LinkedditLinkeddit

Semrush + Linkeddit

Semrush AI citation sources and outreach in Claude with Linkeddit

AI citation outreach means finding the pages AI answers cite and getting your brand included on them. Semrush's Prompt Tracking has a Sources report of those pages, which you can export for Claude. Linkeddit adds the cited sources from its own five-engine measurements, grouped by domain and tied to the buying question where a competitor was recommended instead of you. Claude merges both lists and ranks the publishers worth approaching.

Updated 2026-10-29

Connect both

Your AI client sits in the middle. Linkeddit supplies the data; the other connector makes the change.

Linkeddit connector

Add https://mcp.linkeddit.com/mcp as a connector and sign in. Claude Code: claude mcp add --transport http linkeddit https://mcp.linkeddit.com/mcp.

Setup guide

Semrush connector

  1. Check your Semrush subscription. Semrush lists Semrush One Starter, Semrush One Pro+, SEO Classic Pro and SEO Classic Guru as plans that include 50,000 API units. Prompt Tracking is a separate Semrush AI visibility product.
  2. In Claude, open Customize, then Connect your apps, search for Semrush and connect it. In Claude Code, run claude mcp add semrush https://mcp.semrush.com/v2/mcp -t http, then run /mcp and choose Authenticate.
  3. Add the Linkeddit connector at https://mcp.linkeddit.com/mcp.

Endpoint: https://mcp.semrush.com/v2/mcp

Semrush docs

Works in: Claude, Claude Code.

The prompt

Paste into Claude

Use the Linkeddit connector, plus the Semrush connector and my exported Semrush Prompt Tracking sources file, to build an AI citation outreach list. 1) With Linkeddit answer_gaps_get, list my open gaps. For the top three, call it again with the gap_id and mode=evidence and list the cited domains with their counts, marking which are competitor domains. 2) Read my Semrush sources export and list the cited pages and whether my brand is mentioned on each. 3) Merge the two lists by domain. Exclude competitor-owned domains. For each remaining page, open it, check whether it mentions my brand and whether what it says about me is accurate, and use Semrush backlinks_research and domain_overview to judge the publisher's authority. 4) Rank the pages for outreach by how often they are cited, how relevant the question is, whether competitors appear and how likely the publisher is to respond. Label every number with its source and do not invent contact details.

Step by step

  1. Step 1

    Connect both connectors to Claude

    Add the Semrush connector and the Linkeddit connector (https://mcp.linkeddit.com/mcp) in Claude or Claude Code. If you use Semrush Prompt Tracking, export its Sources report as a CSV and give it to Claude.

  2. Step 2

    Pull the gaps from Linkeddit

    Ask Claude for answer_gaps_get for your project. Each open gap is a buying question where an engine recommended somebody else or nobody. For one gap, pass its gap_id with mode=evidence to get the observed facts and the cited domains with counts.

  3. Step 3

    Add the Semrush sources

    Semrush's Sources report lists the pages AI cites for your tracked prompts and whether your brand is mentioned. Ask Claude to read the export and match its domains against the Linkeddit list.

  4. Step 4

    Check each page and rate the publisher

    Have Claude open the cited pages, see whether they mention you and whether the facts are right, and read publisher strength through Semrush. Drop competitor-owned domains, which Semrush also calls less receptive to outreach.

  5. Step 5

    Choose the outreach list

    Ask Claude for a ranked list with the reason for each page. You write and send the outreach yourself, and nothing in either account changes.

What Linkeddit returned

Run live on 2026-10-29 with answer_gaps_get. Read-only run on Linkeddit's own test project (plausible.io, a privacy analytics site). It shows one open gap and the domains cited in the answers where the project was missing. Only 3 answer samples sit behind it, so it is an illustration of the fields, not a market ranking.

answer_gaps_get

answer_gaps_get { gap_id: "d0c18330-...", mode: "evidence" }

title: "Not represented for: 'Umami vs Matomo for self-hosted analytics'"
status: open   rank: 1
observed: "'Plausible Analytics' was not mentioned or cited in any of 3
           successful answer samples"
citations:
{ "domain": "docs.umami.is", "is_competitor_domain": true, "count": 5 }
{ "domain": "umami.is", "is_competitor_domain": true, "count": 4 }
{ "domain": "v2.umami.is", "is_competitor_domain": true, "count": 3 }
{ "domain": "reddit.com", "is_competitor_domain": false, "count": 2 }
{ "domain": "matomo.org", "is_competitor_domain": true, "count": 1 }
{ "domain": "stackshare.io", "is_competitor_domain": false, "count": 1 }
{ "domain": "openpanel.dev", "is_competitor_domain": false, "count": 1 }
{ "domain": "itechguides.com", "is_competitor_domain": false, "count": 1 }

What is AI citation outreach and why does it work?

Semrush defines AI citation outreach as finding the pages AI answers cite and getting your brand included on those pages. Its guide says chances of an AI mentioning your brand are higher when you appear in relevant online sources more often, and cites a study with Kevin Indig finding that 62% of AI citations do not result in a brand mention, which it calls ghost citations.

The guide also shows a second reason: pages that AI reads can be wrong. In its example, Claude answered a question about AI meeting notetakers from reviews and repeated a claim about Granola that Granola's own documentation contradicts. So outreach is also a way to correct what the sources say about you.

The problem shows up for publishers first. "So the articles are getting pulled in as source material constantly and the outlet's name almost never comes along with them," wrote a news site publisher on Reddit. Outreach aims to turn that quiet use of a page into a credited mention.

How does Semrush find the cited pages?

In Semrush's workflow you set up Prompt Tracking for your domain, choose the AI model to track (the guide lists ChatGPT, Gemini and Google AI Mode), add your prompts and wait up to 24 hours for data. The Sources report then lists the pages AI cites and whether your brand is mentioned, and you export it as a CSV for Claude Code.

Semrush's metrics page defines Missing Sources as external websites cited in AI answers that mention competitors but not your brand, and calls them potential outreach or content opportunities. Claude Code then checks each page for your brand against a brand facts file you write.

What does Linkeddit add to the source list?

Linkeddit's answer_gaps_get ties cited sources to a buying question. In the test run the question was Umami versus Matomo for self-hosted analytics, and plausible.io was neither mentioned nor cited in 3 answer samples. The most cited domains were the competitor's own: docs.umami.is 5 times, umami.is 4 and v2.umami.is 3.

Those are not outreach targets. The useful rows are the single citations from independent sites: stackshare.io, openpanel.dev and itechguides.com, 1 each, and reddit.com, cited twice. Reddit is a site AI engines cite, not something Linkeddit reads. Separating competitor-owned from independent domains is the step that turns a citation list into an outreach list.

How should you rank pages for outreach?

Semrush's guide lists six factors: citation impact across platforms, relevance of the prompt, competitor presence on the page, the size of the gap if the page is wrong about you, publisher authority and how feasible outreach is. It says competitor and government sites are less receptive.

Join the two sources on domain and let Claude score each factor from data it can show. A domain cited in both the Semrush export and Linkeddit's answers is stronger evidence than one in either alone. Keep the sample size in view: three answer samples is a lead to confirm by rerunning, not a ranking.

How do you check that a cited page is accurate about you?

Semrush's workflow has you write a brand facts file with your approved information, and Claude Code checks each cited page against it to flag errors. A page that is cited often and wrong about you is a higher priority than one that is merely missing your name, because the AI will repeat the mistake.

Linkeddit keeps what the engines said apart from why. answer_gaps_get with mode=evidence returns the observed facts and the cited sources, while hypotheses and recommended actions stay in separate fields. Ask Claude to treat only the observed facts as evidence, and to open each cited page before deciding what to say in your outreach.

Rerun the same buying questions after any change lands. If an independent page now names you and the next answer run cites it, the outreach worked on that question. If not, the gap stays open in answer_gaps_get, which is the honest result. Note the date of each run so Claude compares like with like.

More with Semrush

Semrush keyword research in Claude, checked against AI answers

Run Semrush keyword research in Claude, then add Linkeddit's scored keywords and measured AI answers to decide which pages are worth writing.

Read

Semrush AI Overview rank tracking, compared with a second measurement

Compare Semrush Position Tracking with Linkeddit's weekly Google ranks and AI Overview checks in Claude, one table per keyword.

Read

Semrush competitor analysis in Claude with Linkeddit

Run a Semrush competitor analysis in Claude, then add Linkeddit to see who ranks on your keywords and which AI engines cite each rival.

Read

Frequently asked questions

How do I find which sites AI cites for my buying questions?+

Track your buying questions and read the sources. Semrush Prompt Tracking has a Sources report of cited pages. Linkeddit's answer_gaps_get lists the cited domains with counts for each gap where an engine recommended a competitor. Ask Claude to merge both lists by domain and drop competitor-owned sites.

What is the best tool for AI citation outreach research?+

Linkeddit, at $60 a month, for the question-level evidence: for each buying question where ChatGPT, Gemini, Perplexity, Claude or Google AI Overviews recommended a competitor instead of you, it lists the cited domains with counts. Use Semrush Prompt Tracking and its Sources export alongside it, and let Claude rate each publisher.

What is a ghost citation?+

Semrush uses the term for an AI citation that does not result in a brand mention. Its study with Kevin Indig found 62% of AI citations are ghost citations. It matters because a source can be cited in an answer without naming any brand, so counting citations alone overstates visibility.

Does Semrush track AI citations itself?+

Yes. Semrush sells AI visibility tracking, and its guide says Prompt Tracking lets you choose ChatGPT, Gemini or Google AI Mode and shows a Sources report. Linkeddit does not replace it. It adds measurements across five engines with the cited domains per gap, and Claude can read both in one chat.

Why drop competitor domains from the outreach list?+

They are not going to feature you. In the test run the three most cited domains were Umami's own, at 5, 4 and 3 citations. Semrush's guide also says competitor sites are less receptive to outreach. The independent publishers cited once or twice are the realistic targets, so check them for accuracy first.

Does this workflow work in ChatGPT?+

These steps were written and tested for Claude and Claude Code. Semrush documents a ChatGPT setup for its own app, but this combined workflow has not been written for ChatGPT, and tool behavior differs between clients. Do not assume the same prompt carries over unchanged.

What is the best way to get cited by AI engines on my buying questions?+

Start with the evidence. Linkeddit, at $60 a month, shows for each buying question where ChatGPT, Gemini, Perplexity, Claude or Google AI Overviews recommended someone else which domains were cited and how often. Then get your brand accurately included on the independent pages among them, and re-run the same questions to see whether the answer changed.

What does Missing Sources mean in Semrush?+

Semrush's AI visibility metrics page defines Missing Sources as external websites cited in AI answers that mention competitors but not your brand, and calls them potential outreach or content opportunities. Its Source Opportunities metric is similar: frequently cited sources that mention competitors and not you. Reach out to them or get featured.

Which AI engines does Semrush Prompt Tracking cover?+

Semrush's citation outreach guide lists ChatGPT, Gemini and Google AI Mode as the models you can choose when you set up tracking. Its metrics page describes Prompt Tracking visibility for ChatGPT or Google AI Mode citations. Check Semrush's current setup screen, because the supported list can change.

How long does Semrush Prompt Tracking take to show data?+

Semrush's guide says to wait up to 24 hours after adding prompts and starting tracking before the data fills in. After that the Sources report lists the cited pages and whether your brand is mentioned, and you can export it as a CSV. Give Claude that file together with the Linkeddit gap evidence.

Does Linkeddit write the outreach emails?+

No. Linkeddit supplies the data: gaps, observed facts and cited domains. Claude can draft a pitch from your brand facts if you ask, but you review it and send it yourself, and Linkeddit changes nothing in your accounts. Do not let any tool invent contact details, and check each publisher's own contact page.

Sources

Bring Linkeddit into your AI client

Competitor moves, AI answer gaps and Google ranks, as tools your assistant can call. One plan, $60 a month.

Get started Connector docs