# Thin vs Fat Events, Domain vs Integration Events — Event-Driven Architecture & CQRS

Source: https://www.skillbyai.com/en/event-driven-architecture/e-granularity

> Choose how much data an event carries and which events cross service boundaries.

## How much to put in an event

A **thin event** carries identifiers only (`OrderPlaced { orderId }`); consumers call the source for details. It keeps events small and data fresh, but couples consumers back to the source's API and availability, and can cause a burst of calls after every event. A **fat event** (event-carried state transfer) carries the data consumers need; they work independently and can build local copies, at the cost of larger events, a wider contract and copies of data that must be governed (including personal data). Also separate **domain events**, used inside one service's model and free to change, from **integration events**, published for other services, which are stable, versioned and deliberately designed. Do not publish every internal state change: publish the business facts other contexts actually need.

## Thin and fat versions of the same fact

Choose per use case; many systems settle in between.

```json
// thin: consumers must call GET /orders/o-1001
{ "type": "OrderShipped", "orderId": "o-1001" }

// fat: consumers have what they need
{
  "type": "OrderShipped",
  "orderId": "o-1001",
  "customerId": "c-42",
  "carrier": "BlueDart",
  "trackingNumber": "BD123456789IN",
  "shippedAt": "2026-10-03T09:10:00Z",
  "items": [{ "sku": "notebook", "qty": 3 }]
}
```

## Watch personal data in fat events

Events are copied into logs, topics, data lakes and consumer databases. Include personal data only when consumers need it, and plan how a deletion request reaches every copy.

**Quiz:** What is a drawback of thin events?

- [x] Consumers must call back to the source for details, coupling them to its availability
- [ ] They are always larger than fat events
- [ ] They cannot carry IDs
- [ ] They break schema registries

*Answer:* Consumers must call back to the source for details, coupling them to its availability. Thin events push consumers to query the source, reintroducing runtime coupling.