# Data dictionary

| Column | Type / values | Meaning |
|---|---|---|
| `ticket_id` | Text, unique synthetic ID | Identifier for one support ticket; not linked to a real order or person. |
| `created_at_utc` | ISO-8601 UTC timestamp | Time the support ticket was created. |
| `contact_reason` | Category | Main reason recorded for the contact: tracking update, delayed delivery, failed delivery attempt, damaged parcel, address change, payment question, missing item, or other. |
| `channel` | Category | Contact channel: web form, email, phone, or chat. |
| `region` | Category; sometimes blank | Broad fictional operating region. Blank means not captured. It is not a precise location. |
| `parcel_stage` | Category | Parcel stage recorded when the ticket was created: before dispatch, in transit, out for delivery, delivered, exception, or returned. |
| `days_after_promised_date` | Decimal number; sometimes blank | Approximate days after the promised delivery date at ticket creation. Zero or a positive value is used when recorded; blank can mean not applicable or not captured. Do not assume blank equals zero. |
| `status` | `resolved`, `pending`, or `escalated` | Ticket status in this synthetic snapshot. |
| `first_response_minutes` | Decimal number; sometimes blank | Approximate elapsed minutes from ticket creation to first response. Blank means unavailable. |
| `resolution_time_hours` | Decimal number; sometimes blank | Approximate elapsed hours from ticket creation to resolution. Pending tickets have no resolution time; other missing values may also be unrecorded. |
| `escalated` | `yes` or `no` | Whether the ticket is marked escalated. This is related to, but not identical in meaning to, the current `status` field. |
| `customer_rating_1_to_5` | Integer 1–5; often blank | Optional post-resolution rating. Missing ratings are not zero and may not be representative of all customers. |

## Analysis guidance

- Check row counts, date coverage, category spelling, duplicate IDs, and missingness before charting.
- Suitable summaries include ticket volume by day/week, contact reason, channel, parcel stage, and status; response/resolution distributions; and rating counts with the number of missing ratings shown.
- When comparing groups, show counts alongside averages or percentages. Small groups may be unstable.
- A pattern or association does not establish a cause. The data is synthetic and lacks operational context such as route, weather, staffing, or actual delivery scans.
