Find new opportunities across AI search and SEO.

AI search and SEO.

Try free for 7 days
Outperform your search competitors.
Outperform your search competitors.Search now spans Google, ChatGPT, and beyond—and Semrush gives you the SEO and AI visibility tools to stay visible everywhere:
See how AI platforms talk about you
Discover valuable topics to cover
Get AI-powered strategy suggestions
Build keyword plans with personalized insights
Identify fixes to improve technical health
Benchmark yourself against competitors
Outperform your search competitors.
10M marketing professionals have already used Semrush
Outperform your search competitors.
Outperform your search competitors.
Outperform your search competitors.

What is citation core? A guide to AI search visibility

Author:Asif Ali
19 min read
November 16, 2026
Contributor: Faizan Ali

AI platforms often rely on third-party websites when mentioning or recommending brands. That means improving your AI visibility requires work beyond your own website. Citation core gives you a practical way to focus that effort.

In this guide, you’ll learn what citation core is, how to find and prioritize the sites in your citation core, and how to get your brand featured on them.

Download our free citation core worksheet. Use it as you work through the steps below.

What is citation core?

Citation core is the small group of websites AI platforms repeatedly cite when answering questions about a specific category.

When ChatGPT, Gemini, Google AI Mode, and other AI platforms generate answers, they may link to the web pages they use as sources. These links are AI citations.

Across all the answers AI gives about a given category, a handful of websites get cited by multiple AI platforms far more often than the rest. Those websites make up that category’s citation core.

For example, when I searched “best contract management software” on ChatGPT, Juro appeared as a cited source.

A response to “best contract management software” on ChatGPT with the Juro citation highlighted.

Juro also appeared in Google AI Mode for the same prompt.

A response to “best contract management software” on Google AI Mode with the Juro citation highlighted.

And in Perplexity.

A response to “best contract management software” on Perplexity with the Juro citation highlighted.

And even in Claude.

A response to “best contract management software” on Claude with the Juro citation highlighted.

I also found Juro as a cited source on AI platforms for various other related prompts, such as “affordable contract management software” and “how to choose a contract management software.”

Juro is one example of a broader pattern: AI citations tend to concentrate around a relatively small set of recurring sources within a category.

A 2026 study of AI answers in four industries found that a small set of websites earned most of the citations in each industry, across every AI platform tested.

Industry-specific sources can dominate within their categories. In our AI Visibility Index 2026 study of 126 million prompts, we found NerdWallet, Investopedia, and Bankrate received more AI citations in the finance category than general business publishers like Forbes and The Wall Street Journal.

Top sources by AI citations in the finance category which includes brands like NerdWallet, Investopedia, and Bankrate.

In software in the U.S., the citation core included G2, Capterra, TrustRadius, and Forbes Advisor. In beauty, it included Allure and Byrdie. In travel, it included TripAdvisor.

But, these are examples of broad, category-level citation cores. The websites in your citation core will depend on:

  • Your prompt set: Your citation core is based on the full set of prompts you target. The topics, use cases, and buying stages covered by that set affect which websites qualify for your citation core.
  • Your target location: If you sell in several countries, each country can have its own citation core. For example, AI answers about cars lean on Kelley Blue Book and Edmunds in the U.S., AutoTrader and What Car? in the U.K., and mobile.de and Auto Bild in Germany.

Why citation core matters for AI visibility

Citation core matters because the sources AI relies on can influence whether your brand appears in an answer and how it is presented.

AI search often treats third-party sources as a key signal for which brands to mention. So if your brand appears consistently across the sites AI draws on for your category, it has a stronger chance of appearing in relevant AI answers.

Patagonia shows how repeated third-party coverage can support both brand mentions and authority. According to Semrush’s AI Visibility Index, six specialist outdoor-gear sites account for more than 65,000 Patagonia mentions across AI answers — more than all Tier 1 traditional media combined. 

A list showing six specialist outdoor-gear sites accounting for more than 65,000 Patagonia mentions across AI answers.

Coverage across citation core sites can shape how AI presents your brand. If citation core sites describe your brand positively on a consistent basis, AI may be more likely to present it favorably or include it when comparing and recommending options. 

Citation core sites can also influence buyers directly. Review sites, specialist publications, and communities help people compare products and evaluate alternatives, so positive coverage can shape which brands they choose.

How to find the sites in your citation core

Find the sites in your citation core by researching relevant prompts, collecting the sources AI platforms cite in their answers, and identifying the domains that appear across multiple platforms.

Follow these steps:

1. Define your scope

Define your scope by choosing the category, target location, and AI platforms you want to analyze.

Start with the specific market or product category you want your citation core to represent. Keep it narrow enough that the sources you uncover are relevant to the same buying context.

For example, “software” is too broad. “Contract management software” gives you a much clearer category to analyze.

Then, choose the geographic location where you want to improve your AI visibility. Citation patterns can vary between markets, so focus on the country or region where your target customers are based.

Finally, choose the AI platforms you want to analyze. If you already know which AI platforms your target audience uses, focus on those. If you don’t, start with ChatGPT, Google AI Mode, Gemini, and Perplexity to get a broad view across major AI search platforms.

You can also use your existing AI referral traffic as a signal. In Semrush’s Traffic & Market Toolkit, open the AI Traffic dashboard to see which AI platforms are already sending visitors to your site and competing sites.

AI Traffic report showing traffic distribution and traffic trend charts for three competing brands.

Treat AI referral traffic as a starting point rather than a complete picture. People can use an AI platform without clicking through to your website, so traffic alone doesn’t tell you exactly how much your audience uses each one.

Your scope might look like this:

  • Category: Contract management software
  • Location: United States
  • AI platforms: ChatGPT, Google AI Mode, Gemini, and Perplexity

Maintain the same scope throughout your analysis.

2. Build your prompt set

Your prompt set should cover the main ways your audience researches and evaluates products in your category.

Include prompts from different stages of the buying journey:

  • Awareness: Prompts about the problem or category, such as “What is contract management software?” and “How does contract management software work?”
  • Consideration: Prompts that help people explore possible solutions, such as “Best contract management software for startups” and “Juro alternatives”
  • Decision: Prompts that help people evaluate specific options, such as “Juro vs. Ironclad” and “Is Juro good for small businesses?”

You don’t need an equal number of prompts from each stage. Focus on the ones most relevant to how your audience researches the category, and avoid letting one topic or stage dominate your list.

I recommend starting with around 50 prompts. This gives you enough variety to uncover recurring citation patterns while keeping the analysis manageable.

You can get initial prompt ideas from questions customers ask during sales and support conversations, queries in Google Search Console, existing target keywords, and discussions in relevant online communities.

Treat these as starting points because they show what your audience searches for or talks about, rather than necessarily what they ask AI platforms.

You can also use Semrush’s Prompt Research tool to find prompts people use in AI search. Enter your category (e.g., contract management software) and click “Analyze.”

Prompt Research tool start with "contract management software" entered and "Analyze" clicked.

Select the country you defined in step one, then review the prompt data across the four AI platforms Prompt Research supports. If your analysis includes other AI platforms, you can still use the final prompt set you build here. In the next step, you’ll run that same prompt set across every platform in your scope, so you’re comparing citations against the same prompts.

Prompt Research report with the country and AI platform filters highlighted.

Scroll to the “Related topics” section and stay on the “Topics” tab. Review topics that have a “Best” or “High” relevance rating and clearly relate to how your audience researches your category. If two topics cover similar ground, prioritize the one with higher “AI Volume.”

Related topics on Prompt Research with the Relevance and AI Volume columns highlighted.

Click a relevant topic to see the prompts within it. Choose prompts that cover distinct research needs, and add them to a working list.

A topic expanded showing a list of prompts within it along with the AI response, brands, and sources.

Repeat this process across other relevant topics until you have a broad mix of prompts. 

Avoid filling your list with multiple variations of the same underlying query. For instance, I'd keep “best contract management software for startups” but skip similar prompts such as “top contract management software for startups.”

Once you have a working list, label each prompt with a buying stage — “Awareness,” “Consideration,” or “Decision.” Then check how many prompts you have at each stage and which major research areas they cover.

If most of your prompts fall into one stage, or an important area is missing, use the topic search field in Prompt Research to find prompts that fill those gaps. For example, search for a missing research area such as “security,” or use terms like “alternatives” and “pricing” to fill gaps at different buying stages.

"security" entered in the topic search field showing a list of related topics and prompts.

Continue until you have around 50 distinct prompts covering the main ways your audience researches and evaluates products in your category.

Record your final prompt set in a Google Sheet with columns for the prompt, buying stage, research area, and prompt source. Here’s what that can look like:

A spreadsheet showing a list of prompts along with the buying stage, research area, and source.

Further reading: How to do prompt research for AI SEO

3. Track your prompts and collect the citations

Record the domains and pages cited for every prompt in your final set, across each AI platform in your scope.

You can do this manually by running each prompt in each platform and noting every cited source in a Google Sheet. For each citation, record:

  • Prompt
  • AI platform
  • Source domain
  • Cited URL

While manually collecting citations is doable, it gets time-consuming quickly. If you track 50 prompts across four AI platforms, you need to check 200 responses and potentially record hundreds of citations.

You can speed up the process with Semrush’s Prompt Tracking tool, which tracks a custom set of prompts and records the domains and pages cited in responses from ChatGPT, Google AI Mode, and Gemini. Open the tool and configure your project. Select an AI platform, choose a target location, add your prompts, and click "Start Tracking." 

50 prompts added and "Start Tracking" clicked on the Position Tracking setup.

You'll repeat the setup for every supported AI platform in your scope. Check out our page on how Prompt Tracking works for a more detailed explanation.

Once Semrush has collected the results, go to the “Sources” tab. 

The "Sources" tab on Position Tracking showing data like tracked prompts, source pages, brand mentions, and sources by category.

Scroll to the “All sources” section to see which websites were cited across your tracked prompts, and expand any domain to see the individual pages AI cited from that site.

A domain expanded showing a list of pages cited on the "All sources" report.

Copy the citation data into a Google Sheet, including the prompt, AI platform, source domain, and cited URL for every citation.

Alternatively, you can export this report as a CSV. Then upload the file to an AI tool and ask it to return a cleaned table containing only the AI platform, prompt, source domain, and cited URL for each citation.

If you tracked multiple AI platforms, repeat this for each one and add the cleaned data to the same Google Sheet.

By the end of this step, you’ll have one Google Sheet containing the prompt, AI platform, source domain, and cited URL for every citation in your scope. This is how it may look:

A spreadsheet showing list of prompts along with the different AI platforms, source domains, and cited URLs for each.

4. Identify the sites in your citation core

Identify your citation core by determining the domains that appear across most of the AI platforms in your analysis.

There’s no universal cutoff for how many platforms a domain needs to appear on. For an analysis covering four platforms, I recommend starting with domains cited by at least three.

At this point, your existing Google Sheet contains all the citation-level data you collected in step three, including the prompt, AI platform, source domain, and cited URL.

Create a new sheet for your domain-level summary. List each source domain once, then add separate columns for each of your AI platforms. In each platform column, record how many different prompts produced an answer that cited that domain. Then add one more column showing how many of the four AI platforms cited the domain. Sort the sheet by the number of AI platforms from highest to lowest.

A list of domains along with the number of citations for each on ChatGPT, AI Mode, Gemini, and Peplexity.

You can build this summary manually by moving the relevant data into the new sheet and counting the prompts for each domain and platform. Or, use a pivot table to do the grouping for you. Set the source domains as “Rows,” the AI platforms as “Columns,” and the prompts as the “Values,” with prompts counted uniquely. This will show how many different prompts cited each domain on each platform.

The Pivot table editor highlighted on the right-hand side of a spreadsheet with a list of citation core domains.

You can also use an AI tool to create the summary. Simply give your Google Sheet to Claude or ChatGPT and tell it how you want the citation data grouped. Here’s a prompt you can use:

Once you’ve sorted the results, the domains cited across three or four platforms make up your working citation core.

5. Prioritize your citation core opportunities

Prioritize the sites in your citation core based on topic relevance, citation frequency, competitor gaps, and reachability.

Copy the qualifying domains from your domain summary into a new “Citation core” sheet. Add columns for “Topic relevance,” “Citation frequency,” “Competitor gap,” and “Reachability.”

A citation core spreadsheet with columns for source domain, topic relevance, citation frequency, competitor gap, and reachability.

Assess topic relevance

Assess topic relevance to identify which sites in your citation core are most closely aligned with your industry.

Manually check how often each site covers your industry, category, or closely related topics.

Then, in the “Topic relevance” column, rate each site as:

  • High: Regularly covers these topics
  • Medium: Covers them sometimes
  • Low: Rarely covers them
A citation core spreadsheet with the "Topic relevance" column filled out and highlighted.

Check citation frequency

Check citation frequency to identify which sites in your citation core AI platforms rely on most often.

Use the per-platform prompt counts you calculated in step four and add them together for each domain. Because the same prompt can cite a domain on multiple platforms, this total can be higher than the number of prompts in your set.

Record the total in the “Citation frequency” column.

Sites with higher totals are cited more often across your prompt set. That makes them more important to consider when deciding which third-party opportunities to prioritize.

A citation core spreadsheet with the "Citation frequency" column highlighted.

Find competitor citation gaps

To find competitor citation gaps, you’ll identify sites in your citation core that mention competitors but not your brand.

Use Semrush’s Competitor Research report in the AI Visibility Toolkit to find these gaps. Open the tool, add your domain and the competitors you want to compare, and click “Run competitor analysis.”

AI Competitor Gap Analysis with five competing domains entered and “Run competitor analysis” clicked.

Scroll down, switch to the “Sources” view, and select the “Missing” filter to see the domains AI platforms cite in answers that include your competitors but leave your brand out.

Missing sources on AI Competitor Gap Analysis showing domains cited in AI answers that leave out a selected brand but include its competitors.

Compare the domains in this “Missing” tab with the domains in your citation core.

In the “Competitor gap” column in your Google Sheet, mark each site as:

  • Yes: The domain appears in both your citation core and the Missing list
  • No: The domain is in your citation core but doesn’t appear in the Missing list

A “Yes” means the source is part of your citation core and is also being cited in AI answers where competitors appear but your brand doesn’t.

These sites deserve higher priority because competitors are already gaining visibility in answers associated with a source AI platforms cite in your category.

A citation core spreadsheet with the "Competitor gap" column highlighted.

Assess reachability

Assess reachability to identify which citation core opportunities you can realistically influence.

Reachability depends on how much control or access you have. Some sites let you update your presence directly, while others, such as competitor-owned domains, give you little room to influence inclusion.

In the “Reachability” column of your Google Sheet, rate each site as:

  • High: You can directly create, claim, or update your presence (e.g., G2, Capterra)
  • Medium: There’s a clear route through outreach, pitching, or an editorial process (e.g., an industry publication)
  • Low: You have little realistic influence over whether your brand appears (e.g., Wikipedia, a competitor-owned domain, a publication with no clear submission route)
A citation core spreadsheet with the "Reachability" column highlighted.

Rank your opportunities

Rank your opportunities to decide which sites in your citation core to work on first.

You now have the four signals you need in your citation core sheet. Use them to order the list:

  • Move sites with “Low” Topic relevance toward the bottom
  • Sort the remaining sites by Citation frequency from highest to lowest
  • Where citation frequencies are close, prioritize the site with higher Reachability
  • If the opportunities are still comparable, prioritize the site marked “Yes” for Competitor gap

For a short citation core list, you can apply these rules manually in Google Sheets.

For a larger list, upload the completed “Citation core” sheet to an AI tool and ask it to apply the same logic consistently. Here’s an example prompt you can use:

Import the CSV into a new “Ranked citation core” sheet. Your final sheet may look like this:

A ranked citation core sheet showing top-ranking domains along with topic relevance, citation frequency, competitor gap, and reachability.

Start with the top-ranking citation core domain and work down the list, focusing on the highest-priority sites you can realistically influence.

How to get featured on sites in your citation core

To get featured on sites in your citation core, create a reason for each site to mention, review, quote, or include your brand.

Here are some tips:

Claim and update your review platform profiles

Claim and maintain your product profiles on review platforms in your citation core so the information that’s within your control stays accurate and current.

If a platform such as G2, Capterra, TrustRadius, Gartner Peer Insights, or GetApp appears in your citation core, check whether your product already has a profile there. If it does, claim or verify it. If it doesn’t, create one if the platform allows vendors to do so.

Keep the information you control up to date, including your product description, categories, features, integrations, screenshots, pricing details, and other available profile fields.

You should also encourage customers to leave genuine reviews in line with the platform’s rules. Reviews give buyers and AI systems more current third-party context about how your product is used and perceived.

Give publishers original data they can cite

Share original data to give publishers in your citation core a new finding they can reference in relevant coverage.

Review the cited pages you collected earlier to understand the topics, claims, and questions each publisher covers. Then look for a question your company can answer with data that those publishers don’t already have. That could include:

  • Product usage data: How customers use specific features, workflows, or processes
  • Survey data: Opinions, behaviors, priorities, or challenges from a defined audience
  • Business data: Patterns in metrics such as deal size, sales cycle, renewal rate, or contract volume
  • Original analysis: New findings created by analyzing proprietary or public datasets

Package the strongest findings with a short methodology and a link to the research. Send them to the relevant writer or editor and explain where the data fits their coverage.

Build research partnerships with academic sites

Partner with universities, research centers, or .edu sites in your citation core when your company can contribute to the work they’re already conducting.

Look for relevant faculty members, labs, or programs, then explore ways to collaborate on research, industry projects, or educational initiatives. That could include:

  • Joint studies: Co-develop a survey, benchmark, or research project
  • Research access: Provide participants, anonymized datasets, or product access
  • Industry collaboration: Contribute practitioners, case material, or technical expertise
  • Educational projects: Support student research, capstone projects, or guest teaching

For example, HubSpot has partnered with universities such as Howard University and William Paterson University on digital business and marketing education. Those collaborations have led to university pages and programs that reference HubSpot directly.

A HubSpot education partner program page on the William Paterson University website.

Offer partnerships to review and comparison sites

Offer review and comparison sites a commercial partnership when their business model includes recommending products to their audience.

Check the cited pages you collected earlier to see which sites publish reviews, comparisons, or “best X” content where your product could fit. Then approach them with a concrete partnership offer. 

Your partnership offer might include:

  • Affiliate commission: A percentage or fixed amount for referred customers
  • Referral fee: Compensation for qualified leads or purchases
  • Exclusive offer: A discount or promotion for the publisher’s audience
  • Free product access: An account or trial they can use to evaluate the product

Include the commercial details upfront, such as commission or referral rates, average order value, attribution terms, and any audience-specific offer.

Participate in the communities that AI cites

Participate in communities such as Reddit and Quora when they appear in your citation core to build a credible presence in discussions AI platforms already cite.

Use the cited URLs to identify which subreddits, Quora topics, and recurring question themes appear most often. Contribute answers with specific advice, examples, or firsthand expertise that make each response useful on its own.

Mention your product only when it directly helps answer the question, and explain the relevant use case rather than dropping a promotional link.

Further reading: Reddit marketing: How to do it right (+ 7 tips to get results)

Use public relations to build major-media coverage and notability

Use public relations (PR) when major publications or Wikipedia rank highly in your citation core but offer little direct path to inclusion.

For major media, a PR agency or consultant can help develop newsworthy angles, package original research, and connect your experts with journalists covering your category. 

PR can also support Wikipedia notability by helping your company earn substantial, independent coverage in leading publications.

Treat public relations as a longer-term investment when a major publication or Wikipedia scored low on reachability in your analysis, rather than as an immediate citation tactic.

How to track progress on your citation core

Track progress by re-running the same prompts and checking whether placements you earned on citation core sites translated to more brand mentions in AI answers.

You can track these changes manually by running the prompts on a monthly or quarterly cadence. Or, use Semrush’s Prompt Tracking to monitor ChatGPT, Google AI Mode, and Gemini automatically. Semrush runs your tracked prompts daily and records changes in brand mentions, visibility, cited domains, and cited URLs.

Look at “Your brand mentions” to see whether your brand starts appearing in more of the AI answers generated for your tracked prompts.

"Your brand mentions" metric highlighted on the "Sources" tab on the Position Tracking tool.

Then search the “All Sources” table for the specific citation core domains where you earned a placement. Look at “Your mentions” and “Mention rate” to see whether AI answers citing those sources are mentioning your brand more often.

A URL entered along with the mentions and mention rate columns highlighted on the All sources report.

If your citation core strategy is working, your brand should appear on more of the citation core sites you targeted and in more AI answers for relevant prompts across multiple AI platforms.

Frequently asked questions

Is a citation core just a list of the most-cited websites?

No, a citation core is the small group of sources that appear consistently across multiple AI platforms for the prompts you analyze.

Which sites qualify depends on your category, prompt set, target location, and the AI platforms included. A site can be cited frequently on one platform and still sit outside the core if it doesn’t recur across the others.

Is citation core the same for every company in a category?

No, companies in the same category can have overlapping but different citation cores because they may target different markets, use cases, prompt sets, and AI platforms.

Can competitor-owned websites be part of your citation core?

Yes, competitor-owned websites can belong in your citation core if it meets the same citation criteria as any other source.

You may have little ability to earn a mention on that site, so it would typically receive a lower reachability rating when you prioritize your opportunities.

Does citation core still matter when AI tools rely on training data?

Yes, some of the data AI models are trained on comes from web content, so mentions on sites in your citation core can still contribute to what AI systems know about your brand. This can be true even when they aren’t searching the web in real time.

Does getting featured on a citation core site guarantee more AI mentions?

Getting featured on a citation core site can increase your chances of earning more AI mentions because your brand is now present on a source AI platforms repeatedly cite for relevant topics and prompts.

It doesn’t guarantee inclusion in every answer, so track whether your brand mentions increase after the placement.

How do you analyze citation gaps in AI?

Analyze AI citation gaps by comparing the sources in your citation core with the sources that mention competitors.

Look for citation core sites where competitors appear in AI answers and your brand doesn’t. Those gaps show where earning a mention could help close an existing visibility disadvantage.

How do you benchmark your brand’s AI citations against competitors?

Benchmark your brand’s AI citations by tracking the same prompts across the same AI platforms for your domain and competitor domains.

Compare how often each domain is cited, which pages earn those citations, and which prompts and platforms generate them. You can then compare citation core sources that mention competitors to identify off-site visibility gaps.

How often should you refresh your citation core analysis?

Refresh your citation core analysis whenever citation patterns may have materially changed, such as after several months, a major market shift, or changes in the AI platforms you track.

For most teams, re-running the analysis every quarter is a practical starting point.

Share
Author Photo

Asif is a content marketing specialist with over eight years of experience. While working at agencies and startups, he has helped brands drive millions in sales through full-funnel inbound marketing campaigns.

Share
Add Semrush as a preferred source in Google SearchTells Google to show you more from Semrush in AI, Search, and Discover.

Most popular pages