> ## Documentation Index
> Fetch the complete documentation index at: https://docs.marketer.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Write effective requests

> Structure Marketer requests around an outcome, evidence, constraints, and a clear action boundary.

You do not need a long prompt. You need a request that removes ambiguity from the decision and from what Marketer may do next.

## Use the request formula

A strong request contains:

1. **Outcome** — the question to answer or artifact to produce
2. **Source** — the connected system that should provide the evidence
3. **Scope** — the account, campaign, product, audience, or workspace in bounds
4. **Window** — exact dates or completed periods
5. **Measures** — the metrics or quality criteria that matter
6. **Constraints** — definitions, exclusions, and business rules
7. **Output** — the format you want
8. **Action boundary** — read, draft, ask for approval, or execute under policy

## Start from these patterns

### Performance analysis

```text theme={null}
Compare [metrics] for [scope] across [current period] and [comparison period]
using [source]. Name the exact dates, timezone, and freshness. Rank the three
largest changes, explain the evidence, and do not make changes.
```

### Creative brief

```text theme={null}
Draft a creative brief for [product and audience] using [attached context and
Knowledge]. The goal is [outcome]. Include [required sections] and avoid
[claims or themes]. Keep it as a draft for review.
```

### Campaign preparation

```text theme={null}
Prepare a campaign plan for [channel, objective, product, and audience]. Apply
the workspace guardrails and show the proposed budget, targeting, creative,
and measurement plan. Do not publish. List every item that needs approval.
```

## Replace vague language

| Instead of            | Specify                                                      |
| --------------------- | ------------------------------------------------------------ |
| “recently”            | Exact dates or “last 7 complete days”                        |
| “best campaigns”      | The ranking metric and minimum spend or conversion threshold |
| “improve performance” | The metric to improve and the guardrails that must hold      |
| “our data”            | The provider, account, and attribution view                  |
| “launch this”         | Draft, paused launch, or active launch, plus who approves    |

## Ask for decision-ready output

Tell Marketer how you will use the result. A useful structure is:

* **Finding:** what changed
* **Evidence:** source, dates, scope, and values
* **Interpretation:** the most likely explanation and uncertainty
* **Recommendation:** the proposed next step
* **Approval:** what would change and what control applies

## Refine without restarting

Use follow-ups to correct one dimension at a time:

* “Use only Google Ads. Remove Shopify figures.”
* “Recalculate in the workspace timezone using complete days.”
* “Limit the answer to the attached campaign.”
* “Keep the recommendation, but turn it into a draft with no write actions.”
* “Show the source and calculation behind the first claim.”

<Note>
  A prompt cannot expand your role or bypass workspace controls. Guardrails, autonomy settings, provider permissions, and approval gates still apply.
</Note>

Before acting on the output, use [Review results](/essential-guidelines/review-results).


## Related topics

- [Templates and use cases](/templates-use-cases/overview.md)
- [MCP connect](/workspace-admin/mcp-connect.md)
- [Account overview](/workspace-admin/account-overview.md)
- [Chat and context](/working-with-marketer/chat-and-context.md)
- [Review your result](/get-started/review-result.md)