Signup here ends with a real scan rather than a tour. By the time you reach the dashboard, every engine you selected in step 5 has been asked a set of questions about your brand and the answers have been analysed, so the first numbers you see are measurements rather than placeholders.

This walkthrough covers the six onboarding steps, the first thing to check when the scan lands, and how to tell your own data apart from the sample illustration the dashboard shows before you have any.

Step 1: Create the account and confirm your email

Signup is the first of six onboarding steps (Account, Workspace, Brand, Competitors, Engines, First Scan), with the current step tracked in the URL, so a refresh keeps your place.

Email confirmation is on. That means signup does not log you straight in: it shows a “check your email” state with a resend action, and the confirmation link brings you back to step 2. The link usually opens in a new tab, so expect two tabs and continue in the one that just opened. Your answers are preserved.

Creating the account also provisions a workspace for you automatically. Even as a solo user you are the owner of a one-member workspace, which is what everything else (projects, billing, members) hangs off. There is nothing to set up for it.

One thing the product does not yet do, in case you go looking: two-factor authentication is present in Account → Security as a clearly-labelled coming soon toggle. It is disabled rather than hidden so it is not mistaken for a setting you have already applied.

Step 2: Describe the brand you want measured

Steps 2 to 4 collect what the scan needs to ask a sensible question: your organisation, the brand name as customers say it, your domain, and the competitors you expect to be compared against.

Two details save trouble later:

• The domain is normalised server-side. https://Acme.com/pricing and acme.com become the same value, so you cannot accidentally create two projects for one site.
• Use the trading name, not the legal entity. Engines answer in prose, and the analysis matches against that prose. “Acme” finds mentions that “Acme Holdings Ltd.” never will.

Competitors here are names, not domains, for exactly the same reason.

Step 3: Choose the engines to track

Step 5 shows the four free-plan engines as selectable cards (ChatGPT, Perplexity, Gemini and Claude), with a note that paid plans can track any four of the seven models (adding Grok, Meta AI and DeepSeek) and that Google AI Overviews, Google AI Mode and Microsoft Copilot run on the weekly SEO-data pass on paid plans rather than as engines you add. There is no engine picker beyond those four cards; wider engine choice is a dashboard setting, not an onboarding one.

Selecting more engines does not cost more. A credit buys one prompt, checked once, across every model you run, so the same prompt set costs the same on one engine or all the models you’re entitled to. The only thing an engine you will never act on costs you is an extra row to scroll past. You can change the selection at any time in Settings → General under AI Engines, and the Engines Tracked card on the Overview page reports what your scans are actually covering.

Step 4: Run the first scan and watch it finish

Step 6 creates the project and runs a genuine scan. The page polls the job until it is done, which typically takes under a minute.

If the scan fails (a provider outage, or a deployment where the AI gateway is not configured), you are told plainly and signup still completes. The project is already saved; you can run the scan again from the dashboard. The step is also reload-safe: refreshing it adopts the project you already created instead of colliding with it.

Free accounts refresh weekly on their own, so the automated cadence covers you there too. Every paid plan re-scans daily: the cadence is uniform per population, and you can still run a scan by hand whenever you want a fresher reading.

Step 5: Read the Overview page

The dashboard opens on Overview in the Monitor group. Five stat cards run across the top:

• AI Visibility Score: the 0–100 composite for your latest completed scan.
• Engines Tracked: how many surfaces this project is scanned across.
• AI Shelf Share: how often an answer that features you actually cites your domain, as opposed to merely naming you. A mention without a citation means the engine knows you and sent the reader elsewhere.
• Domains Cited: unique domains recorded as citation sources in the latest scan.
• Open Alerts: findings from the same scan that need a decision.

Below them sit the visibility trend, a per-engine breakdown, recommendations and recent alerts.

Read the score next to the mention rate, never alone. The composite folds together how often you are mentioned, how prominently, and how you are described, so two very different situations can produce the same number. Understanding Your Visibility Score works through exactly how, and it is the next thing worth reading.

Step 6: Tell your data from the sample illustration

Before a project has a completed scan, several pages render illustrative charts and tables so the layout is not empty. Those panels show plausible-looking numbers rather than blanks, which is exactly why the rule to learn is look for the label, not for an empty value. Each stat card shows — with a “No scans yet” note beneath it, and a sample-data notice sits above the row. They are display-only: they never reach the database, never count toward usage, and never affect billing.

The moment a real scan completes, those panels switch to your own results and the labelling disappears. A number on its own tells you nothing; a number on a “No scans yet” panel is not about your brand.

The same distinction applies to a scan that failed: a failed job produces no data, and the page keeps showing the last completed scan rather than inventing a gap. The Activity page in the Workspace group lists every job and its outcome, and is where a failed or stuck scan can be restarted.

Step 7: Set up the things a first scan cannot infer

The first scan uses auto-generated discovery prompts, because nothing yet knows what your buyers ask. That is a fine baseline and a poor programme. Three follow-ups, in order of payoff:

1. Replace the prompts with real buyer questions. Auto-generated questions are inferred from your domain and industry; the ones your customers actually type are not. See Setting Up Tracked Prompts.
2. Decide where findings should reach you. Alerts are derived from every scan automatically; what you configure is delivery by email or webhook: Setting Up Smart Alerts.
3. Connect Search Console. It is available on every plan including free, and it is the only source that backfills real history the moment you connect it, so it is the fastest way to have a trend at all: Tracking Google Search Performance.

To monitor a second brand or site, use the + button beside the project selector in the top bar and add a domain and brand name. Each project is scanned, scored and reported independently, subject to your plan’s domain allowance.

Worth knowing before you widen coverage: these are conversational surfaces rather than ranked result lists. Google’s own guidance on AI features makes that explicit, which is why the question you ask is as much a part of the measurement as the brand you ask about.

Step 8: Where to go next

• Visibility score: what the headline number is made of.
• What is AEO?: the discipline this product measures.
• How AI search works: the mechanics underneath every number on the dashboard.
• Verifying Your Domain: optional, and worth knowing what it does and does not unlock.