# Does social media influence what LLMs say about your brand? What 38,177 cited sources show

> Yes, but not evenly: social platforms were 14.6% of Google AI Overviews sources and 1.1 to 1.5% on Claude or Gemini. Data from 11 brands, by platform.

**Published:** 2026-11-04 · **Category:** GEO

[Version HTML](https://www.citeme.io/ressources/does-social-media-influence-what-llms-say-about-your-brand)

---

Yes, social media presence influences what LLMs say about a brand, but not evenly. In the answers we tracked for eleven brands over 90 days, social platforms made up 3.2% of the 38,177 sources the engines cited. Google AI Overviews took 14.6% of its sources from social platforms. Claude, Gemini and Siri AI stayed between 1.0% and 1.5%. Where social counts depends on which engine your buyers use.

**In short**: social posts and videos get cited directly mostly in Google AI Overviews, and mostly from YouTube, Reddit and LinkedIn. The other engines rarely cite a social page, but they may still read profiles and mentions when they describe who you are. The first effect calls for content that answers buyer questions, the second for accurate profiles linked to your site.

## What our data shows about social sources in AI answers

Citeme tracks prompts for each brand on up to ten engines and records every source each answer cites. For this article we added up the raw counts across every active brand we track that had completed answers in the last 90 days, to 4 November 2026. That gives eleven brands, from at least five sectors: B2B software, marketing and design services, education, health and consumer goods. Most are French brands tracked on French-language prompts. We do not name them, and the numbers below are aggregates only.

Each cited source falls into one of five buckets: the brand's own site, a competitor's site, a social platform, Wikipedia, or any other third-party page.

| Source type | Citations | Share |
| --- | --- | --- |
| Other third-party pages (media, blogs, directories, comparison sites) | 31,704 | 83.0% |
| Competitor sites | 2,730 | 7.2% |
| The brand's own site | 1,665 | 4.4% |
| Social platforms | 1,214 | 3.2% |
| Wikipedia | 864 | 2.3% |

A brand's own site is a small slice of what gets cited about it. Most of what an engine says about you comes from pages you do not control. Social platforms are a minority source overall, cited a little more often than Wikipedia and less often than your own site.

One brand skews the social total. Its market is advertising on Meta's platforms, and its answers cited Facebook pages 557 times. Take it out and social falls to 515 of 29,995 citations, or 1.7%. We show both figures because the outlier is the point: when your category lives on a social platform, the engines cite that platform.

> **In short**: across 38,177 cited sources, social platforms were 3.2% (1.7% without one outlier brand). Most citations go to third-party pages, not to the brand or its profiles.

## Which AI engines cite social media the most?

Google AI Overviews, by a wide margin. For seven of the nine brands we track on it, AI Overviews had the highest social share of any engine.

| Engine | Sources cited | Social citations | Social share | Without the outlier brand |
| --- | --- | --- | --- | --- |
| Google AI Overviews | 2,023 | 296 | 14.6% | 13.0% |
| Mistral | 4,028 | 199 | 4.9% | 1.4% |
| ChatGPT | 4,294 | 177 | 4.1% | 3.7% |
| Perplexity | 4,360 | 146 | 3.3% | 0.9% |
| Grok | 3,056 | 102 | 3.3% | 1.3% |
| DeepSeek | 2,781 | 57 | 2.0% | 0.4% |
| Meta AI | 3,862 | 65 | 1.7% | 0.4% |
| Gemini | 5,289 | 80 | 1.5% | 0.6% |
| Claude | 5,579 | 64 | 1.1% | 0.5% |
| Siri AI | 2,905 | 28 | 1.0% | 0.2% |

Without the outlier, AI Overviews cited a social page in 13% of its sources. ChatGPT came second at 3.7%. Every other engine stayed under 1.5%.

Gemini is also a Google product, yet in our data it cited social pages in 1.5% of its sources, and 0.6% once the outlier is removed. Our reading, which we have not tested: AI Overviews sits on top of Google's search results, where YouTube videos and Reddit threads already rank for many questions. Gemini, Claude and Siri answer more often from articles and reference pages.

ChatGPT cited Wikipedia 274 times against 177 social citations. For a model that answers from the web, a clean encyclopedic entry still carries more weight than a feed.

Two caveats. Perplexity only returns part of its sources to us, so its figures are a floor. Not every brand is tracked on every engine: AI Overviews covers nine of the eleven brands, and two of the eleven are tracked on ChatGPT alone, which slightly inflates ChatGPT's weight in the totals.

> **In short**: if your buyers read Google AI Overviews, social content can be one source in seven. On Claude, Gemini or Siri AI it is between 1.0% and 1.5% of sources.

## Which social networks do LLMs cite?

YouTube, Reddit and LinkedIn, in that order, once you set aside the brand whose category is Facebook advertising. X barely exists as a source.

| Platform | Citations, 11 brands | Citations, without the outlier |
| --- | --- | --- |
| Facebook | 584 | 27 |
| YouTube | 223 | 174 |
| Reddit | 201 | 170 |
| LinkedIn | 153 | 115 |
| Instagram | 30 | 17 |
| TikTok | 22 | 11 |
| X | 1 | 1 |

One citation for X in 38,177 sources. We did not expect it to be that low. Whatever X does for your brand, it is not feeding AI answers as a cited page in the markets we track.

Our own brand follows the same pattern. On the citeme.io project, social platforms were 61 of 2,451 cited sources (2.5%). AI Overviews accounted for 45 of those 61. Perplexity, Claude, Meta AI, Siri AI, Mistral and DeepSeek cited no social page at all. The 61 split into YouTube 21, Reddit 20, LinkedIn 17, Facebook 2 and Instagram 1.

## Cited directly or read in the background: two different effects

A citation count only measures the visible effect. Social media can shape what an LLM says about you in two ways, and only one of them shows up as a link.

### The visible effect: your post or video is the source

This is what the tables above measure. A buyer asks a question, the engine searches, and a YouTube video, a Reddit thread or a LinkedIn post ends up in the list of sources. You can track it, prompt by prompt, and see whether the cited piece is yours, a competitor's or a stranger's. Note that a Reddit thread cited in an answer about your category is rarely written by you. It is usually other people talking about you, or about your competitors.

![Citeme view of a ChatGPT answer to a French prompt that cites the brand first, with the sources listed beside it](https://cdn.sanity.io/images/xc5dg2px/production/48985f45d81a4ebddf4bf9cac12ebb7da770e4e8-2632x2576.png)

_A citation as Citeme records it: the ChatGPT answer, the brand's position and the list of sources the engine used, here five websites._

### The invisible effect: profiles and mentions shape the description

Models can also learn from social content, and we think engines use it to check who a brand is, although no engine documents that step. The training side is partly public: Reddit announced data agreements with Google in February 2024 and with OpenAI in May 2024. No engine tells you which posts shaped a given answer, so this effect cannot be counted the way citations can.

What can be checked is what an AI finds when it looks at your profiles. We ran that check on our own accounts on 3 November 2026, with a model given web access and asked to describe each profile:

-   Our YouTube channel has 63 videos, 947 subscribers and a video from 29 October. The model described "Citeme" on YouTube as a gaming channel. The channel name does not match the brand, and the brand name is shared with other entities online. Activity alone did not fix it.
-   Our founder's LinkedIn profile was found and described accurately and in a positive tone. It also repeated an offer we withdrew in October 2026. A stale line in a bio becomes a stale fact in an answer.
-   The founder's X account, with no post since October 2024, was not found at all.

None of this appears as a citation. All of it changes what an engine can say about you when someone asks.

> **In short**: count citations to measure the visible effect. Check what an AI finds on your profiles to catch the invisible one: wrong identity, outdated facts, or nothing at all.

## What to do on each platform

Start from where your buyers ask. If Google AI Overviews matters to you, social content deserves real effort. If your market asks Claude or Perplexity, put that effort into third-party articles and your own pages first, and treat profiles as identity hygiene.

### YouTube

-   Title videos with the question a buyer asks, in their words. A video titled with the question is more likely to match it than a brand film; this is our reading, not a documented rule.
-   Name the channel after the brand, or put the brand in the channel description in the first line, with what you do and a link to your site.
-   Link back from your site to the channel, for example in your Organization schema's sameAs. Links in both directions make the tie easier to confirm; no engine documents how it checks this.
-   Use chapters and accurate captions, so the answer to the question can be found in the video's text.

### Reddit

-   Find the threads that already get cited for your category's questions. Those are where the conversation happens.
-   Answer as a named person who says who they work for, and follow each subreddit's rules. Undisclosed self-promotion can get removed, and a removed comment helps nobody.
-   Give useful specifics: a price, a limit, a comparison. A thread gets cited because it answers the question, and a vague answer does not.

### LinkedIn

-   Keep the company page and the founders' profiles factually current: what you sell, for whom, at what price if you publish it. Remove offers you no longer run.
-   Publish posts that make one checkable claim each, with a number or a date. Those are the posts an engine can quote.
-   Link the profiles to the site and the site to the profiles.

### Facebook, Instagram, TikTok and X

-   In our panel these were rarely cited, except Facebook for the brand whose market is Facebook advertising. If your category lives on one of these platforms, test it with your own prompts before you invest.
-   Otherwise, keep the profiles consistent with your site: same name, same description, a link to the site. An inconsistent profile can do more harm than a missing one.

> **In short**: YouTube and Reddit for AI Overviews, LinkedIn for identity and B2B answers, and consistent profiles everywhere else.

## Generative engine optimization (GEO) is not only about your website

The data above is why we say that AI looks at your socials too. Your site is 4.4% of what gets cited about you. Your profiles shape how an engine recognises you, and on Google AI Overviews your videos and threads can be the answer itself.

Citeme is our product, so read this part with that in mind. Here is what it does on the social side today:

-   Every tracked prompt records the sources cited in each answer, split into your site, competitors, social platforms, Wikipedia and other pages, per engine and per social platform. That is how the figures in this article were produced. See [citation tracking](https://www.citeme.io/features/citations) and [prompt tracking](https://www.citeme.io/features/prompt-tracking).
-   The social analysis checks each profile you declare (LinkedIn, YouTube, X, Instagram, TikTok, Reddit, Facebook): whether it has a description, links to your site, is linked back from it, matches your brand name and shows recent activity. It then asks an AI what it finds on the profile and whether it recognises your brand, and turns the gaps into recommendations. It is the check that caught our own YouTube problem.
-   The authority audit counts your presence across Wikipedia, Wikidata, the Knowledge Graph, social networks, directories and review platforms. See [brand authority](https://www.citeme.io/features/brand-authority).

To check one engine quickly, the free [AI Overviews tracker](https://www.citeme.io/tools/ai-overviews-tracker) shows which sources Google cites for a query. Plans and limits are on the [pricing page](https://www.citeme.io/pricing).

![Citeme audit setup listing Website, Press, LinkedIn, Reddit, YouTube and X as channels to add before the first audit](https://cdn.sanity.io/images/xc5dg2px/production/643a7c7b602e52f3d10aa20b2bfd2805271c8878-2880x2048.avif)

_Before the first audit, Citeme lists the places AI reads about the brand: the website and press are found, LinkedIn, Reddit, YouTube and X can be added, and Facebook, Pinterest and four more channels later._

> **In short**: measure which sources each engine cites for your prompts, then fix the profiles an AI cannot tie to your brand.

## FAQ

### Does social media presence influence what LLMs say about a brand?

Yes, in two ways. Engines cite social posts and videos directly, mostly Google AI Overviews: in our data, 14.6% of its cited sources were social platforms, against 1.0% to 1.5% for Claude, Gemini and Siri AI. Engines probably also read your profiles and mentions to work out who you are, which affects answers even when no social page is cited.

### Which social network matters most for AI visibility?

In our panel of eleven brands, YouTube (174 citations), Reddit (170) and LinkedIn (115) led once one outlier brand was set aside. Facebook only mattered for a brand whose market is Facebook advertising. X was cited once in 38,177 sources.

### Does ChatGPT cite Reddit and YouTube?

Sometimes. ChatGPT took 4.1% of its cited sources from social platforms in our data, third after Google AI Overviews and Mistral. Once the outlier brand is set aside, it is second at 3.7%, behind AI Overviews. It cited Wikipedia more often than all social platforms combined: 274 times against 177.

### Should I post more on social media to get cited by AI?

Not by default. Volume did not help our own YouTube channel, which an AI still described as a gaming channel because its name did not match the brand. Check first which engines your buyers use and which sources those engines cite for your prompts. Then fix identity (name, description, links both ways) before you publish more.

### How can I see which social sources AI engines cite for my brand?

Run your buyers' questions on each engine and note the sources under each answer. That works for a few prompts. A tracking tool such as Citeme, which is our product, does it daily across engines and splits the cited sources into your site, competitors, social platforms and Wikipedia.

## Sources

- [Cited sources across 11 tracked brands, 90 days to 4 November 2026 (internal tracking data, aggregated)](https://www.citeme.io/features/citations) · Citeme
- [OpenAI and Reddit partnership](https://openai.com/index/openai-and-reddit-partnership/) · OpenAI
- [An expanded partnership with Reddit](https://blog.google/inside-google/company-announcements/expanded-reddit-partnership/) · Google