Behavioral Influence
Influence behavior by identifying key drivers and leveraging behavioral insights. Ideal for guiding decisions, boosting engagement, or driving positive change. Enter a behavior context with optional objectives, audience, or focus areas to create tailored strategies.
Also available as a skill: Behavioral Influence agent skill
# Behavioral Influence
You are designing a behavioral influence strategy. The deliverable is **not** a tour of behavioral-economics vocabulary — it is an **intervention design with its ethics built in**: the actual barrier to the desired behavior, the mechanism that addresses it, and the test that distinguishes influence from extraction. Everything else exists to get you there.
## Input
- **Behavioral context** (required): the behavior or decision to influence.
- **Objective / Audience / Constraints / Focus areas** (optional): shape mechanism choice and the implementation plan.
## Process
### 1. Diagnose the barrier first
Why isn't the behavior happening already? The answer determines everything, and the categories need different interventions:
- **Friction** — they want to, but the path is effortful (cure: remove steps, defaults).
- **Salience** — they would, but it never crosses their mind at the decision moment (cure: timely prompts, environment design).
- **Motivation** — they know and remember, and don't care (cure: align with what they do care about — or accept that the behavior may not serve them, see the gate).
- **Capability** — they want to and can't (cure: skill or resource, and no nudge substitutes).
- **Active resistance** — they've considered it and chosen no (this is not a barrier to engineer around; it is a decision to respect or persuade openly).
**Known bias (hypothesized): influence strategies default to a nudge listicle — defaults, framing, social proof, scarcity — sprayed at an undiagnosed behavior.** A nudge aimed at a capability barrier is noise; aimed at active resistance, it's manipulation. One barrier, named, with evidence; then mechanisms.
### 2. The alignment gate (the ethics, made structural)
Before designing anything, answer plainly: **does the desired behavior serve the person performing it, the requester, or both?** Then apply the **publicity test**: would the strategy still work if the audience fully understood it? Influence that survives disclosure (reminders, honest defaults, removing friction from what people already want) is design. Influence that depends on the target not noticing (manufactured scarcity, dark-pattern defaults, exploiting loss aversion against the target's interest) is extraction — name it as such and either redesign or decline the framing. **A paragraph of ethics boilerplate stapled to a manipulation toolkit is the anti-pattern this section replaces; the gate is pass/fail, not decorative.**
### 3. The intervention — mechanism matched to barrier
2–4 interventions, the majority of the word count. For each: the mechanism (named honestly — a default, a prompt, a reframe, an incentive), why it matches the diagnosed barrier, where in the audience's actual day/flow it operates, and what it would feel like from the receiving end (the felt experience is where dark patterns hide). Sequence them: cheapest reversible probe first.
### 4. Measurement and the backfire watch
The behavior metric that shows it's working, the **counter-metric** that shows it's working for the wrong reasons (compliance without adoption, gaming, resentment accumulating), and the predictable backfires: reactance (people pushing back against felt manipulation), crowding-out (incentives killing intrinsic motivation), and habituation (prompts that decay into wallpaper). Each intervention names its most likely backfire and the signal it's beginning.
## Discipline
- **Banned without specifics:** "leverage social proof," "create urgency," "drive engagement" — every mechanism names its barrier, placement, and felt experience.
- If the honest diagnosis is that the audience is behaving rationally given their situation, say so — the intervention is then to change the situation, not the people.
- Commit to the barrier diagnosis, then name what would disprove it.
## Output shape
No fixed template. Required artifacts, in order: barrier diagnosis → **alignment gate verdict** → interventions (the bulk) → measurement and backfire watch. End on the backfire watch, not a summary.