The AVS Rubric: Methodology
How we measure whether your trust infrastructure is strong enough for AI product growth to compound.
Why Trust Infrastructure Determines AI Product Growth
AI-native products face a unique market constraint: buyers can't predict how your product will behave before purchasing.
Unlike traditional SaaS where workflows are deterministic ("click here, get this result"), AI outputs vary based on context, model versions, and user inputs. This unpredictability creates a trust gap that breaks traditional growth loops before they can compound.
The Adaptive Value System (AVS) Rubric measures whether your trust infrastructure is strong enough for growth to accelerate.
The Trust Stack
The rubric assesses eight trust dimensions organized in a hierarchical stack. Gaps in lower layers cascade upward, making upper layers unstable.
The Eight Dimensions
Click any dimension to expand its details.
Layer 1: Product-ICP Clarity
Layer 2: Pricing Architecture
Layer 3: Operational Controls
Layer 4: Enterprise Readiness
How It Works
Input: Publicly Observable Signals
You enter your company URL. The AVS Rubric Agent crawls up to 25 pages across your public digital presence, prioritizing high-intent surfaces using a weighted scoring engine:
Homepage
Primary positioning, outcome claims, ICP signals
Pricing page
Tier structure, unit definitions, overage behavior, billing options
Documentation & API reference
Cost calculators, usage examples, metering details, quickstarts
Blog & changelog
Product updates, case studies with quantified outcomes
Trust center
Security controls, compliance certifications, audit surfaces
Use cases & case studies
Workflow specificity, customer outcomes, proof artifacts
Terms of service
Overage policies, limit behaviors, renewal/cancellation terms
Community & investor content
Public demos, architecture posts, community evidence
Evidence Quality Rules
Not all public content counts as evidence. The rubric enforces strict quality filters:
- • Rejected as evidence: copyright footers, cookie banners, navigation menus, social media links, generic legal boilerplate, partner logos without context, job postings, and auto-generated content.
- • Marketing slogans rejected unless accompanied by specific, concrete details (metrics, features, workflows, pricing numbers).
- • Every citation must be specific: page + concrete fact (e.g., "Pricing page lists 3 tiers: Free, Pro ($49/mo), Enterprise (custom)").
- • Duplicate evidence is counted once — the same fact on multiple pages does not inflate confidence.
The report reflects what your prospects can actually see — not what you believe you're communicating. This is the trust infrastructure gap.
Output
You receive a trust infrastructure report with a total score (0–16, mapped to a maturity band) including:
- Overall AVS Score — Sum of 8 dimension scores (each 0–2), categorized as Nascent, Emerging, Established, or Advanced
- Dimension breakdown — Individual 0–2 scores with subtest-level detail for each of the 8 dimensions
- Strengths & weaknesses — What's working, what's missing, and what it enables or blocks
- Trust breakpoints — Specific gaps that are actively blocking trust and growth
- Confidence labels — How certain each assessment is, based on evidence quality
- 90-day focus — Prioritized actions with measurable outcomes to close trust gaps
The 0–2 Scoring System
Each dimension is scored on a 0–2 scale using a deterministic, subtest-based methodology. Scores are not subjective ratings — they are computed from evidence.
Score Definitions
The trust signal is absent or too vague to be actionable. Fewer than 3 of 6 subtests pass.
This is a critical gap — prospects cannot evaluate this aspect of your product.
Partial evidence exists but key elements are missing. 3–4 of 6 subtests pass, or a hard gate caps the score.
Foundation exists but isn't complete enough to build trust at scale.
Clear, specific, and verifiable evidence across the dimension. 5–6 of 6 subtests pass with no gate failures.
This dimension is actively building trust and enabling growth.
How Scores Are Computed
Each dimension uses 6 subtests (5 for Buyer & Budget Alignment). Each subtest is binary: pass (1) or fail (0). The subtests evaluate specific, observable criteria — not subjective impressions.
Points → Score Mapping
0–2 pts
Score = 0
3–4 pts
Score = 1
5–6 pts
Score = 2
Hard Gates
Certain subtests act as hard gates — if they fail, the dimension score is capped regardless of how many other subtests pass. Gates enforce that critical capabilities (like auditability, measurability, or stated outcomes) cannot be bypassed.
Example: In the "Value Unit" dimension, if the unit definition lacks auditability (no dashboard breakdown or export logs), the score is capped at 1 — even if all other subtests pass. A billable unit that customers can't verify isn't production-grade.
Overall Score & Maturity Bands
The total AVS score is the sum of all 8 dimension scores (max 16). This maps to a maturity band:
Nascent
0–4
Emerging
5–8
Established
9–12
Advanced
13–16
Scores are computed from an evidence ledger — every fact is tracked with its source, reliability, and page reference. This makes scores reproducible and auditable.
Confidence Labels
Each dimension includes a confidence score indicating assessment certainty:
Clear, unambiguous evidence found. This is a confirmed finding.
Act on this immediately.
Partial evidence with some ambiguity. May indicate inconsistent messaging.
Investigate further — needs human validation.
Minimal or conflicting evidence. Automated assessment may be missing context.
Don't act alone — signal is weak.
Example interpretations
A High Confidence gap should trigger immediate action — it's blocking trust. A Low Confidence finding might just mean the AI couldn't access the right information.
What Traditional Analytics Miss
Funnels and product analytics tell you what users do. They don't tell you whether users can predict what will happen before they commit.
When the answer is "no," growth loops leak:
- • Sign up but won't invite teammates — uncertainty about cost allocation
- • Activate but won't share outputs — can't predict if it works for recipient
- • Renew once but won't expand — fear of surprise bills
- • First success but won't scale — no confidence in consistency
The AVS Rubric identifies these gaps before they show up in your retention curve — by measuring whether trust infrastructure exists in your public signals.
Why This Matters for AI Products
Traditional SaaS could rely on free trials and generous freemium tiers to build trust through experience. AI products break this model:
1. Free trials are expensive
LLM inference costs, GPU time, and API call expenses make generous free tiers economically unsustainable. Trust must exist before trial, not during it.
2. One good output doesn't guarantee the next
AI output quality varies by input, context, model version, and even time of day. A single successful trial doesn't give buyers confidence the product will work reliably at scale.
Trust must be built through signals — transparent pricing, clear constraints, explicit guardrails, documented failure modes — rather than just experience.
AVS measures whether those signals exist.
Questions about the methodology? Book a 30-min session to discuss your specific context.