# Citation persistence review worksheet

Follow one exact target URL across registered prompt cells and ordered observation windows. Preserve the original positive cohort, observed losses and collection gaps to describe what remained continuously confirmed.

Worksheet version 1.0 · 25 September 2026. Complete and repeat records as needed. Blank fields request your evidence; these manual files do not collect answers, publish changes or calculate RankEcho product metrics.

Complete guide: https://rankecho.io/resources/prove/citation-persistence

## 1 Freeze the target and schedule

Register the full schedule before collecting later answers. W1 is the anchor-positive post-shipment window, not a reconstructed baseline.

**Target identity and event rule** — Record exact target URL, page version, redirect and normalization rules, qualifying source roles and raw source capture procedure. A citation elsewhere on the same domain is a separate event.

[Enter your response]

**Stable cell context** — Record panel version and cell IDs with exact prompts, surface, locale, session, exposed model label and collection mode. Define how context changes and unclassifiable answers are retained.

[Enter your response]

**Ordered windows and shipment** — Record shipment time and evidence, each planned window and UTC capture schedule, owner and registered endpoint. Keep shipment before W1 and windows strictly ordered; retain every planned cell.

[Enter your response]

## 2 Record every cell and window

Keep one record for each planned cell in every window. Do not extend sampling selectively after a loss or replace a failed capture with an invented negative answer.

**Observation and URL evidence** — Record cell and window IDs, UTC capture time, context, original and resolved URL, source role and raw capture pointer. Every target-positive state must resolve to the same exact normalized URL.

[Enter your response]

**Observed and missing states** — Choose target cited, other owned only, observed without target, observed unclassifiable, unavailable or failed. Retain reasons and gaps; a changed context cannot confirm uninterrupted persistence.

[Enter your response]

**W1 cohort membership** — List only exact-target-positive W1 cells in the inception cohort. Preserve all other cells outside it and record later incidents separately. If W1 has no positives, persistence is unavailable, never 0%.

[Enter your response]

## 3 Reconcile cumulative status

The fictional C01–C04 W1 cohort contains four cells. C02 loses the target at W3; C03 fails at W3; C04 is unavailable at W4. Later returns cannot erase these records.

**Status through each window** — Classify every cohort cell once: continuous when all windows cite the target; lost after any observed target absence; indeterminate when there is no confirmed loss but a gap or context drift remains.

[Enter your response]

**Partition and separate coverage** — Verify continuous plus lost plus indeterminate equals the original cohort size. Separately report observed coverage and target presence among observed follow-ups; keep each denominator explicit. With no observed follow-ups, target presence is unavailable, never 0%.

[Enter your response]

**Returns and later incidents** — Keep reacquisition after loss separate from a target seen after a gap. Neither restores continuous history. Exclude later incident cells from the original cohort throughout the analysis.

[Enter your response]

| Through window | Continuous | Lost | Indeterminate |
| --- | --- | --- | --- |
| W1 | 4 | 0 | 0 |
| W2 | 4 | 0 | 0 |
| W3 | 2 | 1 | 1 |
| W4 | 1 | 1 | 2 |

## 4 Review and report the record

At W4, the example has one continuously confirmed cell, one lost and two indeterminate. C05 and C06 first cite T01 at W2 and remain later incidents outside the W1 cohort.

**Independent reconstruction** — Ask a reviewer to derive cohort membership, every status and follow-up denominator from the full ledger. Preserve evidence for all 24 fictional records when reproducing the six-cell, four-window example.

[Enter your response]

**Descriptive conclusion** — Report endpoint, full cohort partition, coverage, reacquisitions, gaps and later incidents. Do not treat point presence as uninterrupted retention or infer that the preceding change caused the result.

[Enter your response]

**Next decision and corrections** — Record continue, investigate loss, repair collection or stop, with owner and date. Keep the completed schedule intact. Any further monitoring gets a new prospective schedule and version.

[Enter your response]

## Complete guide

Citation persistence asks whether an exact normalized target URL continues to appear for the same registered prompt-surface cells across consecutive planned windows after an anchor-positive W1 citation. Freeze that inception cohort, calculate exact-target presence only among observed follow-ups, publish coverage separately, retain shifts, absences, reacquisitions, later new gains, and unobserved states, and never add later positives to W1. Missing, unavailable, and failed observations are indeterminate, not losses. This is not a durability guarantee, causal effect, citation rate, answer-volatility estimate, or fixed-prompt baseline comparison. Site traffic is a separate outcome and is not required.

Original guide update: 2026-09-07. Worksheet prepared 25 September 2026. The original guide and its evidence records retain their dates. This manual review record complements the original guide and retains its separate method and product boundaries.

## What is the scope of citation persistence?

The unit is an exact target URL within one stable prompt-surface cell observed over ordered post-shipment windows. The method does not ask whether any page on the domain appeared, whether answer prose stayed similar, or whether a page change caused the observation.

Use one predeclared schedule and retain every window. W1 is the anchor-positive post-change cohort definition, not a reconstructed baseline. Do not extend sampling only after a loss, omit an unobserved result, or move a later positive into the original cohort.


## How does this differ from citation rate, volatility, and a fixed retest?

Citation rate pools declared positives across eligible cells within a window. Answer volatility compares unordered feature-signature pairs. A fixed-prompt retest compares baseline with a later window. Persistence begins only when W1 contains the exact target citation and preserves the order of every later scheduled window.

These methods can coexist, but their numerators, denominators, order, missingness rules, and questions are not interchangeable.

| Method | Unit and question |
| --- | --- |
| Wrong-page diagnosis | Does a qualifying owned URL match the predeclared owner for this exact prompt cell? |
| Citation rate | What share of eligible cells meets a declared citation rule within one window? |
| Answer volatility | How often do unordered repeat pairs disagree on a predeclared feature signature? |
| Fixed-prompt retest | What changed between a registered baseline and a later matched observation? |
| Citation persistence | After an exact-target citation creates an inception cohort, which cells remain continuously confirmed through ordered later windows? |

- [Review citation-rate units](https://rankecho.io/resources/prove/citation-rate)
- [Review answer-volatility units](https://rankecho.io/resources/prove/answer-volatility)
- [Review the fixed-prompt retest](https://rankecho.io/resources/prove/fixed-prompt-retest)

## Which frozen windows define the synthetic record?

The five-edge artifact was recorded as shipped at 2026-08-03T16:00:00Z. W1 through W4 are later scheduled observations; the baseline wrong-page ledger is contextual history and does not enter the persistence denominator.

Every cell retains its original prompt, surface, locale, signed-out synthetic session, exposed-model label, qualifying-link rule, and exact-URL normalization rule.

| Window | Collected at | Role |
| --- | --- | --- |
| W1 | 2026-08-10T14:00:00Z | First predeclared post-shipment window and inception window |
| W2 | 2026-08-17T14:00:00Z | Second predeclared post-shipment window |
| W3 | 2026-08-24T14:00:00Z | Third predeclared post-shipment window |
| W4 | 2026-08-31T14:00:00Z | Fourth predeclared post-shipment window |


## What does the worked synthetic persistence ledger show?

SYNTHETIC EXAMPLE — Alderwave Relay, its pages, prompts, surfaces, citations, links, and observations are fictional; this is not RankEcho data, customer data, provider evidence, or a benchmark.

T means the exact normalized T01 URL is cited; O records a shift to another owned URL without T01; F and U retain unobserved failed or unavailable states. An observed answer without T01 would remain a separate target-absence state. C01–C04 form the W1 anchor-positive cohort. C05 and C06 are later new gains at W2 and remain outside that cohort.

| Cell | W1 | W2 | W3 | W4 | W1-cohort interpretation |
| --- | --- | --- | --- | --- | --- |
| C01 | T · exact target cited | T · exact target cited | T · exact target cited | T · exact target cited | Confirmed through W4 |
| C02 | T · exact target cited | T · exact target cited | O · other owned only | T · exact target cited | Shifted at W3 · target reacquired at W4; loss retained |
| C03 | T · exact target cited | T · exact target cited | F · failed | T · exact target cited | W3 failed · target reobserved at W4; continuity unknown |
| C04 | T · exact target cited | T · exact target cited | T · exact target cited | U · unavailable | W4 unavailable · continuity unknown |
| C05 | O · other owned only | T · exact target cited | T · exact target cited | T · exact target cited | Later new gain at W2 · outside W1 cohort |
| C06 | F · failed | T · exact target cited | T · exact target cited | T · exact target cited | Later new gain after W1 failure · outside W1 cohort |


## How does the synthetic persistence arithmetic reconcile?

The primary W1 cohort contains four cells. All four remain confirmed through W2. At W3, C02 has an observed loss and C03 has a failed gap, leaving two confirmed, one lost, and one indeterminate. At W4, C04 becomes unavailable, leaving one confirmed, one lost, and two indeterminate.

C02's W4 target citation does not erase its W3 loss. C03's W4 citation does not fill its failed W3 observation. The partition therefore moves 4/0/0, 4/0/0, 2/1/1, and 1/1/2 across confirmed continuous, confirmed lost, and indeterminate states.

Observed-follow-up denominators remain separate from coverage. Exact T01 is present in 4/4 observed cohort cells at W2, 2/3 at W3, and 3/3 at W4; coverage is 4/4, 3/4, and 3/4. W4 includes one uninterrupted retention, one reacquisition after a shift, and one target observation after a failed gap.

| Through window | W1 inception cohort | Confirmed continuous | Confirmed lost | Indeterminate |
| --- | --- | --- | --- | --- |
| W1 | 4 | 4 | 0 | 0 |
| W2 | 4 | 4 | 0 | 0 |
| W3 | 4 | 2 | 1 | 1 |
| W4 | 4 | 1 | 1 | 2 |


## How do observed-follow-up denominators and coverage stay separate?

The exact-target-present column uses only eligible observed W1-cohort follow-ups as its denominator. Coverage uses the full four-cell anchor-positive cohort. Failed, unavailable, and unclassifiable rows reduce coverage and remain unknown rather than entering the target-absence numerator.

Point presence does not repair continuous history: C02 is reacquired at W4 after an observed shift, while C03 shows T after a failed gap. Both are distinct from C01's uninterrupted retention.

| Follow-up | Observed coverage | Exact target among observed | Continuous from W1 | Other-owned shift | Target absent | Unobserved | Reacquired / after gap |
| --- | --- | --- | --- | --- | --- | --- | --- |
| W2 | 4/4 = 100.0% | 4/4 = 100.0% | 4 | 0 | 0 | 0 | 0 / 0 |
| W3 | 3/4 = 75.0% | 2/3 = 66.7% | 2 | 1 | 0 | 1 | 0 / 0 |
| W4 | 3/4 = 75.0% | 3/3 = 100.0% | 1 | 0 | 0 | 1 | 1 / 1 |


## Which state rules prevent favorable denominator drift?

Apply status cumulatively from W1 through the stated window; every W1 cohort cell must occupy exactly one status.

| Status | Rule | Later treatment |
| --- | --- | --- |
| Confirmed continuous | Exact T01 citation in every planned window from W1 through the stated window | Remains confirmed only while the uninterrupted sequence continues |
| Confirmed lost | At least one eligible observed later window lacks T01 | Loss remains visible even if T01 returns later |
| Indeterminate | No confirmed loss, but a failed, unavailable, unclassifiable, or context-drifted window exists | A later target citation cannot backfill the gap |
| Later incident | Cell was not target-cited at W1 but becomes target-cited later | Report separately; never add it to the W1 cohort |


## How are failures and a zero inception denominator reported?

Failed, unavailable, observed-unclassifiable, and context-drifted windows never become observed losses. They make continuous persistence indeterminate from that point because the planned sequence cannot be confirmed.

If W1 contains zero exact-target citations, report ‘persistence not available: no inception cohort.’ The undefined zero-cohort result must not render as 0%, and later incident citations must not be used to reconstruct W1.


## What can RankEcho report and what remains a companion method?

RankEcho can retain configured prompt observations, sources, and shipment-linked later runs under its product definitions. The broader proof-report template can document planned windows, raw evidence, losses, failures, and interpretation limits.

RankEcho does not currently compute or publish this exact-target-URL inception-cohort persistence classification, its absorbing confirmed-loss rule, or its confirmed/lost/indeterminate table. Attach the companion cell-and-window ledger; do not present it as a native product metric or import format.

The Proof Ledger product held rule collapses engines within each run, counts any target citation, and can remain held when at least half of later run rollups are cited, so intermittent uncited runs may still qualify. This manual method instead keeps exact URL identity, consecutive planned windows, reacquisition, and observation gaps explicit.

The Q3 report's citation stability is full cited-domain-set Jaccard overlap across two matched runs. It is not exact-target URL retention in an anchor-positive post-change cohort.

- [Open the proof report framework](https://rankecho.io/templates/proof-report)
- [Review the product Proof Loop](https://rankecho.io/product/proof-loop)

## What interpretation is supported?

One of four W1 target-cited cells remains continuously confirmed through W4; one has a confirmed loss and two are indeterminate because of observation gaps. Two additional cells first cite T01 in W2 and remain outside the W1 inception cohort.

These are descriptive post-shipment observations. They do not show that the internal links caused a citation, estimate long-run durability, reveal a provider mechanism, or promise a later citation, recommendation, ranking, referral, conversion, or revenue result.


## What deterministic validation completes the persistence record?

Another person must be able to derive the cohort and every cumulative status directly from the complete observation ledger.

| Check | Pass condition | Evidence retained |
| --- | --- | --- |
| Chronology | Shipment precedes W1; W1–W4 are strictly ordered | UTC shipment and collection timestamps |
| Cell identity | All 24 records use the same six prompt-surface-context IDs | Frozen prompt and context registry |
| Target identity | Every T state resolves to exact normalized T01 | Original URL, resolved URL, UI role, and capture |
| State evidence | Each state has a unique raw answer or failure pointer | Complete 6 × 4 ledger |
| Cohort | Only C01–C04 are T at W1; C05–C06 remain later incidents | Mechanical inception projection |
| Partition | Every cohort cell occupies one state and rows reconcile 4/0/0, 4/0/0, 2/1/1, 1/1/2 | Cumulative status projection |
| Zero cohort | No W1 positives returns not available, never 0% | Null-display test |
| Interpretation | No causal, durability, traffic, ranking, or product-computation claim | Semantic check |


## When should monitoring continue or stop?

Complete the predeclared schedule regardless of direction. After W4, choose a new prospective schedule only if the decision still warrants it; preserve the original cohort and do not rewrite its endpoint.

A loss can return the team to Find, while an indeterminate result can justify repairing collection. Neither state authorizes selective deletion or a claim that persistence failed across all future windows.

- [Return to the owner diagnosis](https://rankecho.io/resources/find/issues/wrong-page-cited)
- [Review the deployed graph](https://rankecho.io/resources/fix/internal-link-cluster)

## Who maintains this guide?

Written and maintained by Abiot Y. Derbie. Published 2026-09-07; sources checked 2026-09-07; last updated 2026-09-07. The persistence record is synthetic.

Recheck provider source displays, product reporting definitions, cadence, and evidence captures before applying the method. Send material corrections through the contact page.

- [Send a correction](https://rankecho.io/contact)
- [Review measurement definitions](https://rankecho.io/methodology)

## Is citation persistence the same as citation rate?

No. Citation rate summarizes positives among eligible cells in a window. Persistence follows a frozen cohort that cited one exact target URL at its inception window.

## Does a failed window count as a citation loss?

No. It makes continuous confirmation indeterminate because the planned observation is missing; it is not an observed negative.

## Can a later target citation erase an earlier loss?

No. The return is retained, but the original continuously persistent sequence ended at the observed loss.

## Can later-positive cells join the original cohort?

No. Report them as later incidents or start a separately declared cohort; never add them retroactively to W1.

## Does persistence prove that the internal-link cluster worked?

No. It reports ordered observations after the shipment without isolating the link change from answer variability, provider changes, indexing changes, or other sources.

## Does RankEcho calculate this persistence table?

No. This exact-target inception-cohort method requires a companion ledger; current RankEcho product outputs use different records and units.

## Preserved dated evidence

- [Bing Webmaster Blog: AI Performance](https://blogs.bing.com/webmaster/February-2026/Introducing-AI-Performance-in-Bing-Webmaster-Tools-Public-Preview) — checked 2026-09-07. Bing's public-preview AI Performance report exposes citation counts, cited pages, trends, and sampled grounding-query phrases while warning that citation count does not indicate placement, ranking, authority, or page importance. Public preview announced February 10, 2026 Primary-source documentation check; Bing reports provider aggregates, not this guide's prompt-level owner classification or exact-URL persistence cohort. Confidence: High

- [Google Search: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) — checked 2026-09-07. Google says AI Overviews and AI Mode can use query fan-out, different models and techniques, and varying responses and links; normal Search eligibility applies, but eligibility does not guarantee crawling, indexing, serving, or appearance. Google Search documentation last updated December 10, 2025 Primary-source documentation check; the guidance does not establish that an internal-link change causes an AI citation or that a sampled citation will persist. Confidence: High

- [OpenAI API: evaluation best practices](https://developers.openai.com/api/docs/guides/evaluation-best-practices) — checked 2026-09-07. OpenAI says generative output is variable and recommends task-specific, logged, repeated or continuous evaluation rather than impressionistic assessment. Current OpenAI developer evaluation guidance Developer and API evaluation guidance is not a consumer ChatGPT Search protocol and does not define exact-URL persistence, a cadence, a threshold, or a causal design. Confidence: High

- [NIST AI 600-1: Generative AI Profile](https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.600-1.pdf) — checked 2026-09-07. NIST calls for documented measurement context and limitations and recommends ongoing monitoring of generative-AI sources and citations. NIST AI 600-1, July 2024 Government risk-management guidance; it is not a search-provider protocol and does not prescribe this persistence state model, window count, threshold, or causal interpretation. Confidence: High

## Complete fictional source records

These are the original fictional Alderwave Relay records, not real captures, customer observations or deployment receipts. Evidence pointers label the teaching records; replace them with your retained evidence before use.

```json
{
  "pages": [
    {
      "id": "T01",
      "url": "https://alderwave.example/platform/supplier-risk",
      "path": "/platform/supplier-risk",
      "role": "intended-owner",
      "userJob": "Explain the fictional Alderwave Relay supplier-risk monitoring product.",
      "version": "synthetic-target-v4",
      "evidencePointer": "synthetic-evidence/pages/T01-v4"
    },
    {
      "id": "H01",
      "url": "https://alderwave.example/platform",
      "path": "/platform",
      "role": "ancestor-hub",
      "userJob": "Orient readers across the fictional platform modules.",
      "version": "synthetic-hub-v3",
      "evidencePointer": "synthetic-evidence/pages/H01-v3"
    },
    {
      "id": "W01",
      "url": "https://alderwave.example/blog/supplier-risk-checklist",
      "path": "/blog/supplier-risk-checklist",
      "role": "competing-page",
      "userJob": "Provide a fictional supplier-risk checklist rather than own the product prompts.",
      "version": "synthetic-checklist-v6",
      "evidencePointer": "synthetic-evidence/pages/W01-v6"
    },
    {
      "id": "S01",
      "url": "https://alderwave.example/docs/review-evidence",
      "path": "/docs/review-evidence",
      "role": "supporting-spoke",
      "userJob": "Document fictional review-evidence fields used by the product workflow.",
      "version": "synthetic-docs-v2",
      "evidencePointer": "synthetic-evidence/pages/S01-v2"
    }
  ],
  "promptOwners": [
    {
      "id": "P01",
      "prompt": "What supplier-risk signals does Alderwave Relay monitor?",
      "intendedOwnerId": "T01",
      "ownershipState": "resolved",
      "ownerReason": "The product page owns the product-scope question; the checklist and documentation support narrower tasks.",
      "registrationVersion": "synthetic-owner-map-v1",
      "evidencePointer": "synthetic-evidence/owners/P01"
    },
    {
      "id": "P02",
      "prompt": "How does Alderwave Relay document a supplier-risk review?",
      "intendedOwnerId": "T01",
      "ownershipState": "resolved",
      "ownerReason": "The product page owns the workflow overview and links to field-level documentation.",
      "registrationVersion": "synthetic-owner-map-v1",
      "evidencePointer": "synthetic-evidence/owners/P02"
    },
    {
      "id": "P03",
      "prompt": "What is included in Alderwave Relay supplier-risk monitoring?",
      "intendedOwnerId": "T01",
      "ownershipState": "resolved",
      "ownerReason": "The product page owns the included-capabilities question; the hub remains broader.",
      "registrationVersion": "synthetic-owner-map-v1",
      "evidencePointer": "synthetic-evidence/owners/P03"
    }
  ],
  "contexts": [
    {
      "id": "CTX-A",
      "surface": "Surface A",
      "serviceLabel": "Fictional answer surface A",
      "locale": "en-US",
      "accountState": "signed-out synthetic session",
      "modelLabel": "not exposed",
      "citationRoleRule": "Count only links the fictional interface marks as citations or supporting sources; exclude related and suggested links.",
      "urlNormalizationRule": "Resolve redirects, remove fragments and declared tracking parameters, lowercase the host, preserve the path, and compare the resulting exact URL.",
      "collectionMode": "One scheduled manual capture per prompt and window."
    },
    {
      "id": "CTX-B",
      "surface": "Surface B",
      "serviceLabel": "Fictional answer surface B",
      "locale": "en-US",
      "accountState": "signed-out synthetic session",
      "modelLabel": "not exposed",
      "citationRoleRule": "Count only links the fictional interface marks as citations or supporting sources; exclude related and suggested links.",
      "urlNormalizationRule": "Resolve redirects, remove fragments and declared tracking parameters, lowercase the host, preserve the path, and compare the resulting exact URL.",
      "collectionMode": "One scheduled manual capture per prompt and window."
    }
  ],
  "cells": [
    {
      "id": "C01",
      "promptId": "P01",
      "contextId": "CTX-A",
      "surface": "Surface A",
      "state": "observed",
      "evidencePointer": "synthetic-evidence/baseline/C01-answer",
      "stateReason": "Eligible answer and all qualifying source locations captured."
    },
    {
      "id": "C02",
      "promptId": "P01",
      "contextId": "CTX-B",
      "surface": "Surface B",
      "state": "observed",
      "evidencePointer": "synthetic-evidence/baseline/C02-answer",
      "stateReason": "Eligible answer and all qualifying source locations captured."
    },
    {
      "id": "C03",
      "promptId": "P02",
      "contextId": "CTX-A",
      "surface": "Surface A",
      "state": "observed",
      "evidencePointer": "synthetic-evidence/baseline/C03-answer",
      "stateReason": "Eligible answer and all qualifying source locations captured."
    },
    {
      "id": "C04",
      "promptId": "P02",
      "contextId": "CTX-B",
      "surface": "Surface B",
      "state": "observed",
      "evidencePointer": "synthetic-evidence/baseline/C04-answer",
      "stateReason": "Eligible answer and all qualifying source locations captured."
    },
    {
      "id": "C05",
      "promptId": "P03",
      "contextId": "CTX-A",
      "surface": "Surface A",
      "state": "observed",
      "evidencePointer": "synthetic-evidence/baseline/C05-answer",
      "stateReason": "Eligible answer and all qualifying source locations captured."
    },
    {
      "id": "C06",
      "promptId": "P03",
      "contextId": "CTX-B",
      "surface": "Surface B",
      "state": "failed",
      "evidencePointer": "synthetic-evidence/baseline/C06-failure",
      "stateReason": "Synthetic collection failure before an eligible answer was preserved."
    }
  ],
  "shippedAt": "2026-08-03T16:00:00Z",
  "windows": [
    {
      "id": "W1",
      "collectedAt": "2026-08-10T14:00:00Z",
      "relationToShipment": "First predeclared post-shipment window and inception window"
    },
    {
      "id": "W2",
      "collectedAt": "2026-08-17T14:00:00Z",
      "relationToShipment": "Second predeclared post-shipment window"
    },
    {
      "id": "W3",
      "collectedAt": "2026-08-24T14:00:00Z",
      "relationToShipment": "Third predeclared post-shipment window"
    },
    {
      "id": "W4",
      "collectedAt": "2026-08-31T14:00:00Z",
      "relationToShipment": "Fourth predeclared post-shipment window"
    }
  ],
  "observations": [
    {
      "cellId": "C01",
      "windowId": "W1",
      "state": "target-cited",
      "evidencePointer": "synthetic-evidence/persistence/C01-W1",
      "detail": "Eligible answer cites the exact normalized T01 URL."
    },
    {
      "cellId": "C01",
      "windowId": "W2",
      "state": "target-cited",
      "evidencePointer": "synthetic-evidence/persistence/C01-W2",
      "detail": "Eligible answer cites the exact normalized T01 URL."
    },
    {
      "cellId": "C01",
      "windowId": "W3",
      "state": "target-cited",
      "evidencePointer": "synthetic-evidence/persistence/C01-W3",
      "detail": "Eligible answer cites the exact normalized T01 URL."
    },
    {
      "cellId": "C01",
      "windowId": "W4",
      "state": "target-cited",
      "evidencePointer": "synthetic-evidence/persistence/C01-W4",
      "detail": "Eligible answer cites the exact normalized T01 URL."
    },
    {
      "cellId": "C02",
      "windowId": "W1",
      "state": "target-cited",
      "evidencePointer": "synthetic-evidence/persistence/C02-W1",
      "detail": "Eligible answer cites the exact normalized T01 URL."
    },
    {
      "cellId": "C02",
      "windowId": "W2",
      "state": "target-cited",
      "evidencePointer": "synthetic-evidence/persistence/C02-W2",
      "detail": "Eligible answer cites the exact normalized T01 URL."
    },
    {
      "cellId": "C02",
      "windowId": "W3",
      "state": "other-owned-only",
      "evidencePointer": "synthetic-evidence/persistence/C02-W3",
      "detail": "Eligible answer cites one or more owned URLs but not T01."
    },
    {
      "cellId": "C02",
      "windowId": "W4",
      "state": "target-cited",
      "evidencePointer": "synthetic-evidence/persistence/C02-W4",
      "detail": "Eligible answer cites the exact normalized T01 URL."
    },
    {
      "cellId": "C03",
      "windowId": "W1",
      "state": "target-cited",
      "evidencePointer": "synthetic-evidence/persistence/C03-W1",
      "detail": "Eligible answer cites the exact normalized T01 URL."
    },
    {
      "cellId": "C03",
      "windowId": "W2",
      "state": "target-cited",
      "evidencePointer": "synthetic-evidence/persistence/C03-W2",
      "detail": "Eligible answer cites the exact normalized T01 URL."
    },
    {
      "cellId": "C03",
      "windowId": "W3",
      "state": "failed",
      "evidencePointer": "synthetic-evidence/persistence/C03-W3",
      "detail": "Collection failed before an eligible answer was preserved."
    },
    {
      "cellId": "C03",
      "windowId": "W4",
      "state": "target-cited",
      "evidencePointer": "synthetic-evidence/persistence/C03-W4",
      "detail": "Eligible answer cites the exact normalized T01 URL."
    },
    {
      "cellId": "C04",
      "windowId": "W1",
      "state": "target-cited",
      "evidencePointer": "synthetic-evidence/persistence/C04-W1",
      "detail": "Eligible answer cites the exact normalized T01 URL."
    },
    {
      "cellId": "C04",
      "windowId": "W2",
      "state": "target-cited",
      "evidencePointer": "synthetic-evidence/persistence/C04-W2",
      "detail": "Eligible answer cites the exact normalized T01 URL."
    },
    {
      "cellId": "C04",
      "windowId": "W3",
      "state": "target-cited",
      "evidencePointer": "synthetic-evidence/persistence/C04-W3",
      "detail": "Eligible answer cites the exact normalized T01 URL."
    },
    {
      "cellId": "C04",
      "windowId": "W4",
      "state": "unavailable",
      "evidencePointer": "synthetic-evidence/persistence/C04-W4",
      "detail": "The declared surface was unavailable; no eligible answer was captured."
    },
    {
      "cellId": "C05",
      "windowId": "W1",
      "state": "other-owned-only",
      "evidencePointer": "synthetic-evidence/persistence/C05-W1",
      "detail": "Eligible answer cites one or more owned URLs but not T01."
    },
    {
      "cellId": "C05",
      "windowId": "W2",
      "state": "target-cited",
      "evidencePointer": "synthetic-evidence/persistence/C05-W2",
      "detail": "Eligible answer cites the exact normalized T01 URL."
    },
    {
      "cellId": "C05",
      "windowId": "W3",
      "state": "target-cited",
      "evidencePointer": "synthetic-evidence/persistence/C05-W3",
      "detail": "Eligible answer cites the exact normalized T01 URL."
    },
    {
      "cellId": "C05",
      "windowId": "W4",
      "state": "target-cited",
      "evidencePointer": "synthetic-evidence/persistence/C05-W4",
      "detail": "Eligible answer cites the exact normalized T01 URL."
    },
    {
      "cellId": "C06",
      "windowId": "W1",
      "state": "failed",
      "evidencePointer": "synthetic-evidence/persistence/C06-W1",
      "detail": "Collection failed before an eligible answer was preserved."
    },
    {
      "cellId": "C06",
      "windowId": "W2",
      "state": "target-cited",
      "evidencePointer": "synthetic-evidence/persistence/C06-W2",
      "detail": "Eligible answer cites the exact normalized T01 URL."
    },
    {
      "cellId": "C06",
      "windowId": "W3",
      "state": "target-cited",
      "evidencePointer": "synthetic-evidence/persistence/C06-W3",
      "detail": "Eligible answer cites the exact normalized T01 URL."
    },
    {
      "cellId": "C06",
      "windowId": "W4",
      "state": "target-cited",
      "evidencePointer": "synthetic-evidence/persistence/C06-W4",
      "detail": "Eligible answer cites the exact normalized T01 URL."
    }
  ]
}
```

## Related reading

- [Prove resource selector](https://rankecho.io/resources/prove)
- [Wrong-page diagnosis](https://rankecho.io/resources/find/issues/wrong-page-cited)
- [Internal-link cluster](https://rankecho.io/resources/fix/internal-link-cluster)
- [Fixed-prompt retest](https://rankecho.io/resources/prove/fixed-prompt-retest)
- [Citation rate](https://rankecho.io/resources/prove/citation-rate)
- [Answer volatility](https://rankecho.io/resources/prove/answer-volatility)

## Product next step

Attach this exact URL and ordered-window record to the proof report. RankEcho's Proof Loop and Proof Ledger use separate product rules and do not calculate this inception-cohort persistence table. https://rankecho.io/templates/proof-report