Tracking AI Citation Gaps with Semrush Prompt Tracking

Semrush’s “Sources” report within its Prompt Tracking tool identifies the specific third-party domains and pages that AI platforms retrieve to generate answers. By comparing these citations against actual brand mentions, practitioners can identify “citation gaps”—instances where an AI platform relies on a specific source but does not include the user’s brand in the resulting output. This functionality allows brands to move beyond general visibility and target the specific third-party pages that influence AI responses.

If you already use a Semrush plan to track AI prompts, you can immediately begin identifying which specific websites are shaping the AI’s perception of your brand.

Distinguishing Citations from Brand Mentions

Citations and brand mentions measure fundamentally different aspects of AI visibility. A citation refers to a specific URL or domain used as a source during the retrieval process, while a brand mention indicates the inclusion of a brand’s name in the final AI-generated response. A brand may appear in an answer without its own website being cited, or a website may be cited as a source without the brand being recommended in the text.

Tracking both metrics simultaneously allows users to identify a Citation Gap—the disparity between the sources an AI platform relies on and the presence of a specific brand in the answer. Identifying these gaps is the first step toward understanding why an AI platform is favoring a competitor or a third-party site over the user’s own content.

Building Prompt Lists for Semrush Tracking

Building a target prompt list defines the scope of the AI answers a brand intends to monitor. Users should start by identifying questions that reflect actual customer problems, product evaluations, and specific requirements. For example, a mattress brand might track queries regarding “shoulder pain” recommendations.

Semrush's Prompt Research tool

Identifying these questions requires analyzing real customer conversations found in support transcripts, surveys, and online communities like Reddit. Semrush’s Prompt Research tool can also be used to discover the specific questions people are asking AI platforms to define the monitoring scope. (Prompt Research: a tool used to discover specific questions people are asking AI platforms to define the monitoring scope.) Once these prompts are identified, they are added to a campaign in the Position Tracking tool to begin collecting daily data points.

Analyzing Sources and Citation Gaps

The Sources report provides three primary KPIs to measure AI visibility: Tracked prompts (the total count of monitored questions), Source pages (the number of unique URLs cited across those answers), and Brand mentions (how often the specific brand appears). These metrics provide a baseline for how large the source pool is and how often the brand is successfully appearing in the results.

Users can switch between platforms

Users can switch between platforms, such as ChatGPT and Google’s AI Mode, to see how the source pool changes based on the model’s underlying architecture. (AI Mode: a specific AI platform functionality that retrieves and processes information from the web to provide answers.) For instance, in a campaign tracking 20 prompts, the same set of questions surfaced 49 source pages in AI Mode but only 40 in ChatGPT. Because AI platforms update their retrieval daily, users must check these reports regularly to see if citations or mention rates shift over time.

76% Retail vs 33% Knowledge Base: Platform Differences

Different AI platforms exhibit distinct preferences for the types of content they retrieve during the Retrieval Augmented Generation (RAG)—the process where AI platforms retrieve information from the web to generate answers. Semrush found that ecommerce and retail sites account for 76% of sources in AI Mode, compared to only 38% in ChatGPT. Conversely, knowledge bases account for 33% of sources in ChatGPT but only 6% in AI Mode.

To improve visibility in ChatGPT

These differences dictate where content teams should focus their efforts based on the target platform. To improve visibility in ChatGPT, brands should prioritize educational and informational content resembling a knowledge base. For AI Mode, the data suggests that visibility is more heavily driven by retail and shopping-related content. This distinction allows for a more granular content strategy where the same brand can optimize differently for different AI ecosystems.

Closing Citation Gaps via Outreach and Content

Closing a citation gap involves identifying third-party pages that are frequently cited but do not mention the user’s brand. A high “Cited for Prompts” count paired with a low “Mention rate” identifies a high-priority outreach target where the AI platform trusts the source, but the source has not yet included the brand’s information. (Mention rate: how often the specific brand appears in the AI generated results relative to the total number of prompts tracked.)

Users should analyze these high-leverage pages to determine if they cover a relevant

Users should analyze these high-leverage pages to determine if they cover a relevant use case or if the information provided is outdated. If a third-party site is a primary influence on an AI answer, the goal of the outreach is to get the brand included in that site’s content. If the brand is already cited but has a low mention rate, the focus shifts to optimizing internal product detail pages, applying product schema, and building more comprehensive buyer-focused content. (product schema: structured data code added to web pages to help search engines and AI understand product details.)

SEO Strategists using Semrush

SEO Strategists using Semrush can use the Sources report to identify specific content gaps that require immediate attention. By identifying third-party pages with high citation counts, these practitioners can add those specific URLs to their outreach email sequence to secure brand mentions. This moves the needle from generic link building to targeted AI visibility.

SEO Strategists can also compare AI citation data against their traditional keyword ranking report to identify where they are losing ground in AI-generated answers. This allows them to report on a “Mention rate” metric, showing how often the brand appears in AI answers relative to the total number of prompts tracked.

Beginner students skipping the RAG technicals

Beginner students skipping the RAG technicals should refer to basic “What is SEO” and “How to use keywords” tutorials instead. This analysis assumes a baseline understanding of how Retrieval Augmented Generation works and how to navigate professional SEO tool workflows.

Iker’s mattress brand: a citation gap walkthrough

Iker runs a mattress brand and wants to improve visibility for “shoulder pain” queries. After setting up a Prompt Tracking campaign for 20 prompts, Iker opens the Sources report and identifies a medical review site that is cited for 10 of those prompts, but Iker’s brand has a 0% mention rate. Iker identifies that site as a primary outreach target and creates a plan to get the brand included in that site’s content.

If Iker’s brand were already cited in 9 out of those 10 prompts but still had a low 10% mention rate, the strategy would flip. In that scenario, Iker would stop the outreach focus and instead work on optimizing his own product detail pages and applying more robust product schema to ensure the AI mentions the brand more frequently.

Practical uses for RAG data

• Identify high-leverage third-party sources for outreach by finding pages with high “Cited for Prompts” counts and low “Mention rates”.
• Determine platform-specific content priorities, such as prioritizing knowledge base content for ChatGPT versus retail content for AI Mode.
• Quantify competitive AI visibility gaps by using the “Domains” view to see how many times competitors are cited compared to the brand.
• Audit the impact of product schema on AI answers by monitoring if schema additions correlate with higher brand mention rates.
• Discover new customer questions to track by analyzing Reddit communities and using the Prompt Research tool to find questions people ask AI.

Replacing guesswork with Semrush data

Semrush Prompt Tracking reduces the need for guessing why a brand is missing from AI answers. Instead of manual observation and guesswork, users get empirical data on exactly which sources are being used instead of their own content.

To validate these findings

It is not yet established how these citation shifts impact total click-through traffic or conversion rates, nor is it clear if site owners can opt out of being used as a citation source. While Semrush provides these insights, the specific impact on organic search volume remains unconfirmed. To validate these findings, practitioners should correlate Semrush’s mention rate data with internal conversion data from their own websites.

Original reporting: How to track your AI citations and sources with Semrush (Semrush), on 7 October 2026.