How is the Resolution Integrity observation window defined?

The observation window is defined per task class based on the typical lifecycle of the customer issue the class addresses. For billing disputes: 30 days, because disputed charges surface as repeat contacts within the billing cycle. For technical support: 14 days for recurring issues, 60 days for one-time errors. The window should be long enough to capture all failure modes of false closure and short enough that the measurement cycle informs Steward review at operational tempo. Every window definition is encoded in the [Deterministic Logging](https://arcoventure.studio/lexicon/deterministic-logging) configuration alongside the closure event it governs.

What is the relationship between Resolution Integrity and MTTI?

A falling [MTTI](https://arcoventure.studio/lexicon/mtti) without a change to the [Intervention Threshold](https://arcoventure.studio/lexicon/intervention-threshold) definition is a lagging indicator of Resolution Integrity failure or Knowledge Debt accumulation. If the system is producing more escalations at an unchanged threshold, one cause is that autonomous resolutions are failing: the Steward is being called into cases the system closed incompletely, because the repeat contact or downstream propagation triggered a new escalation. Resolution Integrity measurement distinguishes this from genuine threshold miscalibration: if the integrity rate is falling alongside MTTI, the root cause is output quality failure.

How does Resolution Integrity feed the Exception Architecture improvement cycle?

Every false closure is an exception state: the task class encountered a condition its resolution specification did not cover. When the Steward resolves the false closure, the resolution produces a [Proof of Action](https://arcoventure.studio/lexicon/proof-of-action) record that encodes what a genuine resolution for that condition looks like. The [Exception Architecture](https://arcoventure.studio/lexicon/exception-architecture) is updated to include the new resolution specification. The next time the system encounters the same condition, it either resolves it autonomously against the updated specification or escalates correctly. Resolution Integrity improves because the specification becomes more complete with every false closure the Steward encodes.

Can Resolution Integrity be measured without Deterministic Logging?

Not reliably. Measuring Resolution Integrity requires tracking the downstream state of every closure over the observation window: whether the ticket identifier appears in a subsequent contact event, whether the closure generated a downstream exception, whether rework was attributed to the incomplete resolution. [Deterministic Logging](https://arcoventure.studio/lexicon/deterministic-logging) is the technical foundation: recording the specific conditions of each closure, including output state and downstream identifiers, makes the calculation auditable. A system that logs closures without logging their downstream outcomes can only measure closure rate, which is a less informative metric.

How does Resolution Integrity interact with the 80 Percent Threshold?

The 80 Percent Threshold measures the proportion of cross-departmental handoffs executing without human intervention. Resolution Integrity provides the quality gate: a handoff that executed without human intervention but produced a false closure did not execute correctly. The Threshold measures autonomy; Resolution Integrity measures whether the autonomy produced a valid output. Both are required for a complete quality picture. A business at the 80 Percent Threshold with low Resolution Integrity has achieved autonomous execution volume without autonomous execution quality.

Closed Is Not the Same as Resolved

Resolution Integrity is the proportion of agentic ticket closures that do not reopen, do not resurface as repeat contacts within a defined observation window, and do not propagate downstream as exceptions or rework — distinguishing genuine first-contact resolution from mere ticket-closure containment. The Escalation Rate tells the Steward how often the autonomous system required human involvement. It does not tell the Steward what happened to the cases the system handled autonomously. A low Escalation Rate combined with low Resolution Integrity is the specific signal that the Intervention Threshold is set too permissively: the system closes autonomously at a high rate while producing closures that do not hold.

In a human-staffed operation, false closure has a natural limit: the agent is accountable for the outcome over time. In an agentic operation, the agent closes tickets at maximum speed because that is what it is optimised for. An autonomous system without Resolution Integrity governance generates downstream work faster than it saves it: the tickets it closes incompletely resurface as repeat contacts, downstream exceptions, and rework that consumes the Steward time the autonomous architecture was supposed to free.

The four failure modes of false closure

False closure manifests in four ways. Reopening: the customer contacts again with the same issue within the observation window, indicating the resolution did not address the root cause. Repeat contact: the customer contacts on a related issue within the window, indicating the resolution addressed the surface symptom without identifying the underlying condition. Downstream propagation: the closed ticket generates an exception in a downstream workflow — a billing dispute, a data anomaly, an operations exception — because the resolution was incomplete at the point of closure. Rework: an adjacent workflow must reprocess or correct work the closed ticket was supposed to complete.

Deterministic Logging is the technical foundation for all four measurements: recording not just that a closure occurred but the specific conditions of the closure, including the output produced and the downstream state it left, makes the Resolution Integrity calculation auditable. It cannot be measured without it.

How Resolution Integrity governs the Steward’s quality oversight

Under the Stewardship Model, the Escalation Rate and Resolution Integrity together constitute the primary quality governance instruments. The Escalation Rate measures autonomous execution volume; Resolution Integrity measures whether that execution produced valid outputs. Resolution Integrity targets vary by task tier: T1 closures require near-perfect integrity. A T1 integrity rate below 95% indicates systematic miscalibration that the Continuous Regression Loop should be detecting. T2 and T3 closures involve the Steward precisely because their complexity makes autonomous integrity harder to guarantee.

Designing for genuine resolution

The Resolution Integrity discipline requires encoding what a genuine resolution looks like for each task class before the first autonomous closure occurs. The Exception Architecture specifies which task states the system can resolve autonomously and what the output of a valid resolution must include. A resolution missing required output elements is an incomplete closure that will resurface. Encoding this as a post-closure validation step — that checks the output against the resolution specification before the ticket status updates — is the architectural mechanism that prevents false closure from accumulating.

Every genuine resolution recorded in the Operational Ledger compounds the resolution specification: the Steward can identify which output elements are consistently present in genuine resolutions and absent in false closures, and encode them as the resolution validation criteria. The Resolution Integrity rate improves not just because the model improves but because the specification that governs its output becomes more precise with every cycle.

The Operator’s Verdict

An autonomous system optimised for closure velocity without Resolution Integrity governance does not eliminate support cost. It defers it. The false closures it generates resurface as repeat contacts, downstream exceptions, and rework. The net cost of the autonomous system is higher than the gross cost of its closure rate suggests, and the MTTI compression that results is a symptom of the quality gap rather than a signal of model failure. Architectural Certainty at the resolution layer — the state in which the autonomous system closes the right cases correctly at the right rate — is only reachable when Resolution Integrity is designed in from the first deployment, not retrofitted after the false closure rate becomes visible.

KEY TAKEAWAY

What is Resolution Integrity and why is it the quality metric that Escalation Rate cannot replace?

Resolution Integrity is the proportion of agentic ticket closures that do not reopen, do not resurface as repeat contacts within a defined observation window, and do not propagate downstream as exceptions or rework. It provides the output quality dimension that Escalation Rate cannot: Escalation Rate measures how often the autonomous system handled a ticket without human involvement; Resolution Integrity measures whether those autonomous handling events actually resolved the underlying condition. A low Escalation Rate combined with low Resolution Integrity signals the Intervention Threshold is set too permissively. The system is proceeding to autonomous closure on cases it cannot genuinely resolve, and the false closures create downstream work the architecture was supposed to eliminate. Key metric: T1 task class closures require near-perfect Resolution Integrity. A T1 rate below 95% indicates systematic miscalibration the Continuous Regression Loop should be detecting and the Steward should be prioritising.