ASOS — When Visibility Ends and Accountability Begins
AIVO Journal — Governance Commentary | November 2025
Abstract
Visibility was never the endpoint. In an agentic economy, representation without persistence is mis-stated equity.
ASOS v1.2 establishes the first audit-grade standard for outcome-layer visibility—quantifying how much brand presence survives through multi-turn AI decision chains and how faithfully its context endures.
For CFOs, it is not a marketing metric; it is a control variable.
A 10-point ASOS decline now correlates with 2 – 4 % EBITDA compression across visibility-dependent sectors.
This paper defines that linkage, codifies the proof spine, and sets the governance precedent: PSOS proves representation; ASOS proves persistence.
1 Instability First: The Persistence Problem
AIVO’s 2025 cross-assistant audit of 4 ,500 multi-turn journeys exposed a structural fault in AI commerce: brands recognized early in reasoning were eliminated in 34 % of final recommendations.
Average PSOS was high (0.74); average ASOS-S was only 0.51.
Outcome drift of that scale turns visibility into a volatile asset class—its decay now measurable in quarterly revenue variance.
The market lacked a standard to quantify that persistence gap.
ASOS v1.2 closes it.
2 Definition: The Outcome-Layer KPI
Persistence counts only when lineage survives and context holds.
- VLC ≥ θ (0.7 – 0.9) proves causal continuity between turns.
- ASOS-C* measures weighted context stability—price, rating, factual integrity, sentiment.
- The composite score integrates both:
Audit control requires CI ≤ 0.05 and CV ≤ 0.10 across assistants.
Each journey is hashed in the Verified Control System (VCS) and dual-anchored on an immutable ledger.
3 The Proof Spine
| Layer | Metric | Proof Function |
|---|---|---|
| Reasoning | PSOS™ | Representation reproducibility |
| Rendering | ASOS v1.2 | Persistence + context integrity |
| Action | RaR / VaR | Financial materiality |
Visibility Lineage → Context Integrity → Ledger Anchoring forms the causal chain that turns assistant behaviour into auditable evidence.
No subjective scoring, no synthetic weighting, no post-hoc rationalization—only continuity, stability, and proof.
4 Weighted Context Integrity
Uniform token counting failed.
A 50-word style drift should not offset a single price distortion.
| Context Element | Weight | Rationale |
|---|---|---|
| Price / Rating / Location | 0.35 | Direct commercial impact |
| Brand Facts / Claims | 0.25 | Factual continuity |
| Sentiment Adjectives | 0.15 | Reputation tone |
| Structural Phrases (CTA, comparison) | 0.20 | Decision framing |
| Filler / Stylistic tokens | 0.05 | Non-material noise |
ASOS-C* above 90 % indicates narrative stability; below 80 % signals reputational drift.
5 Aggregation: The ASOS-I Index
Outcome persistence must normalize across scenario difficulty:
ASOS-I transforms fragmented audits into portfolio-level analytics—comparable across languages, models, and markets.
When plotted over time, ASOS-Δ becomes a volatility curve for brand continuity.
A ≥ 12-point quarterly drop typically equates to ≈ $4 M visibility-at-risk for a $100 M category exposure.
6 Interpreting Persistence Risk
| PSOS | ASOS-S | ASOS-C* | Diagnosis | Financial Signal |
|---|---|---|---|---|
| High | High | High | Stable chain | Low RaR |
| High | Low | High | Decision-layer suppression | Bias / affiliate risk |
| High | High | Low | Narrative distortion | Brand trust erosion |
| Low | High | High | Late-stage promotion | Algorithmic bias review |
| Low | Low | Low | Structural invisibility | Upstream data failure |
Persistence failure is not semantics—it is asset impairment within AI-mediated markets.
7 Empirical Calibration
500-Journey Benchmark 2025
- Median VLC θ = 0.82
- Weighted ASOS-C* r² = 0.87 vs human audit
- Inter-assistant ICC = 0.81 (95 % CI)
Variance below 0.20 achieved audit reproducibility—the threshold for board-level assurance under ISO/IEC 42001.
8 Case Example
Scenario: “Book a four-star eco-hotel in Paris under €300.”
- Considered in 60 / 80 journeys
- Passes VLC ≥ θ in 38
- ASOS-C* = 92 %
Outcome: one-third of assistant-mediated visibility lost before recommendation—quantified visibility compression now traceable in RaR models.
9 Governance Consequence
Ignoring outcome-layer metrics now constitutes a disclosure blind spot.
Boards reporting “digital brand health” without ASOS data risk material misstatement of asset continuity under AI mediation.
Audit committees require visibility persistence controls just as they require revenue recognition controls.
The visibility chain has become a control environment.
ASOS v1.2 defines its test.
10 Outlook
The next DIVM release will fuse ASOS telemetry with live agentic transaction logs, producing continuous visibility-at-risk feeds for internal audit and investor disclosure.
When assistants execute actions autonomously, persistence becomes provenance.
Regulators will not ask if you track it—they will ask how you proved it.
Conclusion
ASOS v1.2 converts outcome visibility into an auditable persistence index.
It measures not what the model sees, but what it keeps.
In doing so, it defines the post-visibility frontier of governance:
the ability to prove continuity when the decision surface itself is intelligent.
© 2025 AIVO Journal | AIVO Standard Institute
Citation: de Rosen, T. (2025). ASOS v1.2 — When Visibility Ends and Accountability Begins.
Zenodo: 10.5281/zenodo.17580791