The methodology: VRSD framework

The framework for naming
the user your product never wrote for.

VRSD: Vector, Reachability, Severity, Detectability. 4 dimensions, the combination tells you the one Actor to fix before you ship next.

Human Actor, not an edge case.

An Actor is a person playing a role the product never wrote for them, in a scene the product ignored. Actors are not anomalies, they are evidence that the product's definition of "the user" is insufficient. Anti-ICPs were not even considered.

The methodology refuses vague language and every Actor is specific. Not just "vulnerable users." Not just "bad actors." A specific person in a specific scene with a specific motive or vulnerability the product wasn't designed for.

✗ Wrong

"Vulnerable users might be at risk."

✓ Right

"The Returning Spouse - a divorced parent regaining access to a shared calendar account, with visibility to where their ex now goes on Tuesdays."

✗ Wrong

"Bad actors could misuse this."

✓ Right

"The Inference Squatter - a developer who realizes your free chatbot has unrestricted GPT-5 access and starts using it for unrelated coding work."

When you name an Actor specifically, you can score them. When you can score them, you can prioritize which to take out first. When you can prioritize, you can ship.

Every Actor scores on 4 binary dimensions.

Each dimension forces a call. Never "medium" or "depends." The methodology's value comes from forcing the alternatives that turn soft worry into shippable decisions.

V

Vector

How harm travels : The pathway from product behavior to the Actor's experience.

DIRECT INTERMEDIATE SYSTEMIC

DIRECT: the product harms the Actor directly.
INTERMEDIATE: harm travels through a trusted person (teacher, parent, employer, partner).
SYSTEMIC: harm moves through infrastructure such as data pipelines, algorithms, aggregated effects.

R

Reachability

Probability of encountering the harm.

HIGH LOW

HIGH: encountered by default through normal use.
LOW: requires deliberate access or unlikely conditions.

S

Severity

Magnitude of harm, for this specific Actor (not the "average user").

HIGH LOW

HIGH: irreversible, life-altering, requires external intervention to recover.
LOW: recoverable inconvenience, no lasting impact on autonomy or dignity.

D

Detectability

Will anyone notice, engage?

HIGH LOW

HIGH: flagged by analytics, generates support tickets, recognized by the Actor.
LOW: invisible to the company, the Actor, and possibly regulators.

When contested between HIGH and LOW, lean towards HIGH. Invisible harm and unidentified Actors compound, and the cost of over-flagging is a conversation worth having. The cost of under-flagging though is a shipped exploit.

Reachability x Severity = the call.

Plot every Actor on this 2x2, Reachability x Severity scale. The one in the top-right quadrant, HIGH Reach AND HIGH Severity, that's your blind spot. That's the one you need to prioritize ASAP.

HIGH REACHABILITY LOW
The Friction Systemic improvement
THE BLIND SPOT Fix before you ship
The Noise Monitor
The Hidden Threat Response planning
LOW SEVERITY HIGH

The Blind Spot is the ship-blocker quadrant. When you name an Actor who scores HIGH on both Reach and Severity, that Actor needs to block your roadmap until addressed. It's not because shipping is impossible, but because shipping without addressing them ships the harm with the feature, which at a later time you'll end up scrambling to fix, and may get you in legal trouble.

Every investigation surfaces these Actors, so they're non-negotiable.

An investigation that doesn't name a Bridge User and an Adversarial Actor is incomplete. An investigation of a product touching a vulnerable population that doesn't name the specific vulnerability Actor is akin to 'turning a blind eye' and the VRSD Methodology refuses partial coverage.

Base requirements for every investigation

The Bridge User

Is someone who experiences the product through someone else's use, who never opted in. This is the most-missed category - the recipient of automated outreach, the face in a generated image, the voice cloned without consent.

The Adversarial Actor

Is someone deliberately misusing your product - the cheater, the scammer, the prompt-injection prober, the social engineer. They are no longer considered edge cases, they are guaranteed users of any sufficiently useful tool.

Vulnerability-class stacking

When your product touches any of these populations, the corresponding Actor types stack on top of the base requirements:

Children

Predator pattern Actor + Exploitation-by-peers Actor. Required for any product where minors plausibly access, including via a parent's or family account.

Elderly users

Scam-targeting Actor + Isolation-exploitation Actor. Required for any product targeting retirement, healthcare, eldercare, or with 65+ user base.

People in crisis

Crisis-specific exploitation Actor. Focused on mental health, addiction, recovery, abuse survivors, financial distress, bereaved, terminally ill.

Regulated populations

Regulatory-violation Actor. Triggered by HIPAA, COPPA, GDPR special-category, FERPA, PCI-DSS, EU AI Act high-risk categories.

User-to-user products

Asymmetric power Actor. Messaging, trading, teams, comments, shared resources - basically any product enabling one user to act on another.

Refusing a framing.

If your product is "just for fun" but a 12-year-old can use it, the Predator-pattern Actor isn't optional anymore. The VRSD Methodology refuses the "just for fun" framing every such scenarioe.

Same framework, 4 ways in.

The methodology runs through every Villain's Hat product, so you can pick the depth that matches your situation.

The Quick Check

Free.
Claude Skill or ChatGPT GPT. Names 3-6 Actors, scores VRSD, identifies the blind spot to prioritize. 15 minutes.

Try it now →

The Workshop

€497. 3h live.
Your product, walked through the methodology with Pavitra, on a shared board. You'll receive a written verdict in 48 hours in your email.

Apply for the next cohort →

The Investigations Tool

Waitlist open.
8 specialist analysts run on your product. Web-searched precedent. Downloadable PDF dossier. Launches second half of 2026.

Join the waitlist →

The Fractional Advisor

€6,000/month. Two slots maximum.
Embedded in your product planning. Included is VRSD investigation for every new feature, pre-spec.

Book a discovery call →

Find the villains in your product,
before they find your revenue.

Start with the free Quick Check. Just 15 minutes and you'll have real actors named and the blind ppot identified.

Try on ChatGPT → Install Claude Skill →