START WITH THE PRODUCT STEPS
Read START-HERE.txt first. It states the actual source acquisition, access/privacy limits, manual field mapping and exact next prompt. The original file templates and synthetic examples below are supporting review aids. Some paths need separately supplied documents or source text; a blank worksheet is not a native export or completed integration.

G15 - Prepare evidence about a disputed Google review
Status: Proposed / not tested

WHAT THIS KIT DOES
Provides exact-header blank CSVs, separate fabricated examples, required text-input scaffolds, and expected output examples for one offline review. It does not configure an integration or run a workflow.

PREREQUISITES
1. GBP access or public review URL, original screenshots and timestamp, approved incident-log storage, attorney confidentiality review. No client list needs to leave the firm.
2. Use synthetic inputs first. Real records require firm-approved access, processing terms, retention, and confidentiality controls. IDs alone do not anonymize a record.
3. Verify product plan, available exports, coverage, and the named reviewer. The prompt does not configure an integration.
Name a reviewer and keep actual prospect/client data within firm-approved systems. IDs alone are not anonymization.

INPUTS
- reviews.csv: review_id, public_url, posted_at, captured_at, review_text
- evidence.csv: evidence_id, review_id, nonconfidential_observation, source_ref, captured_at
- report-history.csv: review_id, status, submitted_at, decision_ref, appeal_used; platform-policy.txt: current official policy categories and source references, evidence requirements and report/appeal limits

OUTPUT SHAPE
review-evidence.csv: review_id, policy_category, supporting_evidence_refs, missing_evidence, assessment, current_report_status, human_next_step; report-draft.txt: factual nonpublic draft for authorized review

SETUP AND FIRST TRIAL
1. Keep the original exports unchanged. Copy this kit into an approved working folder.
2. Start with only the files ending -example. They contain fabricated data. CSV files without -example contain headers only; text inputs without -example contain empty preparation scaffolds.
3. For the synthetic trial, copy each example into a separate trial folder under its canonical input filename. Do not mix example rows with real exports. Supply the matching example policies too. Output examples are for comparison after drafting, not evidence inputs.
4. Import CSVs as UTF-8 comma-delimited data. Preserve identifiers as text and date/timestamp strings in their specified ISO format. Boolean examples use true/false. Empty cells mean unknown or not supplied, never automatic zero/no. Amounts are numeric; use the stated currency and do not combine currencies. No formulas or macros are included.
5. Paste starter-prompt.txt into the firm's approved assistant with only the input files. A staff member can also perform the comparison in a spreadsheet. No account connector is needed for this file-based trial.
6. Check the draft against fixture-expectations.txt and the matching output examples. Use exact arithmetic in native spreadsheet tools or code and inspect source IDs manually. An AI response has not been run or validated for this kit.
7. Resolve missing policies, uncertainty, permissions and coverage with the named reviewer. Live platform configuration, messages, uploads and other changes require their own approval and testing.

FIXTURE TARGET
Neither review supports a verified policy violation. REV1 is negative sentiment only; REV2 has missing corroboration and an already-used appeal. No public reply, fake-review accusation or duplicate appeal.

LIMITS
- Negative sentiment alone is not grounds for removal. Do not solicit mass reports or disclose representation/matter details in a public response. No public reply generated here, avoiding the existing review-draft workflow.
- Only file/header and synthetic fixture consistency are checked. No workflow, connector, AI response, product feature or integration is tested.
- Every example is fabricated. IDs alone do not anonymize actual firm data. Use approved minimized data, access, processing terms and retention controls.
Source-policy text scaffolds are not legal advice or production configuration. No real review link, phone destination, contact address, API token or credential is included in examples. example.invalid and fixture: references are deliberate non-live placeholders.

SOURCES
Pain evidence: Public discussion: https://www.reddit.com/r/LawFirm/comments/1by2woo
Method: Official method documentation: https://support.google.com/business/answer/4596773?hl=en-419
Method: Official method documentation: https://support.google.com/business/answer/7400114?hl=en

VALIDATION SCOPE
Header and fixture validation passed; workflow not tested. Read the exact checks and limits in fixture-expectations.txt.
