> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tryrekkal.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Set up your brand
> One domain in, and rekkal derives your brand profile, topics and prompts. You review, approve, and measurement starts immediately.
Setup asks you for **one thing: your domain**. Everything else — your brand name, what you
sell, your market, your language, the topics you compete on and the questions rekkal asks —
is derived and put in front of you to correct.
That is deliberate. A setup form with fifteen fields is a setup form people come back to
later and never finish.
Create an account with an email address, or with **Continue with Google**. Until a brand
is set up, the dashboard shows a **Set up your brand** invitation instead of an empty
analytics page — that link is where setup starts. Nothing can be measured until there are
prompts to run.
Just the domain — `example.com`. rekkal reads your homepage once and asks a model to
summarise what it found.
A country-code domain settles your market on its own: a `.fr` domain is registered in
France, and that is a fact rather than an inference, so it is not overridden by anything
the model thinks it read.
This step takes a few seconds. If your site cannot be reached over HTTPS, rekkal says so
rather than guessing at a profile.
Brand name, description, industry, how you position yourself, your products and
services, market, language and timezone all arrive **pre-filled**. Positioning
attributes and products are individual chips you can remove with one click, not a
paragraph you have to rewrite.
Correct anything wrong here. Everything downstream is built from it.
rekkal groups the questions into five topics — the themes buyers ask about in your
category. Deselect a topic you do not compete on and every prompt under it goes with it.
Reading five topics is the point of this step. Approving forty prompts one at a time is
not approval, it is authoring.
Under those topics sit 40 generated prompts, pre-selected and spread across the topics
rather than spent on the first one, so you start with coverage rather than depth in one
corner. Your [plan's](/plans) allowance — 50 on Starter, 100 on Pro — is the ceiling on
what can be saved, and setup writes 40, so on both plans you finish with room to add
more of your own.
Swap any prompt for another; the count against your allowance is on screen. You cannot
save more prompts than your plan allows.
Saving writes your workspace, your brand and your prompts, starts measuring straight
away, and lands you on the dashboard.
## What happens on the first run
rekkal measures on a **daily slot**: every day has one slot, dated to that day in your
workspace's timezone.
**Saving starts a measurement immediately.** Your prompts are dispatched the moment setup
completes, not at the next scheduled pass, so answers begin landing while you are still
looking at the dashboard. They are recorded against today's slot — finish setup at three in
the afternoon and the results land under today's date.
That first pass is a real measurement, not a preview: the same questions, the same engines
and the same arithmetic every daily refresh uses. Nothing about it is a sample or a demo,
and you can open any figure on it and read the individual answers behind it.
The runner also wakes up every hour and picks up any prompt that still has no answer for
today's slot, so a large prompt set that does not finish in one go keeps filling in over
the following hours rather than waiting for tomorrow. A retry later the same day reuses
that slot rather than spending a second answer.
While that is happening the dashboard says so. It shows how many answers have arrived, how
many are still in flight, and — if the pass could not collect anything at all — what went
wrong and whether it cost you any of your allowance. It never shows a zero in place of a
measurement that has not happened.
Each prompt is sent to each engine your plan covers. For every answer, rekkal stores:
* the answer text;
* which brands were named in it, and where in the text;
* which sources the engine cited.
**Your first day is one data point.** One daily slot of a handful of prompts is a small
sample, and rekkal will tell you so: figures below the reporting threshold are labelled
**Below this measure's reporting threshold**, and there is no previous period to compare
against yet, so the change on each card reads **No comparison**.
This is the product working, not the product missing. See
[Reading a change](/metrics/reading-a-change).
## What you see on the dashboard
Screenshot placeholder: `/dashboard`
* **`/dashboard`** — visibility, share of voice and average position over a window of 1, 4
or 12 weeks of daily slots, with a daily chart and a per-engine breakdown for each, plus
the competitor ranking.
* **`/dashboard/sources`** — which domains the engines actually cited, how often, and which
of them cover a tracked competitor but not you.
* **`/dashboard/settings`** — your brand, your account and your billing.
## Changing things later
Correcting a wrong brand profile is a one-click edit during setup. Editing your brand,
prompts or tracked competitors **after** setup is not in the product yet — today, setup runs
once per account, and there is no way to change any of it afterwards. Review carefully while
you are in setup; that is the one place corrections are free.