TL;DR Summary:
Ghost Citation Problem: Semrush says 62% of AI citations never mention the brand behind the cited page, which creates a visibility gap where your content gets used but your name stays hidden.Workflow Breakdown: The new Semrush and Claude Code process starts with Prompt Tracking, exports the Sources report, and then uses three local files to sort each citation into categories like not mentioned, outdated, or underrepresented.Outreach That Works: The strongest pitch is a small, polite correction request sent to publishers, and Semrush recommends reviewing every AI-generated classification by hand because the model can change its answer from one run to the next.Measure Monthly: Re-export the Sources report each month to see whether your brand is being named more often, which pages are still cited, and how your overall AI visibility changes over time.How do you get AI tools like ChatGPT or Claude to mention your brand instead of just citing the page it appears on? Semrush published a new workflow on September 21, 2026, that pairs its Prompt Tracking tool with Claude Code to answer that question. The workflow matters right now because AI citation outreach fixes a gap most brands don’t know they have: getting cited without getting named.
Why AI Citation Outreach Solves a Real Visibility Problem
Semrush’s research with Kevin Indig found that 62% of AI citations never mention the brand behind the cited page. These are called ghost citations. An AI model reads a page, uses it to answer a question, and skips your brand name entirely. This shows up often in evaluation-stage prompts, the questions people ask when comparing products before buying. Semrush tested this with the prompt “What are the best AI meeting notetakers?” in Claude. The response pulled from Medium reviews and G2 ratings, then named four brands as the top picks: Otter, Fathom, Granola, and Fireflies. One review Claude used was outdated. It claimed Granola doesn’t separate speakers, but Granola’s own documentation says it does, through speaker tags on Zoom and Google Meet. AI citation outreach exists to catch and fix these gaps. Since ghost citations are so widespread, knowing exactly which sources AI models pull from for a given brand or competitor is the first step toward fixing them, which is the specific problem AI Mentions: See Which Sources AI Cites Over You is built to surface, giving teams a starting point for outreach before running deeper analysis.
How the Semrush and Claude Code Workflow Runs
The process starts in Semrush’s Prompt Tracking tool. You add your domain, pick a model such as ChatGPT, Gemini, or Google AI Mode, and upload your prompt list. After up to 24 hours, you export the “Sources” report, which shows every page AI cited and whether your brand got mentioned. Next, you load that file into Claude Code, a version of Claude that reads files on your computer and follows plain English instructions. You need a paid Claude plan, since Claude Code isn’t available on the free tier. The setup requires three files: CLAUDE.md, which explains your project, brand-facts.md, which lists your pricing and product details, and analyze-sources.md, which tells Claude how to sort each page. Semrush recommends running the analysis with Claude’s Opus model.
What AI Citation Outreach Categories Reveal About Your Brand
Claude sorts each cited page into one of six categories: not mentioned, underrepresented, outdated or inaccurate, negative, no clear opportunity, or manual review. In a test using H&M, Claude flagged a roundup of affordable office clothing as “underrepresented,” since H&M appeared once among roughly 50 brands. A human check found H&M actually appeared twice, in both the blazer and tops sections. This is why Semrush insists on human review of every output. Claude Code can also produce different results if you run the same file twice, since the tool is non-deterministic, meaning it doesn’t give the exact same answer every time.
What Makes AI Citation Outreach Pitches Work
Sean Markey, who runs outreach for Semrush, tested a pitch style built around one small ask instead of a full list of changes. Out of his last 14 responses, 8 publishers agreed to receive updates without asking for anything in return. His approach opens by thanking the publisher for the mention, then asks permission to send a short correction. Semrush recommends applying the same structure across every AI citation outreach pitch, regardless of brand size.
Tracking Results Over Time
Semrush recommends re-exporting the Sources report every month and comparing it against the last one. You check whether a page is still cited, whether your brand now appears, and whether new citing pages have shown up. The main metric to watch is your overall AI visibility and sentiment over a quarter.
Start by exporting your Sources report from Prompt Tracking and checking how many citations exclude your brand name. That single number tells you whether AI citation outreach deserves a place in your monthly routine. If you’d rather skip the manual Claude Code setup, tools like AI Mentions automatically surface the exact sources AI models cite for your competitors, so you know precisely where to focus outreach before ghost citations quietly steer buyers elsewhere.
