# Cloudbeds Data Retention Plan

## Goal

Ensure Cloudbeds-derived data is not just used once, but retained and grown over time for cumulative business intelligence.

## Retention layers

### 1. Raw file archive
Store every daily Cloudbeds XLSX here:
- `/Users/bobeagle/.openclaw/workspace/reports/cloudbeds/raw/`

Naming rule:
- `occupied-room-count-YYYY-MM-DD.xlsx`

### 2. Parsed structured archive
Store parsed daily JSON here:
- `/Users/bobeagle/.openclaw/workspace/reports/cloudbeds/parsed/`

Naming rule:
- `occupied-room-count-YYYY-MM-DD.json`

### 3. Human-readable report outputs
Store generated daily report artifacts here:
- `/Users/bobeagle/.openclaw/workspace/reports/cloudbeds/output/`
- published copies under `/Users/bobeagle/.openclaw/workspace/shared/reports/`

### 4. Durable markdown read log
Append every successful Cloudbeds read to a markdown log file:
- `/Users/bobeagle/.openclaw/workspace/CLOUDBEDS_READ_LOG.md`

## Markdown log rule

Every time Cloudbeds data is successfully read/parsed, append:
- read date/time
- source email/report reference if known
- source file path
- total occupied rooms if available
- property count if available
- top key notes
- whether the read was full or partial

This creates a durable narrative history in addition to raw/structured machine-readable storage.

## Why this matters

This gives three forms of historical growth:
1. raw original files
2. parsed structured data
3. human-readable audit trail

## Desired behavior going forward

Whenever Cloudbeds data is read:
- archive the source file if available
- parse into JSON if possible
- append a markdown entry to the read log
- use archived history for trends

## Long-term outcome

Over time, this should create a steadily improving occupancy history that can support:
- daily deltas
- weekly averages
- monthly trends
- quarterly trends
- anomaly detection
- executive reporting