The VPSI Framework: A Practical Scoring Model for AI Visibility

Why VPSI exists

Most teams ask for one number: "How are we doing in AI?"

A single number is useful for reporting, but it cannot diagnose the problem.

You can have:

high mention frequency but low rank,
good rank but weak framing,
or positive framing with factual errors.

Each requires a different fix.

VPSI separates those realities so teams can prioritize correctly.

The four dimensions

Visibility

Question: Are we included in answers?

Track two signals:

Unprompted visibility: AI mentions your brand without being asked by name.
Prompted visibility: AI mentions your brand when the prompt includes your name.

Unprompted visibility is the stronger demand signal for category discovery.

Position

Question: Where do we appear when options are ranked?

Position captures relative prominence when AI returns lists or comparisons.

Practical interpretation:

Position 1-2: strong commercial advantage
Position 3-4: visible but losing consideration share
Position 5+: present but weak influence

Perception

Question: How does AI describe us?

Classify mentions by decision impact:

Recommended
Mentioned
Hedged
Warned

A brand can be visible and still underperform if mentions are mostly hedged.

Integrity

Question: Are claims accurate and current?

Focus on risk-heavy fields first:

pricing,
availability,
certifications,
timelines,
policy-sensitive statements.

Integrity protects trust and reduces downstream friction for sales and support.

How to read an actual score profile

Example:

Visibility: high
Position: medium
Perception: low
Integrity: high

This means awareness exists, but recommendation strength is weak.

The right move is not "publish more." The right move is to improve evidence quality and source trust that affects framing.

Another example:

Visibility: low
Position: high
Perception: medium
Integrity: high

This means when AI finds you, you perform well. The issue is discoverability breadth.

The move here is source expansion and entity coverage.

Recommended weight logic

Default weighting can be:

Visibility: 30%
Position: 25%
Perception: 25%
Integrity: 20%

Why this works:

Visibility is prerequisite.
Position and Perception drive commercial outcome.
Integrity acts as trust control.

You can adjust weights by category maturity, but keep the structure stable.

Implementation guidelines

Keep a fixed prompt baseline by funnel stage.
Segment by market (for example India vs global) where demand language differs.
Track per engine before blending into composite.
Attach each recommendation to one measured gap.
Re-measure after every action cycle.

Without this loop, scores become reporting noise.

What VPSI is not

VPSI is not a sentiment-only system.

It is not a one-time audit.

It is not a substitute for product positioning.

It is an operating framework that connects measurement to execution.

Final takeaway

If you want AI visibility work to survive beyond one campaign, you need a model that leadership can track and teams can execute.

VPSI gives you that bridge.