How to Use the Ranqo Docs: Find the Answer, Check the Math, Pull the Data

Start From the Page You Are On

From the dashboard

Every dashboard page has Help in its header, apart from the full-screen setup for adding a brand. Read the guide opens that page's guide in a new tab, so Help on the Competitors page lands on the Competitors guide and Help in the Action Center lands on the Action Center guide. The same menu has Take the page tour on pages that have one, Watch the tutorial when the page has a published video, and Documentation for the docs' front page.

From search

Search reads the text of every page, not only the titles. The label you saw on a card, such as Share of Voice or No detectable change, is usually enough to reach the page that explains it.

From the front page

The docs open on an Introduction that groups the main entry points into four sets of cards: Get started, Work the loop, Understand the numbers and Connect your stack. The Guides sidebar is grouped by the dashboard's own sections, Monitor, Analyze, Act and Create, so a page's guide sits under the name it has in the app.

Plenty of questions have a page of their own. These are the common ones, and where each is answered.

Check How a Number Is Calculated

The four numbers on the Home scoreboard each have a methodology page with the formula behind them. The definitions below are the ones the dashboard shows when you hover each metric, so the docs and the product describe a metric in the same words.

Two numbers that are easy to misread

Sentiment and Share of Voice are the two most likely to look wrong if you work them out by hand from the counts on a page.

Sentiment is not the share of positive mentions. Neutral mentions count half, and a negative mention costs twice what a neutral one does. A brand AI only ever describes neutrally scores 50, the middle of the scale, rather than 0. Say AI names you in 10 answers on one engine: 6 positive, 3 neutral and 1 negative. Your sentiment is (6 + 1.5) / 10 × 100 = 75. The positive share would have said 60%. The Sentiment page works a larger example.

Share of Voice does not count every brand AI names. It divides your mentions by your mentions plus those of qualified competitors. A competitor qualifies if you track it, if AI named it more than once in the window, or if AI gave a website for it that exists and matches its name. A brand named once with no website stays out of the total, so a long tail of one-off names cannot shrink your share. The Share of Voice page has the full rule and a worked example.

What counts as a mention

An answer names you when it uses your brand's name or one of your aliases to refer to you, and an answer that names you five times still counts once. Ranqo reformats each answer before reading it. If your name appears only in the reformatted text and not in what the engine wrote, the mention is withdrawn.

The prompts Ranqo writes for a new brand never name it, because a question that names you almost always gets an answer that does. That would tell you nothing about whether AI recommends you to someone who has not heard of you. How Ranqo measures walks through every step an answer goes through.

Citations and source types

Two pages explain the Sources page. Citations covers what one cited page is, which sites count as your own, and the difference between the % Answers and % Citations columns. Source and article types explains how every cited site is sorted by its relationship to you, from your own site to competitors, editorial sites and forums, and how every cited page is sorted by format.

Find Out Why a Number Moved

Noise between runs

AI answers change from run to run even when nothing about your brand has. A mention seen for the first time is still there in the next run only about half the time. That churn is a large part of why the overall score can swing 14.5 points with nothing changed.

The Action Center allows for it. In a recommendation's drawer, a measured lift shows as a number only when it is bigger than the noise band for its metric. A smaller reading shows as No detectable change. Noise bands has the current bands and how they were measured.

One point per run, and the As of date

Runs are weekly, and the trend charts plot one point for each run, joined by a line. They never invent a value for a day nobody measured, and they show a change only once there are two runs to compare. When the date range you pick holds no run at all, most pages show the last completed run with an As of date and no change, instead of an empty page. One point per run and Carried values explain both.

Why a past week can change

A figure you read last week can read differently today. Usually a new run has entered your date range and the oldest has left it. A week that is over can move too: filters read each prompt as it is today, deleting a prompt removes its past answers, a competitor's spellings can be grouped under one name, and your time zone decides which day a run lands on. Why numbers change goes through every case, including why a run with failed queries reads differently from one page to another.

Comparing Ranqo with another tool brings in more again: engines disagree with each other, answers vary from run to run, and each tool defines visibility its own way. Our post on why AI visibility tools disagree covers that.

Start from the symptom

The Troubleshooting pages start from what you see rather than from a feature: Visibility is zero, A score dropped, No data for an engine, Site Tracking shows no events, Google Analytics is not syncing and WordPress connection fails. Each lists the likely causes and what to change.

Set Up and Connect Your Stack

Your first brand and first week

Quickstart goes from adding a brand to reading its first visibility score. Your first week covers what runs by itself after that, and what to do first. Plans and limits shows what each plan includes. Its tables are rendered from the same billing configuration the product enforces, so they change when a limit does.

Traffic: Google Analytics and Site Tracking

Google Analytics 4 connects with read-only access and no code changes, and shows the sessions it recorded from AI assistants and search engines. Site Tracking is a small piece of server code, with setup guides for Next.js, Express, a Cloudflare Worker and any other backend. It reports every page request, including those from AI crawlers that never run JavaScript and so never show up in GA4.

The two see different things, which is why the Traffic guide puts them side by side.

Publishing to WordPress

Generate Blog can send a post to your WordPress site as a draft or a live post, and updates it in place when you publish again. The connection signs in with an application password, never your main password. WordPress covers connecting, and Publishing to WordPress covers sending a post.

Pull Your Data With the API

The Ranqo API gives your own tools read access to what the dashboard shows: your brands, their visibility and competitors, the prompts you track and the answers AI gave them, tracking runs, Site Access, page audits, Action Center recommendations and Outreach targets.

Who can use it

It is read-only JSON over HTTPS at api.ranqo.ai/v1, and every request carries an API key in its Authorization header. The account owner creates keys in Settings, under API keys. Access is included on Pro and above, and a plan in its trial has the same access.

The same numbers as the dashboard

The reports run on the same code as the Visibility page, over the same window and filters, and are rounded the way the page prints them. Send your time zone in tz: the dashboard counts days in your browser's zone, and the API counts them in UTC when you leave it out. Dates and time zones has the detail, and the API introduction says where else a figure can differ from a page.

Try it before you write code

Every endpoint's reference page, such as Get visibility, has a Try it button that sends a real request with your own key. Paste the key and fill in the parameters. Choose lists the ids in your own account, such as your brands and prompts, so you never hunt for one, and the time zone starts as your browser's. Then press Send.

The live answer comes back with its status, how long it took, how many requests your key has left this minute and a request id to quote to support. Beside it, the code that makes the same request updates as you type, in cURL, JavaScript and Python. Your key stays in that browser tab only, and never goes into the code or a URL.

Your first requests

When you move to your own code, the API quickstart takes about five minutes: create a key, check it, find your brand's id and read its visibility. These are its requests.

export RANQO_API_KEY="paste-your-key-here"

# The account the key belongs to: your plan, its rate limit and the key's details
curl https://api.ranqo.ai/v1/account \
  -H "Authorization: Bearer $RANQO_API_KEY"

# The brands the key can read, each with the id the brand endpoints take
curl https://api.ranqo.ai/v1/brands \
  -H "Authorization: Bearer $RANQO_API_KEY"

# One brand's visibility over the last 30 days, counted in your time zone
curl "https://api.ranqo.ai/v1/brands/BRAND_ID/visibility?period=30d&tz=Europe/Berlin" \
  -H "Authorization: Bearer $RANQO_API_KEY"

Use it to feed a BI dashboard that puts AI visibility beside your other channels, a client report built from live numbers (our report template lists what belongs in one), or an AI agent that reads your visibility data. The whole API is also described in an OpenAPI document, ready to load into an API client or a code generator.

Read the Docs With an AI Assistant

Every page in the docs is also plain markdown, so you can hand it to ChatGPT, Claude or your own agent as it is. There are three ways to get it.

• 01Add .md to the address. Any page's markdown lives at its own address plus .md. Tables the page renders from live product settings, such as plan limits, come through as plain markdown tables.
• 02Take everything at once. llms.txt at the root of the docs lists every page with a one-line description and a link to its markdown. llms-full.txt holds every page in one file.
• 03Use the button on the page. The Copy page button beside each page's title opens a menu: Copy page puts the markdown on your clipboard, View as Markdown opens it, and Open in ChatGPT or Open in Claude starts the assistant with a prompt pointing it at the page's markdown, so you can ask about the page straight away.

Markdown also makes it easy to ask about several pages at once. For example:

Read these two pages from the Ranqo docs:
https://ranqo.ai/docs/methodology/share-of-voice.md
https://ranqo.ai/docs/guides/troubleshooting/score-dropped.md

My Share of Voice fell this week while my Visibility held steady. Using only those two pages, list the likely reasons in the order I should check them, and tell me where in the dashboard to look for each one.

We wrote the docs for two readers: the person on the dashboard, and the assistant that person asks. If you want the same for your own site, our llms.txt guide covers the format and what it does and does not do.

How We Keep the Docs Accurate

Documentation that drifts from the product is worse than none, because people act on it. So the numbers that live in product configuration are not typed into the pages by hand. Plan limits, the engines each plan can track, the Action Center's thresholds and the crawler lists are rendered from the same configuration the product enforces, and they change when it does.

The pages also describe the product as it works today, uneven parts included. Where two pages count the same thing differently, the docs say so instead of picking one: the methodology overview sets out which pages count an answer reused after a failed query and which leave it out. If you find a page that does not match what you see, write to support@ranqo.ai and we will correct it.