# Miami Shortlist Index: methodology and evidence-status appendix
Edition 1: 23 October 2026. Published by SignalTide.

## Scope as reported in the original index
31 independently owned South Beach hotels, between 1st and 23rd Street. Six consumer-facing systems: Google AI Mode, Gemini, ChatGPT, Grok, Perplexity and Copilot. Three exact prompts:
1. boutique hotels in miami beach
2. best hotels in south beach
3. south beach art deco hotels

The study reports logged-out tests from Michigan, with no account history, on 23 October 2026. It reports 15 of 31 hotels not named in these tested prompts on that date. This does not imply permanent absence or measure bookings.

## Evidence availability and verification limits
The underlying raw response archive, individual answer timestamps, model/version identifiers, hotel inclusion worksheet, and original screenshots were **not available for independent verification in the current site source** when this appendix was prepared. We do not fabricate them. The date above is the reported study date, not an invented per-response timestamp. Platform interfaces are identified by product name only; exact versions are unverified.

## Inclusion rules and limitations
The reported geography and independent-ownership criteria are stated above. Property-by-property ownership verification and the original selection worksheet have not been supplied for independent review. Tests used only three phrases, one date, and a Michigan search location. AI outputs are variable, potentially personalized, and not a representative sample of traveler booking behavior.

## Anonymized response archive
Not published: original response transcripts and consent/privacy review are pending. No reconstructed or synthetic responses are presented as observations.

## Replication protocol for next edition
Record each hotel, ownership source, inclusion decision, query, platform/interface, account/login state, locale, search location, timestamp with timezone, full response text, citations, screenshot identifier, and whether the property was merely mentioned or explicitly recommended. Run repeat trials on multiple dates; retain original evidence before reporting new results.

## Interpretation
A non-mention is limited to the tested prompts and date. Changes in AI responses do not independently establish causation, incremental bookings, or revenue. Hotel and spa results must not be conflated.
