LLM visibility vs. SEO KPIs: Bridging the gap

 Author: Crystal Carter

Large language models (LLMs; e.g., ChatGPT, Perplexity, Gemini, etc.) have emerged as the new frontier for digital marketing. While it’s a very exciting time, this nascent tech also has SEOs scratching their heads over how to gauge their brand visibility in LLMs—and tie that visibility to business outcomes.

When it comes to measuring the impact of SEO for generative search and LLMs, there are a lot of overlaps between KPIs for traditional search engines and how you measure brand visibility in generative search tools.

In this blog post, I’ll walk you through these nuances so that you can start tracking your visibility in LLMs and use that data to optimize for even more mentions and, ultimately, site traffic.

Table of contents:

• LLM visibility for brands: The metrics that matter

• Brand mentions vs. Share of voice

• Quality of brand mentions vs. Keyword rankings

• LLM referral traffic vs. Organic search traffic

  • Track LLM traffic in GA4

  • Track LLMs referrals in your website analytics

    • How to track AI referrals in Wix Analytics

• Monitor LLM crawl bots in your bot log reports

• Retrieved pages vs. Indexed pages

• LLM website citations vs. Backlinks

• Conversions and leads from LLMs

LLM visibility for brands: The metrics that matter

Below is a table of LLM KPIs and their analogous SEO KPIs. Some may seem similar, but there are important nuances to keep in mind, especially when contextualizing your reporting.

Metric category	LLM KPIs	Traditional SEO measurement	Bridge the gap
Brand mentions	Brand mentions in LLM responses for relevant topics	Share of voice	Benchmark yourself with competitors for prompts around your core topics.
Value of brand mentions	Frequency, accuracy, sentiment, and positioning of brand mentions in LLM responses	Keyword ranking	Use keywords research to create prompts.
Traffic	Website traffic from LLMs	Organic traffic from search engines	Set up GA4 explorations to identify your traffic. Include it in your SEO reports.
Bot log analysis	Bot traffic from LLMs	User agent analysis	Monitor traffic from these bots.
Retrieved pages	Total number of pages that are known to LLMs	Total number of indexed pages	Identify the most valuable pages for LLM impact and ensure they appear.
Website citations	Website sources in LLM responses	Backlinks, referring domains	Audit your backlinks to understand which percentage of your backlink sources are included in LLM training.
Engagement/ Conversions	Conversions and leads from LLMs	Goal completions, eCommerce transactions, form submissions	Update your “how did you find us” surveys to include an option for ChatGPT or AI Assistant

Brand mentions vs. Share of voice

An example of a brand mention for Wix.

In generative search, brand mentions refer to how often your brand name or website appears in LLM responses to relevant topics. This is similar to share of voice in classic SEO, where you measure how visible your brand is compared to competitors in search rankings. SEOs who are already monitoring share of voice can bridge the gap by benchmarking their brand mentions against competitors. 

To do this, you would analyze LLM responses to prompts related to your core topics, then compare how frequently you appear compared to your close competitors. You can do this using LLM monitoring tools or manually with an LLM brand visibility tracker sheet.

Example of mentions tracking in AI brand monitoring tool Peec AI

This will help you see how often your brand is mentioned relative to others, giving you an understanding of your performance in the LLMs that you can work to improve.

Many tools for LLM brand monitoring include an option for tracking competitors. Just as in classic SEO SERP analysis, it’s worth reviewing business-critical LLM responses to see if there are competitors that are known to LLMs but not known currently on your radar. 

Quality of brand mentions vs. Keyword rankings

Not every mention is equal. Ideally, the quality of brand mentions in LLMs should depend on whether it’s a ‘RAPP’ mention:

• Regular — How often your brand or website are mentioned in outputs over time.

• Accurate — How factually accurate and up-to-date the information is about your brand.

• Prominent — Whether you are frequently listed first in responses.

• Positive — Whether your brand is mentioned favorably.

Here, the combination of frequency and positioning most closely aligns with keyword ranking in classic SEO. And just as higher SEO rankings generally lead to more traffic, more frequent mentions are likely to drive more clicks and potentially more searches of your brand via LLMs. By analyzing how frequently and where your brand appears in these responses, you can gauge the value of your brand’s visibility within the LLM.

Have LLM brand mentions replaced keywords?

Rather than replacing keywords entirely when you’re trying to understand the quality of your brand mentions in LLMs, you can use your target SEO keywords and PAAs to generate prompts that would most likely output a response that includes mentions of your brand or website. 

In the example below, I used keywords related to Site of Sites (a website that showcases web designs) and asked ChatGPT to “create a collection of prompts that a user might enter into ChatGPT to find a website where they could showcase their website design.” Then, I explained that the user might be “a freelancer or an agency professional,” provided a list of keywords, and then asked it to create a number of prompts. 

You can scale this process by bulk uploading and/or using tools like GPTforWork, but the value remains the same. At the moment we have limited visibility on search volume regarding LLM queries; however, the extensive historic keyword research data that you have still has value from a research perspective with regards to user demand and interest. 

LLM referral traffic vs. Organic search traffic

Measuring website traffic from LLMs is new, but we can use many of the tools we use to track organic traffic from search engines and crawl bots to track LLM traffic as well. 

Track LLM traffic in GA4

Analytics tools like Google Analytics 4 (GA4) can be configured with explorations and reports to identify and analyze traffic coming from LLMs. Include this data in your SEO reports to get a holistic view of your website traffic. This will help you understand how LLMs are interacting with your website. You can use this data to inform plans for growth and to serve as proof of concept before investing in generative search optimization. 

Using a regex like the code below, you can isolate page referrer traffic to see sessions in real time from a range of AI tools:

^https:\/\/(www\.meta\.ai|www\.perplexity\.ai|chat\.openai\.com|claude\.ai|chat\.mistral\.ai|gemini\.google\.com|bard\.google\.com|chatgpt\.com|copilot\.microsoft\.com)(\/.*)?$

SEO expert Jes Scholz has been able to use this to incorporate LLM traffic into her SEO reports. 

Track LLMs referrals in your website analytics

Within your web platform’s tools, you can see where you are getting referral traffic from LLMs. Simply go to your referral report and filter by the relevant LLM.

How to track AI referrals in Wix Analytics

To see your referrers from LLMs and generative AI search tools in Wix Analytics, go to the Top Traffic Sources report, and filter by sources that include AI tools.

When reporting to stakeholders, it’s important to note that not all of the traffic from LLMs is attributed as “referral” traffic. In some cases, clicks from ChatGPT and other LLMs are seen as “direct” traffic. This means that information from these reports should be seen as a trend line rather than a hard metric. You should expect to see the amount of traffic increase, but the exact amount of traffic may be different in reality. 

Monitor LLM crawl bots in your bot log reports

Analyzing your bot logs can help you:

• Assess whether LLM bots are visiting your website

• See which pages they are (and are not) crawling

• Make technical SEO recommendations

Similar to how you analyze user agents for SEO, your bot logs can help you see if LLM crawlers are encountering errors or issues when accessing certain pages or content, how often and which page they are crawling. Additionally you can compare them to traditional search engine crawlers, like Googlebot or Bingbot, to see if they are behaving differently. For example, looking at the Wix Bot Traffic over Time report, I can see that Common Crawl (a bot that is regularly used for LLM training) is crawling most of the pages on this small site. With this information, I may choose to update my robots.txt to manage this crawl. 

Retrieved pages vs. Indexed pages 

In classic SEO, we prioritize making sure that pages are crawled and indexed before they can rank. For LLMs, it’s worth understanding which pieces of content are retrieved and used for LLM responses, and which ones are not. 

For instance, if you have a very important product guide with some critical information that is not being retrieved by an LLM, then you might need to review why that’s happening. Additionally, if you have pages that are regularly retrieved by LLMs, then it might be that you need to improve the internal links from that page in order to support search-assisted LLMs. 

LLM website citations vs. Backlinks

Is there some overlap between how you assess website citations in LLM responses and how you track backlinks and referring domains in classic SEO? Certainly. In both cases, you’re getting an indication of whether your website is considered a relevant source of information. And in both cases, quality and volume matters. 

Additionally, the work you do on your backlink strategy will support how your brand performs in LLMs. To measure the impact here, you can audit:

• The percentage of your backlinks that come from web sources that are known to be included in LLM training data
  

• How frequently your backlink referrals appear in LLM responses
  

• Whether the context of your links and mentions is reflected in LLM responses

Auditing backlinks for LLM impact can help SEOs illustrate the value of their work for generative search. This will help you understand the correlation between your backlink profile and your website’s visibility in LLMs. Furthermore, this can inform your backlink strategy, so you can potentially improve your website’s chances of being cited in LLM responses.

Finally, you can compare the sites you are receiving backlinks from with sites that are confirmed content partners for LLMs. Major platforms have direct content partnerships with major publishers. This means that getting mentions in these publications can help you get included in the training data for these models. Perplexity, for instance, has revenue sharing partnerships with over a dozen publishers, including ADWEEK, Blavity, Los Angeles Times, Mexico News Daily, Prisa Media, and World History Encyclopedia. And OpenAI has partnerships with some of the web’s largest publishers as well. 

OpenAI Media Partnerships

Media Company	Notable publications/websites
Axel Springer	Bild
Associated Press	AP News
Atlantic Media	The Atlantic
Vox Media	Vox
News Corp	The Wall Street Journal, MarketWatch, The Times
Dotdash Meredith	Various outlets
Financial Times	Financial Times
Le Monde	Le Monde
Prisa Media	El País, Cinco Días, As, El Huffpost
TIME	TIME Magazine
Condé Nast	Vogue, The New Yorker, GQ, Wired, Vanity Fair
Hearst	Houston Chronicle, ELLE, San Francisco Chronicle, Runner's World
Future	Marie Claire, TechRadar, Cycling Weekly
Axios	Axios
Schibsted Media Group	Various Nordic publications

Those working on building brand authority with digital PR should prioritize links and coverage in these publications when assessing the value of backlinks.

Conversions and leads from LLMs

The nature of organic traffic is changing rapidly due to Google algorithm updates and new SERP features, like AI overviews, so conversion rate optimization has never been a more important consideration for SEO professionals. Whether you’re tracking goal completions, eCommerce transactions, or form submissions, it’s worth monitoring for the events that drive genuine value for your business. The challenge with generative search is that there are fewer established attribution methods and tools currently available. 

To bridge the gap, update your ‘how did you find us’ forms to include options like “ChatGPT” or “AI Assistant.” This will help you attribute conversions and leads specifically to LLMs, allowing you to measure the direct impact of your LLM optimization strategies on your business goals. 

Seer Interactive’s, Wil Reynolds has been using these forms to monitor this traffic because , as he explains, “I was seeing traffic from ChatGPT and Perplexity growing, but I wanted to be able to track if that traffic was more likely to turn into qualified leads than search traffic, so I had to add that in the drop-down to be able to track it through to closed deal.” 

LLM optimization is an evolution of SEO

This era of LLMs presents both challenges and opportunities for brands and SEOs. By embracing the right tools and strategies, search marketers can effectively monitor and enhance their LLM visibility. From leveraging dedicated LLM brand monitoring tools to optimizing website content and backlink profiles, proactive steps can ensure your brand remains relevant and discoverable in this new paradigm of search. 

As LLMs continue to evolve, staying informed and adapting your SEO strategies will be crucial for maintaining a competitive edge and maximizing your brand's reach in the AI-driven future of search.

---

Crystal Carter - Head of SEO Communications, Wix 

Crystal is an SEO & digital marketing professional with over 15 years of experience. Her global business clients have included Disney, McDonalds, and Tomy. An avid SEO communicator, her work has been featured at Google Search Central, Brighton SEO, Moz, DeepCrawl, Semrush, and more. Twitter | Linkedin