A campaign is a single round of analysis: Mentova sends your prompts to every selected AI model, collects the responses, and extracts the brands mentioned. Launching a campaign is the primary action that keeps your visibility data fresh.

When to launch

Launch a campaign whenever you want an up-to-date snapshot of how AI assistants talk about your brand. New campaigns layer on top of previous ones - your charts and KPIs update automatically after each run.

If you prefer automated runs, configure a schedule in Campaigns › Scheduling (STARTER and above — weekly at most on Starter, up to daily from PRO).

How to launch

  1. Open your brand in the dashboard.
  2. Click the Launch analysis button in the header's top-right area — it appears on every dashboard page, not just the main overview (icon-only on mobile).
  3. The Launch analysis dialog opens on a single-screen quote.

1. Scope (if you have more than one topic)

A dropdown offers All topics (the default as soon as the brand has several topics) or one specific topic. With a single topic, its name is simply shown instead of the dropdown.

2. Number of prompts

Four chips set the depth of the analysis:

  • Quick — 5 prompts
  • Standard — 10 prompts (the default)
  • Deep — 25 prompts
  • Custom — a field where you type the exact number

Each chip shows its credit cost. A chip your balance cannot pay for is greyed out, and the count is always capped at the prompts actually available in the chosen scope. The prompts that run are drawn from the scope, unless you pick them yourself (see Advanced options).

3. Assistants and cost

Below the depth, a box recalls which assistants will be queried ("Analysis with N models") with a Change link to the Models page. Under it, the cost line shows N credits · M remaining: the cost is prompts × slot cost of your assistant selection (1 slot for most assistants, more for a few — see Plans & credits). If the quote exceeds your balance, the line turns orange, a link offers to buy credits and the confirm button stays disabled.

4. Advanced options

Collapsed by default:

  • Avatar — an audience persona injected into prompts. Five preset personas (Alex 20, Marie 40, Jean 60, Sophie 35, Thomas 28 — see Avatars) are available on STARTER and above; custom avatars (PRO and above) are created in Settings → Avatars, and the one marked as default is preselected.
  • Choose exact prompts — tick the prompts to analyze (a search box filters the list; pinned prompts come pre-ticked). This selection replaces the depth and works in every scope, "All topics" included: only the ticked prompts are analyzed.

5. Confirm

The Launch (N credits) button sends the analysis. The dialog closes and a notification sums up what was sent ("Analysis launched: N questions × M assistants").

If the button opens a message instead

When an analysis cannot start, the dialog shows a screen naming the reason instead of the quote, with the action that clears it:

  • Trial ended — "Choose a plan to run a new analysis": your results stay readable, a button leads to choosing a plan. If you are a team member rather than the account owner, the screen asks you to have the owner choose one.
  • Suspended brand — "This brand is no longer measured": it exceeds the number of paid brands. The owner can reactivate it in place of another brand or increase the brand count; a member is asked to request it from the owner.
  • Viewer role — "Launching is reserved for editors": your role on this brand lets you read, not launch. Ask an editor or an admin of the brand.
  • No questions — "No questions to analyze": a button leads to the Prompts page to add some.
  • No assistant — "No assistant selected": a button leads to the Models page.

Two refusals appear inside the quote itself, as a message, after you click Launch:

  • Insufficient credits — the message states the credits remaining and the credits needed.
  • Too many attempts — several launches in a short time: wait a minute, then try again.

If an analysis is already running, the button shows its progress and the dialog shows the progress instead of a second quote.

After launching

The Launch analysis button turns into a progress pill (e.g. "Analysis 12/30") — visible from any dashboard page, not just the one you launched from. Click it to open a small progress summary. Results update progressively as each model responds - you do not need to stay on the page.

Credits are reserved at launch. If a model returns no answer for a prompt, that slot is refunded.