Help Center
Get the most out of Linkeddit
Quick guides for Compete, Search, Answer Radar and the MCP connector.
Compete
Tracking competitors, reading the weekly brief, and the customer reviews collected from review sites.
Getting Started With Compete
The Compete plan tracks up to 12 competitors, with up to 6 sources and 3 monitored URLs each.
Reading Your Compete Weekly Brief
One brief per week, Monday to Sunday, for all your competitors together.
Competitor Customer Reviews in Compete
Reviews keep their text, title, date, link and rating. Five star reviews are kept too.
Search
Where your site ranks on Google, AI Overviews, keyword research and watched pages.
Getting Started With Search
Search is part of the Compete plan and shares its sites with Answer Radar.
Reading Search Rankings and AI Overviews
Ranks are checked weekly and read from the first ten results.
Search Keyword Research: Topics, Map, Competitors
Counts are checks from your monthly allowance, which defaults to 500 per account.
Watching a Page Change in Search
It is free. No checks or credits are used, only Search Console data already synced.
Keyword Research
Finding the Google searches your buyers make, and how each keyword is scored.
Answer Radar
How AI visibility is measured, from creating a project to verifying a fix. Every article is checked against the shipped feature, and the definitions behind the numbers are on the methodology page.
Getting Started With Answer Radar
Projects live at /answer-radar. Each has Overview, Actions, Questions, Fixes and History.
Creating an Answer Radar Project
Creation writes the project, its primary brand and its competitors in sequence, with an idempotency key so a retry finishes rather than duplicates.
Adding Competitors to Answer Radar
Share of voice counts a vendor once per answer, however many of its products the answer lists.
Choosing and Approving Buying Questions
Add your own buying question puts a question you wrote into the set.
How Buying Questions Are Chosen and Saved
Linkeddit does not generate questions. The writing is yours or your assistant's.
The Audit Run and What It Costs in Credits
1 credit = 1 measured sample. Customers see credits and engine names, never model ids or currency.
Reading the Visibility Metrics
Metrics are per segment: engine, model, tool version, geography, language. Never averaged across engines.
Reading the Mean Position Metric
Per vendor, per segment, over the samples that name the vendor.
Reading the Appearance Rate Metric
Only Google AI Overviews produces absent samples today. The assistants answer every query, so their appearance rate is 1.
Track and Win ChatGPT With Answer Radar
In 47 measured ChatGPT answers to 4 September 2026, 74.5% searched the web and 72.3% cited at least one source, 3.72 sources on average.
Track and Win Perplexity With Answer Radar
In 44 measured Perplexity answers to 4 September 2026, 100% searched and 100% cited, 9.8 sources per answer. One in five answers cited a reddit.com thread.
Track and Win Gemini With Answer Radar
In 62 measured Gemini answers to 4 September 2026, 93.5% searched, 56.5% cited at least one source, 5.27 sources on average.
Track and Win Claude With Answer Radar
In 67 measured Claude answers to 4 September 2026, 85.1% searched and 85.1% cited, 3.93 sources per answer. None cited a reddit.com thread.
Track and Win Google AI Overviews With Answer Radar
Appearance rate is shown divided by shown plus absent. Provider errors sit on neither side.
Track and Win Google AI Mode With Answer Radar
Every AI Mode row carries the Searched pill, because the answer is built over a live Google search.
Segments and Topics: How Results Are Grouped
One set of numbers per segment. A project on six engines has at least six segments.
The Query Fan-Out View
Grouped by prompt, then by segment: engine, instrument, locale.
The Content Gap View
Agreement is counted in engines, not samples, so one engine repeating itself does not create a requirement.
Drafting a Fix From the Evidence
Fix types: brief, article, response.
Verifying a Fix: Re-Measurement After a Change
A null report means not measured yet, never no change.
The Action Queue
Impact keys: answers without you, intent, evidence, competitor citations, competitor mentions.
Share Links: Showing a Result Without a Seat
Shareable: one gap, one run, one fix, or a project's metrics.
The Email Brief and Its Preferences
Off by default for every account. Only your own toggle enables it.
Scheduling Recurring Audits
Daily exists but is restricted to internal accounts; the response lists what your account may pick.
Usage, Credits and Limits
Configured per-run limits on the day: 50 questions, 5 samples per question, 200 samples.
Connecting Google Search Console
Linkeddit never holds a customer credential; the grant lives in your Google account.
Answer Radar From Claude, Cursor and Other MCP Clients
Claude.ai connectors use OAuth 2.1 with PKCE; the docs cover Claude and Cursor step by step.
What Answer Radar Does Not Do
Measured through official provider APIs with grounded web search: a controlled proxy, disclosed with every number.
Frequently asked questions
How do I get started with Answer Radar?+
Create a project from your domain, approve at least one buying question, and run an audit. The Overview then fills with gaps, cited sources and an action queue. Nothing is measured until a question is approved.
What is an Answer Radar credit?+
One measured sample: one question asked of one engine, once.
How do I use Linkeddit from Claude or Cursor?+
Add mcp.linkeddit.com/mcp as a connector and sign in once with OAuth. There is no API key. The MCP documentation has step-by-step guides for Claude, Cursor, VS Code and Windsurf.
Which plan do I need?+
Compete at $60/month, the only plan sold. It includes Compete, Answer Radar and the MCP connector.
How are the visibility numbers defined?+
The methodology page defines every metric, what counts as a valid measurement, and how fixes are re-measured. Failed measurements are reported, never counted as zero.