition is a salience-and-self-relevance event. The amygdala flags the issue as personally threatening, and the medial prefrontal cortex re-encodes the issue from “exists” to “applies to me.” The brain machinery is the same machinery the self-referential processing literature has documented for autobiographical memory: when a stimulus is encoded as self-relevant, the medial prefrontal cortex shows reliably stronger activation, and the encoding is deeper and more durable. Stage 2 generic messages don’t engage that circuit because they don’t trigger the self-relevance encoding. Stage 3 personalized messages do.

The Stage 3 deliberation runs through the ventromedial prefrontal cortex and the orbitofrontal cortex, which together do the value-integration math that turns “X has this benefit, Y has this cost” into a single comparable utility. The same circuits handle every other cost-benefit decision the person makes — whether to buy a car, take a job, accept a date. The Stage 3 to Stage 5 transition fires when the integrated value crosses a threshold. The Stage 3 to Stage 4 branch fires when the integrated value lands clearly below it and the person commits to the negative decision.

The Stage 5 to Stage 6 transition is the most-studied neural event in the precaution literature, because the intention-behavior gap is itself a research field. The relevant circuits are the dorsolateral prefrontal cortex (which sustains the intention across delay) and the basal ganglia’s action-initiation network (which converts the intention into an actual motor plan at the moment the behavior is performed). Sheeran & Webb’s 2016 update on the intention-behavior gap traces the failure modes neurally: forgotten intentions, intentions that lost their salience under competing demands, intentions overwhelmed by habit reflexes that pull the person toward the older non-action default.

The Stage 6 to Stage 7 transition is the classic habit-formation arc, and the relevant literature is Graybiel’s work on the dorsolateral striatum and the chunking of action sequences. As the behavior is repeated, the cognitive control required to perform it drops from prefrontal to striatal, the metabolic cost of performance decreases, and the behavior becomes resistant to disruption. This is why Stage 7 maintenance interventions need to protect the routine, not motivate the actor — the actor is barely involved anymore.

PAPM vs Other Theories

PAPM vs the Transtheoretical Model (Prochaska & DiClemente)

TTM has five stages (Precontemplation, Contemplation, Preparation, Action, Maintenance) plus an explicit relapse mechanism that returns the person to an earlier stage. PAPM has seven stages and no relapse loop. The two models cover overlapping territory but were optimized for different problems. TTM was built on smoking cessation, where the relapse loop is the dominant feature of the data. PAPM was built on radon testing, where relapse is almost never the operative concern.

The cleanest contrast is at the front end. TTM’s Precontemplation is one stage; PAPM’s Stages 1, 2, and 4 are three. If you’re working on a behavior where the bulk of your audience already knows the issue but has heterogeneous reasons for not acting, PAPM is the better-resolved map. If you’re working on a behavior where the bulk of your audience is in some kind of cycling lapse pattern, TTM is the better fit. The post-2010 consensus in the stage-theory literature is that the two models should not be treated as rivals but as differently-tuned instruments — Glanz, Rimer & Viswanath’s Health Behavior Theory presents both as canonical and notes the complementarity explicitly.

PAPM vs the Health Belief Model (Rosenstock)

The Health Belief Model is older (1950s origin), simpler, and not staged. It treats four perceived variables — susceptibility, severity, benefits, barriers — plus cues to action as continuous predictors of behavior. PAPM doesn’t replace those variables. It puts them inside a stage architecture and claims they’re operative at different stages: susceptibility and severity dominate the Stage 2→3 transition, benefits and barriers dominate the Stage 3→5 transition, cues to action dominate the Stage 5→6 transition, and none of the original HBM variables really speak to Stage 7 maintenance at all.

The practical implication is that HBM-only campaigns over-index on susceptibility and severity messages, which are powerful in the right stage and noise in the wrong stage. A PAPM-overlay on an HBM campaign tells you when to use which HBM lever and when not to. We covered HBM in depth in the Health Belief Model pillar; this one extends that material rather than replacing it.

PAPM vs the Health Action Process Approach (Schwarzer)

HAPA, developed by Ralf Schwarzer in the 1990s, is the closest direct competitor to PAPM in the stage-theory literature. HAPA compresses PAPM’s seven stages into two broad phases — a motivational phase (deciding to act) and a volitional phase (planning and acting on the intention) — and adds continuous variables for outcome expectancies, self-efficacy, action planning, and coping planning. The HAPA literature argues that the binary motivational/volitional split is the empirically supported stage boundary and that PAPM’s seven-stage carving is more granular than the data justifies.

The HAPA-PAPM debate is not fully resolved, but the position that most stage-theorists have converged on is that HAPA’s binary phases describe the largest, cleanest empirical boundary, while PAPM’s seven-stage granularity is more useful for prescribing specific interventions. In practice, our HAPA pillar (post 46083) and this PAPM pillar are sister maps of the same territory at different resolutions, and a careful campaign uses HAPA’s binary phases for high-level targeting and PAPM’s seven stages for message design.

PAPM vs the Theory of Planned Behavior (Ajzen)

Ajzen’s Theory of Planned Behavior predicts behavioral intention as a function of attitudes, subjective norms, and perceived behavioral control, and then treats intention as the proximate predictor of behavior. TPB does extremely well at predicting where someone is at the Stage 5 line and almost nothing else — it is silent on Stage 1, Stage 2, and Stage 4, and it doesn’t engage the intention-behavior gap that dominates Stage 5 to Stage 6. PAPM’s relationship to TPB is that PAPM tracks the journey before and after the formation of intention, while TPB owns the moment of intention formation itself. The two are complements, not rivals.

PAPM in the Real World

Vaccination Programs

The 2021 to 2023 vaccination communications literature is the largest natural experiment in PAPM-style segmentation that has ever been run. By mid-2021, “vaccine-hesitant” had become a load-bearing category in public-health planning, and the campaigns that worked at scale (the United Kingdom’s NHS targeted-segment outreach, the United States CDC’s “We Can Do This” campaign in its later phases, Israel’s Maccabi Healthcare Services automated stage-classification calling) all converged on PAPM-style segmentation even when they didn’t cite the model. Stage 1 audiences (those unaware of new variants or new boosters) got pure-information messaging. Stage 2 audiences got geographically and demographically personalized risk content. Stage 3 audiences got decision-support content with peer-comparison data. Stage 4 audiences — the genuinely hesitant rather than the uninformed — got messages keyed to changed circumstances, primarily through trusted community messengers rather than centralized media. Stage 5 audiences got friction-reduction content, including default appointment booking and on-site clinic availability. Stage 6 and Stage 7 audiences got annual-reminder protection, not persuasion.

The lesson the field took from this period is that “the message” is the wrong unit of analysis. The right unit is the stage-message pair, and the campaigns that segmented at the stage level outperformed campaigns that segmented at the demographic level by margins large enough to detect in administrative data.

Disaster Preparedness

Hurricane preparedness research is the second-richest PAPM application area. Bostrom and colleagues’ 2018 work on coastal-zone evacuation behavior used a PAPM-derived stage classifier to segment residents into pre-decision, deciding, decided-to-evacuate, and evacuating populations. The intervention that worked at scale was a stage-classified text message system that sent different content to different stages during the warning period — initial-warning content to the unaware, personalized-risk content to the unengaged, decision-support content (specific destinations, specific routes, specific check-in points) to those deciding, and friction-removal content (free shuttles, hotel-voucher pre-confirmation) to those who had decided to evacuate but hadn’t yet left.

The before-after comparison was striking: stage-matched warning systems produced earlier evacuation by an average of roughly 8 hours per household compared to the broadcast warning systems that preceded them, which translates directly to capacity at evacuation shelters and survival probability in the highest-risk zones.

Software-as-a-Service Conversion Funnels

Outside health, the cleanest applied use of PAPM is in SaaS conversion analytics. The standard SaaS funnel (Awareness, Consideration, Trial, Purchase, Activation, Retention) is a relabeled PAPM staircase, and the SaaS practitioners who realize this and route stage-specific messages outperform the practitioners who broadcast one onboarding email to every signup. Activation campaigns at companies like Duolingo, Notion, and Wikipedia (yes, even Wikipedia runs PAPM-style fundraising-donor segmentation) are PAPM at its most legible, even when the product teams have never read Weinstein.

Climate-Action Interventions

The climate-behavior literature has been late to adopt stage-targeting, in part because the climate-action audience has been treated as a single mass for the first three decades of the movement. The newer work, especially the climate-communication studies coming out of Yale’s Program on Climate Change Communication and George Mason’s Center for Climate Change Communication, has shown that the “Six Americas” segmentation maps closely onto PAPM stages, and that interventions tailored to the segment significantly outperform broadcast climate messaging. The lesson is the same lesson the radon literature taught in 1992: stage-matched interventions are not optional ornaments. They are the active ingredient.

The Elephant in the Room

The seven stages don’t care whether the behavior you’re moving someone toward is good for them. Stage-matched intervention is a technology, and the same technology that doubled radon-test uptake in 1992 can be used to double credit-card sign-ups, raise smoking initiation rates among teenagers, or move a Stage 4 climate skeptic deeper into the Stage 4 position by exploiting the reactance pattern instead of avoiding it.

This is the unspoken tension at the center of the PAPM literature, and Weinstein himself addressed it only obliquely. The model was published into a public-health environment where the implicit assumption was that the precaution in question was unambiguously the right one. Radon kills people. Vaccines save lives. The implicit ethical framework was that any technology that increased uptake of those behaviors was therefore ethical.

That assumption doesn’t survive contact with private-sector deployment. The same model that segments hesitant vaccine adopters into Stage 4 versus Stage 2 can segment hesitant car buyers, hesitant cryptocurrency purchasers, or hesitant gun owners. The Stage 4 face-saving on-ramp that works for vaccination (“the new variant changes the calculation”) works for crypto (“the new regulations change the calculation”). The model is morally neutral; the only thing that’s not neutral is the campaign’s choice of what behavior to push.

The honest design protocol is to apply the publicity test that Thaler & Sunstein articulated for libertarian paternalism (we covered this in the Libertarian Paternalism pillar). If your stage-segmentation strategy is one you’d be willing to publish in a press release, fully named and stage-keyed, then ship it. If it’s the kind of strategy that would generate a backlash if a journalist explained the back-stage to your target audience, you are running Black Hat Octalysis at a level you should not be running.

The behavioral-design community talks about Black Hat and White Hat Core Drives as if the only ethical concern is the Drive being used. The PAPM literature reveals a second concern: the stage being targeted matters too. Pushing a Stage 4 audience aggressively is a more invasive act than pushing a Stage 3 audience, because the Stage 4 audience has already done the deliberation and you are now overriding their conclusion. A campaign that segments by stage and treats the stages with equal force is using more pressure on its Stage 4 audience than the audience consented to, even if the underlying Core Drive looks White Hat.

How to Apply PAPM with the Octalysis Framework

Below is the Octalysis Framework with Game Techniques diagram — the eight Core Drives arranged around the octagon, each Drive paired with the techniques that activate it. This is the master reference for everything that follows.

The unique contribution this pillar makes is a per-stage Octalysis prescription. Stage-matched intervention is the model’s central thesis, but the model itself doesn’t say which lever to pull at each stage. Octalysis does. Reading the stage-to-Core-Drive map below is the only crosswalk in print that lines up Weinstein’s seven stages with the eight Drives that move them.

Stage 1 → Stage 2 (Unaware → Unengaged): Core Drive 7 + Core Drive 1

The dominant Drive at the front end of the funnel is Core Drive 7 (CD7): Unpredictability & Curiosity. The Stage 1 audience hasn’t heard of the issue yet, which means the information has to function as a curiosity hook before it functions as anything else. Headlines that promise an answer to a question the reader didn’t know they had outperform headlines that announce facts they already think they know. Secondary Drive: Core Drive 1 (CD1): Epic Meaning & Calling, which provides the why-this-matters context once the curiosity hook has caught attention. The mistake at Stage 1 is to lead with CD1 (epic-meaning-first messaging) before CD7 has done its work; readers don’t engage with epic meaning when the underlying issue is still pre-attentive.

Stage 2 → Stage 3 (Unengaged → Deciding): Core Drive 8 + Core Drive 4

The dominant Drive here is Core Drive 8 (CD8): Loss & Avoidance, but only in its personalized form. Generic loss framing (“radon causes lung cancer”) doesn’t move Stage 2 readers; what moves them is loss attached to their identity (“homes in your zip code show elevated radon levels”). The personalization mechanic is Core Drive 4 (CD4): Ownership & Possession — the person needs to feel that the risk is theirs, not a generic public-health problem. Risk calculators, zip-code maps, family-history tools, neighbor case studies. The intervention endows the person with a personalized risk score they then own and can’t easily disown.

Stage 3 → Stage 5 (Deciding → Decided to Act): Core Drive 5 + Core Drive 2

The dominant Drive at the deliberation stage is Core Drive 5 (CD5): Social Influence & Relatedness, in its peer-comparison form. The Stage 3 deliberator is doing cost-benefit math under uncertainty, and the highest-impact uncertainty-reducer is evidence that their reference group has already decided. Secondary Drive: Core Drive 2 (CD2): Development & Accomplishment, which manages the self-efficacy question (“can I actually do this?”). Testimonials from people the reader sees as similar, completion-rate dashboards, peer-pressure-without-shame norm cues. The mistake at Stage 3 is to lean on CD8 instead of CD5; CD8 worked at the Stage 2 to Stage 3 transition and is now mostly noise.

Stage 3 → Stage 4 Off-Ramp Prevention

The Stage 4 off-ramp is the highest-risk transition in the model, because it’s terminal. The prevention play is to remove the reasons a Stage 3 deliberator would conclude “no.” Most often that means reducing perceived barriers (cost, time, complexity) before reducing perceived benefits. This is also where sludge-removal (covered in the Sludge pillar) does its highest-yield work — friction at Stage 3 doesn’t just slow conversion, it actively converts undecideds into decided-not-to-act through the path-of-least-resistance reflex.

Stage 5 → Stage 6 (Decided to Act → Acting): Core Drive 6 + reduce friction

The dominant Drive at the intention-behavior gap is Core Drive 6 (CD6): Scarcity & Impatience. The person has decided. They are now in danger of indefinitely delaying. The lever is a deadline, a window, a “do it now” mechanic. Implementation intentions that pre-commit to a specific moment (“right after my dentist appointment Friday”), expiring offers, scheduled appointment defaults. The simultaneous lever is friction-removal: every step removed from the action path is worth more at Stage 5 than at any earlier stage, because every step is now a potential indefinite-delay anchor.

Stage 6 → Stage 7 (Acting → Maintenance): Core Drive 4 + Core Drive 1

The dominant Drive at the maintenance transition is again Core Drive 4 (CD4): Ownership & Possession, but in a different form than at Stage 2. At Stage 2, CD4 endowed the person with a risk score. At Stage 7, CD4 endows them with an identity (“I’m someone who tests for radon”). Reinforcing the identity reinforces the behavior, because behavior consistent with identity is easier to repeat than behavior that requires a fresh decision. Secondary Drive: Core Drive 1 (CD1): Epic Meaning & Calling — reconnecting the now-habitual behavior to the larger story keeps it from drifting into rote and getting dropped.

The 6-Step PAPM × Octalysis Audit

This is the audit I run before signing off on any stage-targeted campaign. It’s the protocol that puts the per-stage Octalysis crosswalk into practice.

1. Segment by stage, not by demographic. Before any message design, build a stage classifier. The classifier can be a five-question survey, a behavioral inference from past actions, or a self-segmenter (“Which best describes you right now?”). Without a classifier you don’t have a stage-targeted campaign; you have a guess.
2. Estimate the stage distribution. Run the classifier on a sample. Most campaigns are surprised by how heavily their audience is concentrated at one stage (often Stage 2, sometimes Stage 4). The distribution determines where to spend.
3. Map each stage to its dominant Core Drive. Use the per-stage table above as the starting point. Adjust for behavior-specific deviations (some behaviors have a CD8 Stage 5 that overrides the default CD6).
4. Build one message variant per stage. Not one variant per demographic, one variant per stage. The variants will look surprisingly different from each other, which is the point.
5. Route stage-matched messages through stage-classified channels. The route matters as much as the message. Stage 4 audiences don’t trust centralized media; route Stage 4 content through community messengers. Stage 5 audiences need friction-removal channels; route Stage 5 content through one-click-action surfaces.
6. Run the publicity test on every variant. Before shipping, ask whether you would be willing to publish the full stage-to-message map on the front page of your newsletter. If a stage-message pair fails that test, redesign it. The campaigns that fail the publicity test are the ones that produce backlash when a journalist eventually documents the back-stage strategy.

Practical Steps: Running PAPM This Week

You don’t need a grant or a research budget to apply PAPM to whatever campaign you’re shipping next. Here is the minimum-viable PAPM rollout.

1. Write the 5-question stage classifier. Question 1: “Have you heard of [the issue]?” Question 2: “Do you think [the issue] applies to you personally?” Question 3: “Are you actively considering [the behavior]?” Question 4: “Have you decided whether or not to do [the behavior]?” Question 5: “Have you done [the behavior]?” The branching logic gives you a clean Stage 1 through Stage 7 segmentation in under 90 seconds of respondent time.
2. Insert the classifier into one channel. The fastest place is your existing email list. A single-question pop-up on your highest-traffic landing page works equally well. The classifier is the campaign’s most important asset; protect the data.
3. Draft five message variants, one per stage (combine Stages 1 and 2, combine Stages 6 and 7 if you must). Each variant should be obviously different in tone, evidence type, and call-to-action. If two variants look almost the same, you’ve under-differentiated.
4. Build a Stage 4 reactivation track separately. Most teams build four variants and treat Stage 4 the same as Stage 2 or Stage 3. Don’t. Stage 4 needs its own track, keyed to a face-saving reframe (changed circumstances, new evidence, lowered cost). If you don’t have one, document the gap and don’t run Stage 4 outreach until you do.
5. Route based on classifier output. Email sequences, push notifications, in-app messages all branch by stage. If your CRM can’t branch, this is the single most important infrastructure upgrade for your behavior-change work.
6. Measure stage progression, not just conversion. Track the percentage of your audience moving from Stage N to Stage N+1, not just the percentage that hit Stage 6 this week. The stage-progression metric reveals which variants are doing work and which are noise.
7. Re-classify on a regular cadence. Stages aren’t stable on long horizons. Re-run the classifier quarterly for active behaviors, annually for one-time precautions. People move between stages; your campaign should track the movement.

PAPM Was the Beginning, Not the End

The seven stages are not the last word on behavior change. They are the cleanest stage-typology in the precaution literature, and the stage-matched intervention thesis is the cleanest causal claim about why stage-matching matters. But the field has moved on in three specific directions that any modern behavior-change practitioner should know.

First, Schwarzer’s Health Action Process Approach (HAPA) compressed the seven stages into a binary motivational-volitional split and added continuous variables for outcome expectancy, self-efficacy, action planning, and coping planning. Where PAPM stops at “decided to act,” HAPA goes further into the volitional phase and asks the planning question PAPM under-specifies. We covered HAPA in the HAPA pillar, and the two pillars together cover the stage-theory territory at complementary resolutions.

Second, Susan Michie and colleagues’ Capability-Opportunity-Motivation-Behavior (COM-B) model and Behavior Change Wheel reframed the question from “what stage is the person in?” to “what does the person need in order to act?” COM-B’s diagnostic-first stance is a different methodological commitment from PAPM’s typology-first stance. We covered COM-B in the COM-B pillar; a modern stage-targeted campaign should run a COM-B diagnostic alongside the PAPM stage classifier to figure out whether the stage transition is blocked by capability, opportunity, or motivation.

Third, the experimental-policy literature in the Test-Learn-Adapt protocol from the UK Behavioural Insights Team, has shifted the epistemics from “which model is correct” to “which intervention works in this setting, measured by RCT.” PAPM-derived hypotheses are unusually good fodder for that protocol because they generate sharp predictions about stage-message interactions. A campaign that ran a stage-matched intervention through a Test-Learn-Adapt protocol would produce the strongest evidence the field has ever seen on stage-matched advantage in the wild.

The honest summary is that single-model orthodoxy is dead. PAPM is the right model for the stage-typology question, HAPA is the right model for the binary motivation-volition question, COM-B is the right model for the capability-opportunity-motivation diagnostic, and Test-Learn-Adapt is the right epistemic frame for “did the intervention actually work.” A practitioner who uses all four is doing modern behavior-change work. A practitioner who picks one and ignores the others is doing 1990s behavior-change work.

Frequently Asked Questions

What is the Precaution Adoption Process Model?

The Precaution Adoption Process Model (PAPM) is a seven-stage map of how people move from being unaware of a health risk to fully maintaining the behavior that addresses it. It was developed by Neil Weinstein at Rutgers in the late 1980s, validated through a series of studies on radon testing and other precaution behaviors, and formalized in Weinstein, Sandman & Blalock’s 2008 chapter in Health Behavior and Health Education. The model’s central thesis is that different stages need different interventions, and that aiming the wrong intervention at the wrong stage is the single biggest waste of effort in behavior-change campaigns.

How does PAPM differ from the Transtheoretical Model?

PAPM has seven stages; the Transtheoretical Model (TTM, Prochaska & DiClemente) has five. PAPM splits the “not yet acting” group into three distinct stages (Unaware, Unengaged, Decided Not to Act) where TTM lumps them into a single Precontemplation stage. PAPM treats Stage 4 (Decided Not to Act) as terminal; TTM treats non-action as part of a cycling lapse pattern. The simplest decision rule: use PAPM for one-time precautions like radon testing or vaccination; use TTM for cycling behaviors like smoking, dieting, or exercise. The two models are complementary rather than competitive.

What are the seven stages of PAPM?

Stage 1 (Unaware of the issue), Stage 2 (Unengaged by the issue), Stage 3 (Deciding about acting), Stage 4 (Decided not to act — terminal off-ramp), Stage 5 (Decided to act), Stage 6 (Acting), and Stage 7 (Maintenance). Stage 4 branches off from Stage 3; all other stages form a forward sequence. The qualitative-difference claim is that the psychological barriers between adjacent stages are not interchangeable, so the message that moves Stage 2 to Stage 3 will not move Stage 5 to Stage 6.

Why does Weinstein separate “Unaware” and “Unengaged”?

Because the interventions that work for each are completely different. Stage 1 (Unaware) needs awareness content — the person doesn’t know the issue exists, so any persuasion message that assumes prior knowledge bounces off. Stage 2 (Unengaged) needs personalization content — the person knows about the issue but treats it as someone else’s problem, so the lever is personal-risk anchoring. Lumping the two into one “Precontemplation” stage, as TTM does, forces the campaign to choose between awareness content (which wastes Stage 2 readers) and personalization content (which goes over Stage 1 readers’ heads).

What is Stage 4 and why is it important?

Stage 4 (Decided Not to Act) is the stage where the person has weighed the decision and concluded that the precaution isn’t for them. It’s structurally different from Stage 2 (Unengaged) because the Stage 4 person has already done the deliberation and has a sunk-cost defense around the negative conclusion. The Stage 4 reactivation playbook requires a face-saving on-ramp — usually changed external circumstances, new evidence, or a lowered cost — rather than the awareness or personalization content that works at earlier stages. Treating Stage 4 audiences as if they were Stage 2 is one of the most common and most expensive mistakes in vaccination outreach, climate communication, and product re-engagement.

How do you measure which PAPM stage someone is in?

Through a short self-report classifier. The standard five-question algorithm asks whether the person has heard of the issue (Stage 1 vs others), whether they think it applies to them personally (Stage 2 vs others), whether they’re actively considering the behavior (Stage 3 vs others), whether they’ve decided yes or no (Stage 4 or Stage 5), and whether they’ve performed the behavior recently (Stage 6 or Stage 7). The reliability is acceptable at the extremes and weaker at the middle distinctions, particularly Stage 2 versus Stage 3 and Stage 3 versus Stage 5. Improving the classifier is the single highest-leverage investment in any stage-targeted campaign.

Does stage-matched intervention actually outperform single-message campaigns?

In controlled studies with good classifiers, yes — Weinstein, Lyon, Sandman & Cuite’s 1998 radon study and Lippke and colleagues’ 2009 physical-activity work both showed reliable stage-matched advantage. In field studies with noisy classifiers, the advantage shrinks and sometimes disappears. The honest summary is that stage-matched intervention reliably beats single-message broadcast when stage classification accuracy is high and stage differences are large; it doesn’t reliably beat single-message broadcast when classification is noisy or stages are close together. The implication for practitioners is that the classifier matters as much as the message.

How does PAPM compare to the Health Belief Model?

The Health Belief Model (HBM) treats perceived susceptibility, severity, benefits, and barriers as continuous predictors that combine additively to predict behavior. PAPM doesn’t replace those variables; it places them inside a stage architecture and claims they’re operative at different stages. Susceptibility and severity dominate the Stage 2 to Stage 3 transition. Benefits and barriers dominate Stage 3 to Stage 5. None of the original HBM variables speak strongly to Stage 7 maintenance. A PAPM-overlay on an HBM campaign tells you when to use which HBM lever and when not to.

Can PAPM be applied to non-health behaviors?

Yes, and it routinely is, even when practitioners haven’t read Weinstein. Software-as-a-Service (SaaS) funnels (Awareness, Consideration, Trial, Purchase, Activation, Retention) are a relabeled PAPM staircase. The civic-engagement literature uses PAPM-style segmentation for voter outreach. The climate-communication literature increasingly uses PAPM-derived segments (Yale’s “Six Americas” maps closely onto PAPM stages). The seven-stage logic generalizes to any precaution-style or adoption-style behavior where the audience contains a meaningful share of pre-deliberation and post-deliberation members.

How does the Octalysis Framework integrate with PAPM?

Octalysis tells you which Core Drive to activate; PAPM tells you when each Core Drive is operative. Core Drive 7 (Curiosity) dominates Stage 1 to Stage 2. Core Drive 8 (Loss) in its personalized form, paired with Core Drive 4 (Ownership), dominates Stage 2 to Stage 3. Core Drive 5 (Social Influence) and Core Drive 2 (Accomplishment) dominate Stage 3 to Stage 5. Core Drive 6 (Scarcity) plus friction-removal dominates Stage 5 to Stage 6. Core Drive 4 (Ownership of identity) and Core Drive 1 (Epic Meaning) dominate Stage 6 to Stage 7. Pulling the wrong Drive at the wrong stage is the campaign’s single biggest leak. The 6-Step PAPM × Octalysis Audit in the Apply section above is the protocol that makes the crosswalk actionable.

References

1. Weinstein, N. D. (1988). The precaution adoption process. Health Psychology, 7(4), 355–386.
2. Weinstein, N. D., & Sandman, P. M. (1992). A model of the precaution adoption process: Evidence from home radon testing. Health Psychology, 11(3), 170–180.
3. Weinstein, N. D., Lyon, J. E., Sandman, P. M., & Cuite, C. L. (1998). Experimental evidence for stages of health behavior change: The precaution adoption process model applied to home radon testing. Health Psychology, 17(5), 445–453.
4. Weinstein, N. D., Sandman, P. M., & Blalock, S. J. (2008). The precaution adoption process model. In K. Glanz, B. K. Rimer, & K. Viswanath (Eds.), Health behavior and health education: Theory, research, and practice (4th ed., pp. 123–147). Jossey-Bass.
5. Blalock, S. J., DeVellis, B. M., Giorgino, K. B., DeVellis, R. F., Gold, D. T., Dooley, M. A., Anderson, J. J. B., & Smith, S. L. (1996). Osteoporosis prevention in premenopausal women: Using a stage model approach to examine the predictors of behavior. Health Psychology, 15(2), 84–93.
6. Hammer, G. P. (1996). A stage model of behavior change for hepatitis B vaccination among injection drug users. Public Health Reports, 111(Suppl 1), 99–101.
7. Sniehotta, F. F., Luszczynska, A., Scholz, U., & Lippke, S. (2010). Discontinuity patterns in stages of the precaution adoption process model: Meat consumption during a livestock epidemic. British Journal of Health Psychology, 10(2), 221–235.
8. Lippke, S., Ziegelmann, J. P., Schwarzer, R., & Velicer, W. F. (2009). Validity of stage assessment in the adoption and maintenance of physical activity and fruit and vegetable consumption. Health Psychology, 28(2), 183–193.
9. Schwarzer, R. (2008). Modeling health behavior change: How to predict and modify the adoption and maintenance of health behaviors. Applied Psychology, 57(1), 1–29.
10. Sheeran, P. (2002). Intention–behavior relations: A conceptual and empirical review. European Review of Social Psychology, 12(1), 1–36.
11. Sheeran, P., & Webb, T. L. (2016). The intention–behavior gap. Social and Personality Psychology Compass, 10(9), 503–518.
12. Bridle, C., Riemsma, R. P., Pattenden, J., Sowden, A. J., Mather, L., Watt, I. S., & Walker, A. (2005). Systematic review of the effectiveness of health behavior interventions based on the transtheoretical model. Psychology and Health, 20(3), 283–301.
13. Brewer, N. T., Chapman, G. B., Rothman, A. J., Leask, J., & Kempe, A. (2017). Increasing vaccination: Putting psychological science into action. Psychological Science in the Public Interest, 18(3), 149–207.
14. Bostrom, A., Morss, R. E., Lazo, J. K., Demuth, J. L., Lazrus, H., & Hudson, R. (2018). Eyeing the storm: How residents of coastal Florida see hurricane forecasts and warnings. International Journal of Disaster Risk Reduction, 30, 105–119.
15. Rogers, R. W. (1975). A protection motivation theory of fear appeals and attitude change. The Journal of Psychology, 91(1), 93–114.
16. Rosenstock, I. M. (1974). Historical origins of the health belief model. Health Education Monographs, 2(4), 328–335.
17. Prochaska, J. O., & DiClemente, C. C. (1983). Stages and processes of self-change of smoking: Toward an integrative model of change. Journal of Consulting and Clinical Psychology, 51(3), 390–395.
18. Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50(2), 179–211.
19. Wood, W., & Neal, D. T. (2007). A new look at habits and the habit-goal interface. Psychological Review, 114(4), 843–863.
20. Graybiel, A. M. (2008). Habits, rituals, and the evaluative brain. Annual Review of Neuroscience, 31, 359–387.
21. Chou, Y. (2015). Actionable Gamification: Beyond Points, Badges, and Leaderboards. Octalysis Media.

Related Reading

• Behavioral Framework Library — the full hub of behavioral and psychological frameworks every designer should know.
• Health Action Process Approach (HAPA) Explained — the binary-phase model that compressed PAPM’s seven stages.
• Health Belief Model Explained — the continuous-variable model that PAPM extends.
• COM-B and the Behavior Change Wheel — the diagnostic-first complement to PAPM’s typology-first stance.
• Protection Motivation Theory Explained — the threat-and-coping appraisal model PAPM dovetails with at the Stage 2 to Stage 3 transition.
• Sludge Explained — the friction inventory PAPM Stage 5 to Stage 6 needs to address.
• Libertarian Paternalism Explained — the ethics frame for stage-matched persuasion.

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