# Agency playbook

> How agencies run AI visibility for clients with GetIntel, from onboarding a client and setting a baseline to reporting in Looker Studio and showing progress.

Source: https://getintel.ai/docs/academy/playbooks/agency-playbook/

This playbook covers running GetIntel across a roster of clients. See [white-label AI visibility for agencies](/solutions/agencies/) for how the account is built for this.

## Onboard a client

<Steps>

1. **Add the client.** In the brand switcher, choose **+ Add client** and complete the setup with the client's website, product, audience and geography. See [Managing clients](/docs/agencies/managing-clients/).
2. **Review competitors with the client.** Clients know their real rivals. Remove wrong ones, add missing ones.
3. **Check the questions.** Make sure the buyer questions match how the client's customers really search. Add any that are missing.
4. **Record the baseline** after 7 days: every topic's Found and win rate.

</Steps>

## Report

- **Looker Studio:** generate one key for your whole account, then create a report per client from the template. Each report asks which client it covers. See [Looker Studio](/docs/looker-studio/).
- **What to show:** topic scores against baseline, wins and losses by topic, the sources that started citing the client, and the work shipped.

## Run the work

- **One weekly plan per client.** Switch to the client and work their **Weekly plan** with the [weekly routine](/docs/academy/take-action/weekly-routine/).
- **Connect agents per client.** An MCP token only sees the client it was created for, so create one per client for your team's agents.
- **Invite your team.** Team members see every client on the account. See [Team members](/docs/agencies/team/).

## Show progress

Clients respond to evidence. Pair each report's numbers with one or two real AI answers from **Prompts**: an answer that named a rival before, and names the client now.