Reports & Export

Use the simplest output that matches the audience. A client may need a scheduled branded report. An analyst may need CSV. A dashboard may need a Sheets or Looker connection. Software and AI assistants need the API or MCP.

Report delivery

Brand Reports keep completed reporting history and support client-ready output. Agency Reports adds portfolio schedules and a history tab across client brands.

A recurring agency schedule can set the client, cadence, reporting window, recipients, included sections, white-label link, attached PDF, and approval before send. A one-off report can be requested from the Agent and then found in report history.

Report templates now support a Results section. It gives the audience measured outcomes alongside visibility, competitor, citation, and recommendation sections. Keep the result wording honest: Earned, No movement, Couldn't measure, Reverted, or Coincided with a drop. A report must not turn timing into a claim that an Action caused a change.

CSV and JSON

Use an on-screen export for a one-time analysis, board pack, or client attachment. The export shape and available formats depend on the page you start from.

Before downloading, check:

Brand or portfolio scope.
Date range.
Active model, tag, status, or search filters.
Whether the page exports all filtered rows or only selected rows.
Whether client export permission is on if the request comes from a portal user.

JSON is better when software needs nested data. CSV is better for a person working in a spreadsheet.

Google Sheets and Looker Studio

Google Sheets sync suits teams that build working tables and charts in Sheets. Give the service account editor access to the target sheet, verify the connection, and choose the sync cadence available to the account.

Treat Trakkr-managed tabs as replaceable output. Put formulas, notes, and presentation charts on separate tabs so a refresh cannot overwrite them.

Looker Studio suits reusable BI dashboards. It uses API access and reflects the latest completed Trakkr data when the dashboard queries it. “Live” means live access to the newest saved data, not a new AI check each time someone opens the dashboard.

REST API and MCP

The REST API is for software. MCP presents Trakkr as tools an AI assistant can call. The current endpoint reference, authentication guide, limits, and examples live at /learn/api.

Four new REST endpoints expose Suggested actions, decisions, Results, and Pages:

Method	Endpoint	What it does	MCP tool
GET	`/get-opportunity-pool`	Lists Suggested actions waiting for a decision	`list_opportunity_pool`
POST	`/commit-opportunity`	Commits, dismisses, or snoozes one suggestion	`commit_opportunity`
GET	`/get-results`	Lists measured Results for completed work	`get_results`
GET	`/get-pages`	Lists Pages for the brand	`list_pages`

The endpoint and tool names are compatibility names. In the product and your client copy, call the first collection Suggested actions and call measured outcomes Results.

Cursor pagination

These four endpoints use cursor pagination rather than the offset pattern used by some older endpoints.

Send limit and no cursor on the first request.
Read meta.next_cursor from the response.
Pass that opaque value as cursor on the next request.
Stop when next_cursor is null.

Do not parse the cursor or turn it into an offset. Save and return it exactly as received.

Suggested actions and decisions

GET /get-opportunity-pool lists suggestions for one brand. It can narrow by work type, kind, impact, and cursor. These are not committed Actions yet.

POST /commit-opportunity makes one explicit decision: commit, dismiss, or snooze. A commit creates tracked work and fixes the before-and-after measurement setup. Dismissal can include a reason. Snooze accepts a bounded number of days.

The write does not raise an Automation's permission. Permissions remain controlled per brand in Trakkr.

Results

GET /get-results returns measured outcomes. Its verdict filters use API values such as earned, no_change, harm, and couldnt_measure; map those to the exact user words in your interface. The old /get-proof endpoint and get_proof MCP tool remain as deprecated compatibility aliases.

The response says what happened during the measurement window. It does not prove that the Action caused it. Preserve that distinction in any client report or alert built on the endpoint.

Pages

GET /get-pages returns one row per Page and can narrow by ownership or tracked state. A Page may include its current journey bottleneck and latest verdict. Use the product journey words when displaying stages: Available, Reached, Understood, Relevant, Selected, and Visited.

Authentication and plan access

REST API access is Scale-equivalent or granted by an explicit API entitlement. MCP is available on paid plans. Managed contracts can set different limits.

Create personal credentials in Settings: Developer. Each teammate should use their own connect token or API key so access can be removed without breaking everyone else's integration.

Choose an output

Need	Best starting point
Recurring client update	Agency Reports schedule
One-off client PDF	Ask Agent to generate, then use report history
Quick analysis	CSV or JSON
Shared working model	Google Sheets
Reusable BI dashboard	Looker Studio
Warehouse or internal software	REST API
AI assistant access	MCP

Common questions

Why does an export differ from the Dashboard?

Check brand, date range, report freshness, and filters first. An export can be correct for one window while the Dashboard is showing another.

Why is a client unable to export?

The shared portal Allow data export setting and the client's Can export data permission must both be on.

Does the API calculate new data on request?

Most read endpoints return saved data from completed checks and reports. Freshness follows the source's collection cadence.

Should an integration display API verdict values directly?

No. Map them to the product's exact result words and preserve the no-causation rule.

Where is the full endpoint reference?

Open /learn/api for parameters, response examples, authentication, and the interactive playground.