---
title: "The “Indexed” Outcome Category"
url: https://develop.sentry.dev/application-architecture/dynamic-sampling/outcomes/
---

# The “Indexed” Outcome Category

Data types which are subject to dynamic sampling are represented in two different outcome data categories: "total" and "indexed". You can think of the "total" category as counting items that are stored in *aggregated* form, in other words, as metrics. The "indexed" category counts items that are stored individually.

If an item is dropped *before* metrics extraction & dynamic sampling, it will show up as an outcome in both categories (for example, `"transaction"` and `"transaction_indexed"`). If an item is dropped *by* dynamic sampling, it will show up as a "filtered" outcome in the "indexed" category, but it lives on in the form of metrics, for which we emit an "accepted" outcome in the "total" category.

## [Illustration](https://develop.sentry.dev/application-architecture/dynamic-sampling/outcomes.md#illustration)

These charts illustrate the flow of data categories for transactions:

For a sampled transaction:

```mermaid
flowchart LR
  SDK --transaction--> Relay
  Relay --transaction--> DS[Dynamic Sampling]
  Relay --transaction_indexed--> DS
  DS --transaction-->Store
  DS --transaction_indexed-->Store
```

For a transaction filtered by dynamic sampling:

```mermaid
flowchart LR
  SDK --transaction--> Relay
  Relay --transaction--> DS[Dynamic Sampling]
  Relay --transaction_indexed--> DS
  DS --transaction-->Store
  DS --transaction_indexed-->Reject
  style Reject stroke:#f00
```

## [Data Types](https://develop.sentry.dev/application-architecture/dynamic-sampling/outcomes.md#data-types)

The following data categories have a corresponding `"*_indexed"` category:

* transactions
* spans
* profiles