Quick Start

This is the path almost everyone takes on day one. Five steps, about fifteen minutes, most of it waiting on the first scan. For why the numbers look the way they do, read Core Concepts next.

Before you start

Have these to hand:

• Your primary domain, such as nike.com. It is the only thing the first screen asks for.
• Your brand name the way customers say it (Nike, not Nike, Inc.), to confirm on the next screen.
• Two or three competitors you want to be compared against.

Adding a brand needs an active trial or subscription, because the first scan spends real model calls. If you do not have an account yet, sign up first. The rest of this guide assumes you are logged in.

Step 1: Add your brand · about 3 minutes

Go to Settings → Brands and click Brand in the top right. On a brand new account the same flow sits behind Add your first brand. Either way you land on the setup flow, which has six screens.

You are not asked to type a brand name, a description, or a competitor list. Trakkr derives all of it from the domain and asks you to correct it. The brand icon is the favicon of the domain you entered.

Step 2: Check your prompts · about 2 minutes

Setup writes a starting prompt set for you. Open Prompts from the sidebar and read it before you trust any number, because these are the questions Trakkr will ask every model every day.

Good prompts read like something a person would type into a chat window. Keyword queries do not work, because a model answers a full question very differently from a fragment.

To change the set, click Add in the top right of the Prompts page. It opens two choices: Add prompts to write, paste, or upload your own, and Suggest prompts to have Trakkr look for gaps. Pressing ⌘K anywhere and typing add prompt opens the same modal.

Aim to cover three angles:

• Discovery, as in "What's the best [category] for [use case]?"
• Comparison, as in "How does [you] compare to [competitor]?"
• Reputation, as in "Is [you] good for [specific need]?"

Growth and Scale both include 50 active prompts per brand, and the Prompts page shows how many you have used. Do not try to be exhaustive on day one. Add more once you have seen the baseline.

Step 3: Watch the first scan · about 3 minutes

You do not start the scan. Trakkr queues the first report the moment the brand is created and sends you to the Dashboard to watch it.

A First scan panel takes over the page. Its headline moves through getting the brand ready, scanning how AI models talk about it, scoring the results, and almost ready, and a chip lights up for each model as it answers. It usually finishes in about three minutes. You can close the tab; the scan keeps running.

If something goes wrong the panel says so rather than spinning. A slow scan says it is taking longer than usual and retries on its own. Two things stop it outright: a brand with no active prompts, and a brand that is paused. Both are named in the panel with the fix.

Step 4: Read your baseline · about 1 minute

Six numbers run across the top of the dashboard. The first four come straight from the model answers. The last two stay empty until you connect something.

Below the hero, the models row shows each model separately. Scoring 80 on Claude and barely registering on Perplexity is normal: different training data, different retrieval, different audiences. That split is where most diagnoses start.

For the formula behind every score, see Metrics. For the rest of the page, see the Dashboard guide.

Step 5: Find a page to fix · about 1 minute

Open Pages. This is where site-wide numbers turn into a specific URL and a clear next step.

Each page moves through six stages: Available, Reached, Understood, Relevant, Selected, and Visited. Start with the leftmost stage marked Degraded or Stalled. That is the first thing holding the page back.

Open a row to see the full page record. It gathers the evidence, the current work, the history, and the measured result for that URL. If Trakkr suggests a fix, you can open the action from the same record.

Not measured is not a failure. It means Trakkr needs a check or a connection before it can judge that stage, and the record says which.

Common questions

Do I have to write prompts myself?

No. Setup generates a topic-grouped prompt set from your domain, and the first scan runs on that set. Editing it is worth doing, but nothing is blocked until you do.

Can I add a brand without a subscription?

No. The first report is only dispatched for a brand with an active trial or paid plan, so a brand added without one will sit waiting rather than scanning. Start the 14-day trial and the scan begins straight away.

How long until the dashboard has a trend line?

The chart needs more than one day of data to draw a line, and the 14-day and 30-day ranges only appear once the brand has that much history. Expect a usable trend after a week.

What if the first scan finds nothing?

That is a real result, not a bug. The dashboard will say there were no tracked mentions in the latest scan. Start with Pages and Citations to see which sources the models used instead, then work on being in them.

Should I run Research on day one?

Only if you already suspect your prompt set is missing something. Research is a separate market scan of about 500 prompts and takes about five minutes to run, so most people do it in week two, once the baseline has shown them where the gaps are.