> For the complete documentation index, see [llms.txt](https://docs.aloop.icustomer.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.aloop.icustomer.ai/concepts/decisions-and-traces.md).

# Decisions and traces

The decision is the unit of work, and the trace is its receipt. What counts as a decision, and what a trace records.

A **decision** is a moment where the system chose one option over others, spent something finite doing it, and said what it expected to happen. All three parts matter:

* Choosing without spending is analysis.
* Spending without stating an expectation is just activity.
* The decision is the unit of work. Everything else is a step along the way.

"Something finite" means anything there is a limited amount of and running out matters: credits, email sends, ad budget, a person's attention, a slot in a contact's outreach cadence, or room under your audience ceiling.

## Decision or step?

The test is simple: **it is a decision if it spends capacity, claims a record, or changes what is under management.** Everything else is a step, real work, worth recording, but nothing was at stake in choosing it.

| Example                                                            | Verdict                                                          |
| ------------------------------------------------------------------ | ---------------------------------------------------------------- |
| Enrich one record                                                  | A step                                                           |
| Choose *which* 1,200 records to enrich, instead of the other 3,000 | A decision                                                       |
| Run a query to build a list                                        | A step. The *criteria* were the decision                         |
| Import a list                                                      | A decision: it changes what you manage                           |
| Activate an audience to a channel                                  | A decision: it spends sends or budget, and claims those contacts |

## The trace is the receipt

A decision is a specific, auditable action, reversible where the channel allows. It is never just a recommendation. Every decision writes a **trace** before it runs:

* **Why** it was made, and on what evidence.
* **Under what authority**: proposed, approved by you, or approved by a policy you set.
* **For whom**, and what it expected to happen.
* **What came back.** When the outcome arrives, it attaches to the same trace.

Ask "why did we do that?" about anything, and the answer is on record. It was written before the action ran, never reconstructed afterward.

## The ordering

One arc runs from a signal or a question to an outcome, and it contains several decisions along the way: choose the audience, choose the channel, choose the budget. Each decision can produce activations, and each activation its steps. The outcome attaches to the decision that caused it.

## Expected versus actual

Every decision states what it expects: which number moves, by how much, by when, at what spend. The outcome arrives in the same shape, and a **learning is the subtraction**, actual minus expected, never an essay. That is what makes the system's judgment inspectable: where was it wrong, by how much, and in which direction.

The most important question this makes answerable is the one no dashboard asks: **what did we decide and never find out about?** An unmeasured decision is worse than a wrong one. A wrong decision teaches something. An unmeasured one spent something real and taught nothing, so the system surfaces these rather than burying them.

Where you read traces day to day is covered in [Traces and receipts](/measurement-and-outcomes/traces-and-receipts.md).


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:

```
GET https://docs.aloop.icustomer.ai/concepts/decisions-and-traces.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.