Skip to documentation content
One workspace for search and AI visibility.
DocumentationGetting started

AI Marketer

Ask a focused question, inspect the evidence, and turn a reviewed recommendation into your next marketing step.

Product guide4 min read

Start with a useful question

Use AI Marketer to investigate available project data, explain findings, and prepare content proposals. Start with a decision you need to make, not a broad request to “improve marketing.” Connected data and available tools depend on your workspace and permissions.

Start your first conversation

  1. 1

    Open AI Marketer

    Choose AI Marketer in the sidebar. You can also use the floating AI Marketer button on supported dashboard pages to open a conversation panel.

  2. 2

    Give the task a clear scope

    Name the project, audience, language, period, and desired output. Include relevant source text or an image if needed. When attaching an image, wait for its preview before sending; do not include credentials or private customer data.

  3. 3

    Send, then inspect

    Press Enter to send, or Shift+Enter for a new line. Read the response and any displayed evidence or tool activity. If the run is no longer useful, choose Stop; stopping does not undo actions already completed.

    You should see

    The response uses the intended project and scope, and makes missing evidence clear.

Try a focused request

Copy an example, replace the bracketed details, and paste it into AI Marketer. These are starting points, not promises of a particular result. Ask the assistant to state what data is unavailable.

Investigate AI visibility

For [project] in [market], summarize the available AI visibility results for [period]. Separate mentions from citations, identify competitor gaps, and show the supporting answers and source URLs. State any missing data and separate observations from hypotheses. Do not make changes.

What to look for

A scoped comparison you can check against collected answers. Verify the project, period, engines, and cited URLs before choosing an action.

A useful follow-up

Which three checks would help validate these hypotheses?

Understand a traffic change

Compare [project]’s available search and website traffic for [period] with the previous equivalent period. Identify the pages or queries driving the change, name the data source for each finding, and suggest three checks. If GSC or GA4 data is unavailable, say so rather than estimating it. Do not make changes.

What to look for

An explanation tied to the correct connected property and dates. Search clicks, website visits, AI referrals, and AI citations are different signals; do not treat them as interchangeable.

Prepare a content brief

Using this approved audience and source material: [context], propose a content brief about [topic] in [language]. Include reader intent, an outline, required sources, and claims needing verification. Return a proposal only. Do not create, schedule, or publish content.

What to look for

A reviewable brief, not a published article. Check brand facts, audience fit, and source requirements before moving the approved brief into your content workflow.

A useful follow-up

Which claims still need a primary source before we draft?

Review evidence and proposed actions

  • Open the underlying answer, page, or report when available. Check dates and sources, and verify important brand or product claims independently.
  • When an approval request appears, read the proposed action and its inputs before choosing Approve or Deny. Do not approve an action with an unclear destination or scope.
  • A reply is not proof that a change was saved, scheduled, or published. Confirm the resulting state in the relevant screen. AI recommendations do not guarantee rankings, citations, or revenue.

Continue without losing the thread

Use follow-up questions to narrow the same task. On the AI Marketer page, Select a chat opens Recent conversations; use New when you want a separate conversation. Restate the project and scope when switching to a different task.

If the assistant lacks context, provide the missing source or check the relevant integration. Do not mistake unavailable data for a zero value. Review conversation content and tool details for sensitive information before sharing an export.

A little guidance goes a long way.

Need help choosing a workflow or setting up your workspace? Talk to the Alef team.