Synthesis Practitioner Guide #

This guide translates the synthesis into a practical sequence for selecting a behavior, testing fit, designing a solution, and deciding whether to continue, revise, or stop.

Evidence rule: Pre-register thresholds for the named domain, population, context, observation window, and stakes. Ground them in a relevant baseline, pilot data, or cited external evidence. A result from another company is a case-specific observation, not a universal benchmark.

Part 1: Select and Validate the Behavior #

Step 1.1: Observe What the Population Already Does #

Set a sampling plan that reflects the population’s heterogeneity and the stakes of the decision. Continue until the evidence is sufficient to distinguish meaningful patterns and important exceptions. Record why the sample is adequate instead of relying on a universal interview count or timeline.

Use interviews and direct observation to answer:

1. What outcome is the person trying to achieve?
2. What do they currently do to pursue it?
3. Which workarounds, substitutes, and competing behaviors appear?
4. When and where does the behavior occur?
5. What interrupts, delays, or prevents it?
6. What happens immediately after completion?

Prefer observed behavior, records, and traces over stated preference alone. Document where the evidence came from, which population it represents, and which contexts were absent from the research.

Map the behavior chain:

• Antecedent: What makes the opportunity to act available or salient?
• Action: What observable behavior occurs?
• Immediate consequence: What changes for the person or environment?
• Repetition conditions: What supports or obstructs another occurrence?

Useful signals include repeated workarounds, existing spending, recurring attempts, and a clearly described problem. These signals generate hypotheses. They do not establish fit by themselves.

Step 1.2: Compare Candidates with the Behavior Fit Assessment #

Create a short list of candidate behaviors. Define each behavior at the same level of specificity and rate it for the same population, decision horizon, and context.

The Behavior Fit Assessment is a practitioner decision tool for comparing candidate behaviors across Dispositional Fit, Capability Fit, and Context Fit.

It is not a validated measurement instrument.

• Dispositional Fit: The degree to which the behavior matches the population’s relatively enduring tendencies and preferences over the decision-relevant time horizon.
• Capability Fit: Whether the population has the actual abilities and skills required to perform the behavior.
• Context Fit: Whether the external social and physical environment supports the behavior. Time, tools, access, material resources, social expectations, and environmental constraints belong here.

Treat the minimum dimension as a bottleneck and prioritization heuristic; it is not a deterministic probability of behavior.

A score of 6 out of 10 on each Behavior Fit Assessment dimension is a starting threshold that must be calibrated by domain, population, context, stakes, and observed behavior.

Identity Fit is a legacy alternate name for Dispositional Fit. Preserve historical BFA v1.x assessments under their original version. Rescore them under BFA v2.0 before comparing them with current ratings.

Characteristic motivational priorities may inform Dispositional Fit. Active motivation, current emotion, and momentary perception require the fuller Behavioral State Model when they matter. Context Fit remains limited to the Social and Physical Environments. BFA is informed by BSM, but it is not a one-to-one condensation or reassignment of BSM components.

Use the comparison to identify the most plausible behavior and the evidence gaps most likely to change the decision. A high rating is a hypothesis to test in realistic conditions.

Step 1.3: Pre-register the Behavior Market Fit Test #

Before exposing results, write down:

• Target behavior and its observable completion event
• Population and relevant segments
• Social and physical context
• Observation window
• Decision stakes
• Primary outcome and diagnostic measures
• Success, revision, and stopping thresholds
• Source for each threshold
• Planned exclusions and missing-data treatment
• Decision that each possible result would support

Choose measures that match the behavior. Common options include:

• Time to first completed behavior
• First-behavior completion rate
• Behavior frequency within the observation window
• Return to the target behavior
• Abandonment point in the behavior chain
• Segment-level differences
• Unintended outcomes or harms

Retention is useful only when returning is part of the behavior and the observation window matches its natural frequency. A monthly or episodic behavior should not inherit a daily-product retention schedule.

Step 1.4: Interpret the Evidence #

Compare the observed results with the pre-registered criteria. Analyze important segments separately and report uncertainty, missing contexts, and plausible alternative explanations.

Do not infer the cause of a miss from one metric. A low completion rate could reflect a skill gap, an environmental barrier, a weak preference, an active motivational state, measurement error, or several factors together.

Case-specific evidence: In the Proposify onboarding case, the company reported that about 14% of trial users completed its earlier onboarding steps. The redesign focused the flow on sending a first proposal. The source did not disclose a comparable post-redesign completion rate, so the baseline supports the diagnostic story but does not supply a general activation benchmark.

Part 2: Design the Solution Around the Behavior #

Step 2.1: Locate Friction in the Behavior Chain #

Identify where the current attempt slows, fails, or ends. Classify the evidence without moving person-side states into Context Fit:

• Capability gap: The person lacks an ability or skill required by the behavior.
• Context barrier: The social or physical environment restricts time, tools, access, resources, privacy, cues, or opportunity.
• Dispositional mismatch: The behavior conflicts with relatively enduring tendencies or preferences over the decision horizon.
• Active state: Current perception, emotion, or motivation changes the immediate likelihood of action and may require BSM analysis.
• Solution friction: The product introduces unnecessary steps, choices, delays, or uncertainty.

Measure where abandonment occurs and gather qualitative evidence about why. Treat the initial category as a diagnosis to test.

Step 2.2: Specify the Minimum Viable Behavior #

Define the smallest observable action that delivers the intended outcome or a credible first benefit. Remove requirements that do not contribute to completion or learning.

For each remaining step, ask:

1. Is it required for the behavior?
2. Does it demand a new skill?
3. Does the environment make it available at the moment of action?
4. Can a default remove a pre-value decision?
5. Is the consequence visible soon enough for the person to interpret what happened?

The appropriate number of steps and acceptable time depend on the behavior, population, accessibility needs, and stakes. Establish those criteria from baseline evidence and usability research.

Step 2.3: Test the Behavior Path #

Recruit participants from the target population and test in representative contexts. Determine sample adequacy from expected heterogeneity, measurement precision, risk, and the decision being made.

Observe:

• Whether the target behavior is completed
• Time and effort required
• Errors and requests for help
• Environmental interruptions
• Points of hesitation or abandonment
• Whether the intended consequence is understood

Compare results with the registered baseline and thresholds. Revise the smallest plausible cause, then test again. If interface changes do not improve the result, revisit the behavior, population, or environment instead of assuming that more interface work will solve it.

Case-specific evidence: The Instagram pivot case reports company-stated growth from 25,000 users on launch day to 7 million within nine months after the product narrowed around photo sharing. Those figures are adoption signals from one historical case. They do not isolate the effect of behavior selection or define a target for another product.

Part 3: Validate Solution Market Fit #

Step 3.1: Pilot by Relevant Segment #

Define segments before analysis when prior evidence suggests that preference, ability, skill, or environment differs meaningfully. Avoid labels such as “early adopter” unless they correspond to observable selection criteria.

For each segment, measure the same registered outcomes and report:

• Baseline and observed value
• Sample and missingness
• Uncertainty
• Contexts represented
• Adverse or unintended outcomes
• Evidence that supports each diagnosis

An average can hide a strong result in one segment and a harmful or unusable result in another. Use segment differences to refine the target population, behavior, solution, or context.

Step 3.2: Diagnose Abandonment #

Combine behavioral data with follow-up research. Ask when the person stopped, what happened in the environment, which ability or skill was required, what they expected, and what would make another attempt possible.

Classify reasons with the BFA and BSM boundaries:

• Lack of skill or ability: Capability Fit
• Lack of time, access, tools, privacy, or opportunity: Context Fit
• Repeated preference mismatch: Dispositional Fit
• Current emotion, perception, or active motivation: BSM Personal Components
• “This is not for me”: adjacent self-concept or status evidence that requires investigation rather than automatic assignment to Dispositional Fit

Prioritize a fix by decision relevance, expected impact, reversibility, cost, and risk. A frequently mentioned issue is not automatically the most important one.

Part 4: Decide Whether and Where to Scale #

Step 4.1: Register Scaling Criteria #

Scaling criteria should state:

• Which segment is eligible
• Which behavior and outcome must remain stable
• Acceptable uncertainty
• Maximum acceptable harm or failure rate
• Contexts in which evidence applies
• Monitoring interval and rollback condition
• Source and rationale for every threshold

Use thresholds supported by pilot evidence, economics, safety requirements, contractual obligations, or relevant external data. Do not import generic retention bands or satisfaction cutoffs.

Step 4.2: Expand in Evidence-Bounded Increments #

Expand first to populations and contexts represented in the evidence. Treat a new segment, channel, geography, or use environment as a new validity question.

Monitor for:

• Cohort degradation
• Longer or more variable completion paths
• New skill demands
• Environmental barriers absent from the pilot
• Incentive dependence
• Unintended behavior or harm
• A change in who is entering the population

Pause expansion when a registered rollback condition is met. Diagnose before choosing a solution.

Case-specific evidence: The Quibi case records press estimates of roughly 8% to 10% trial-to-paid conversion and a shutdown about six months after launch. Quibi did not publish detailed funnel data, and the pandemic removed much of its intended viewing context. The case illustrates the need to state context and uncertainty; its figures are not universal stopping rules.

Part 5: Continue, Revise, or Stop #

Use the pre-registered criteria together with evidence quality and downside risk.

Continue #

Continue when the named population meets the registered criteria in representative contexts, the evidence is adequate for the stakes, and important harms remain within the stated limits. Record which conditions have been established and which remain uncertain.

Revise #

Revise when the evidence points to a specific and testable bottleneck. Possible revisions include:

• Select a behavior with stronger Dispositional Fit
• Teach or simplify a required ability or skill
• Change the social or physical environment
• Remove solution-created friction
• Narrow the population to a supported segment
• Change the observation window to match the behavior’s natural frequency

State the causal hypothesis and the result that would disconfirm it before retesting.

Stop or Return to Behavior Selection #

Stop the current path when repeated, adequately powered tests miss the registered criteria after plausible corrections, the expected downside exceeds the limit, or another candidate has materially stronger evidence. Stopping is a decision under stated criteria, not the mechanical output of a universal percentage or iteration count.

Document the evidence, uncertainty, attempted revisions, and conditions under which the decision could be reconsidered.

Review Checklists #

Before the Behavior Test #

• Target behavior is observable and specific
• Population, segments, context, window, and stakes are named
• Dispositional, Capability, and Context Fit use the v2.0 definitions
• Historical Identity Fit assessments remain labeled BFA v1.x or are rescored
• Primary outcome and diagnostic measures are defined
• Thresholds have sources and rationales
• Missing-data and exclusion rules are registered
• Possible results map to decisions

Before the Solution Pilot #

• Minimum viable behavior is defined
• Capability gaps are limited to abilities and skills
• Time, tools, access, resources, and opportunity are treated as Context Fit
• Current emotion, perception, and active motivation are analyzed through BSM when relevant
• Participants and contexts represent the intended population
• Baseline and uncertainty will be reported
• Unintended outcomes are monitored

Before Scaling #

• Evidence supports the named segment and context
• Scaling and rollback criteria are registered
• Thresholds reflect domain evidence and stakes
• Segment averages do not conceal important failures
• Monitoring can detect cohort or context drift
• Expansion can be paused or reversed
• Remaining uncertainties are documented

Closing Principle #

Behavioral Strategy starts by selecting a behavior that fits the population and its environment, then tests whether a solution improves the path to that behavior. Each decision should identify the population, context, window, stakes, evidence, and uncertainty. Case studies can suggest mechanisms and measurements. Local evidence determines whether those ideas apply.