Contents
July brings a new AI search visibility pattern
Top individual sources: Citation volume & trends
YouTube remains the strongest citation source
LinkedIn is now a primary GEO channel
Social and professional sources now sit closer to the center of AI visibility
Wikipedia, NIH and Statista lost ground together
Perplexity now needs a social and professional strategy
Earned media needs closer monitoring
What marketers should do now
Conclusion: GEO now depends on source mix
FAQ
July brings a new AI search visibility pattern
Generative AI platforms are creating new paths for customers to discover brands, and our July data gives PR and marketing teams a clear message: AI visibility now depends on matching content to the source ecosystems each model uses.
Meltwater’s July 2026 analysis is based on thousands of prompts across eight major LLMs: Claude, ChatGPT, Copilot, Google AI Mode, Google AI Overviews, Gemini, Perplexity and xAI Grok. The research compares July against June and places the results in the context of our ongoing tracking of AI search behavior - you can find our previous analysis here.
This research is carried out using GenAI Lens, Meltwater's AI search visibility tracking tool. Find out how GenAI Lens can help track your brand's performance in AI search.
In July, social, professional and video sources gained ground together, while several institutional and reference sources declined. YouTube reached a new high, LinkedIn moved from a supporting channel into the top five, Facebook held the #3 position, and Instagram stayed in the top 10 after its June surge.
At the same time, Wikipedia, NIH and Statista all fell. This creates a sharper divide between the social sources models use for broad visibility and the reference sources they use for institutional authority.
This changes the GEO playbook for marketing professionals. Visibility in AI answers now depends on a wider source mix that includes video, LinkedIn, social platforms, owned content, earned coverage and reference assets.
Top individual sources: Citation volume & trends
| Rank | Source | Total Citations | Vis. Score | Cit. MoM | Cit. Rate | Visibility Score MoM |
|---|---|---|---|---|---|---|
| 1 | youtube.com | 229.3k | 26% | +24.3% ↑ | +12.5% ↑ | +13% ↑ (18% rate score) |
| 2 | reddit.com | 128.2k | 18% | +8.24% ↑ | +9.09% ↑ | +5.88% ↑ (12% rate score) |
| 3 | facebook.com | 88.7k | 14% | +19.6% ↑ | 0% → | 0% → (9% rate score) |
| 4 | nih.gov | 81.1k | 12% | -9.65% ↓ | -16.7% ↓ | -7.69% ↓ (5% rate score) |
| 5 | linkedin.com | 80.3k | 14% | +67.8% ↑ | +28.6% ↑ | +27.3% ↑ (9% rate score) |
| 6 | wikipedia.org | 66.6k | 12% | -14.5% ↓ | -12.5% ↓ | -7.69% ↓ (7% rate score) |
| 7 | forbes.com | 62.0k | 12% | +15.9% ↑ | 0% → | +9.09% ↑ (7% rate score) |
| 8 | yahoo.com | 54.8k | 11% | +5.98% ↑ | 0% → | 0% → (7% rate score) |
| 9 | instagram.com | 53.2k | 11% | -2.37% ↓ | 0% → | 0% → (6% rate score) |
| 10 | statista.com | 50.5k | 11% | -7.32% ↓ | -14.3% ↓ | 0% → (6% rate score) |
YouTube remains the strongest citation source
YouTube produced the strongest single-source result in July, reaching 229.3K citations, up 24.3% month over month, making it the highest-volume source tracked so far.
This July rebound also resolves one of the open questions from June, when we saw YouTube’s first recorded dip, raising the possibility that its citation growth had started to slow. July’s data showed YouTube clearly gaining in citation volume, citation rate and visibility score at the same time.
Google AI Mode generated 65,340 YouTube citations in July. Google AI Overviews followed with 54,873, Gemini generated 51,761 and Perplexity generated 44,103. That spread shows how broadly YouTube now supports AI visibility across high-reach models.
Brands should treat YouTube as a core GEO asset. The main work is practical: improve transcripts, use clear titles, write detailed descriptions, cover topics thoroughly and structure videos so AI systems can extract the key answer quickly.
LinkedIn is now a primary GEO channel
LinkedIn produced the largest multi-metric gain of any established top-10 source since we began tracking this data in April. Citation volume grew 67.8%, citation rate increased 28.6% and visibility score rose 27.3%. LinkedIn also jumped from #8 in June to #5 in July, reaching 80.3K citations.
The model-level data shows where that growth came from. Perplexity generated 31,311 LinkedIn citations, while Google AI Mode generated 21,202. AI Overviews added 12,530, Copilot generated 8,283 and Grok generated 5,269.
This shows that LinkedIn now functions as an AI citation source across high-visibility models, especially for professional topics, business content and social proof. We explored the importance of LinkedIn as a B2B AI citation driver in a recent report.
That report, along with this more recent research, highlights some best practices for getting your LinkedIn content cited: Posts should state the core point early, include clear claims, use specific data where possible and avoid burying useful information in vague commentary. Executive posts, company research, industry explainers and practical POV content all have stronger GEO potential when they use direct structure and plain language.
Social and professional sources now sit closer to the center of AI visibility
| Source Type | July Count | July % | June % | MoM Change | Multi-Month Trend | Trend |
|---|---|---|---|---|---|---|
| Other | 4,500,000 | 56.4% | 55.2% | +1.2 pts | Third consecutive month of growth | ↑ Confirmed ascending |
| Earned/News | 2,700,000 | 33.1% | 35.0% | -1.9 pts | Second consecutive decline from 37–40% band | ↓ Under pressure |
| Social | 626,700 | 7.8% | 6.9% | +0.9 pts | New 5-month high; corroborated by individual sources | ↑ Confirmed ascending |
| Reviews | 126,700 | 1.6% | 1.6% | 0 pts | Stable across all 5 months | → Structural floor |
| Wiki | 68,200 | 0.9% | 1.0% | -0.1 pts | Second consecutive pullback from May’s 1.1% peak | ↓ Declining |
| Press Release | 17,700 | 0.2% | 0.2% | 0 pts | Flat for 3 months at new low base | → Stabilised |
Social reached 7.8% of citation domain count in July, its highest share so far, and unlike April’s short-lived spike, July’s growth came from several sources moving together.
Facebook held the #3 position overall with 88.7K citations, up 19.6% month over month. LinkedIn surged into the top five. Instagram held at #9 with 53.2K citations, down slightly after June’s large increase but still elevated compared with earlier months. Reddit also continued its impressive growth run, reaching 128.2K citations and holding the #2 position.
Social content now affects AI visibility in more than one way and the practical takeaway is this: you should track which social platforms feed which AI systems. Reddit may support category conversation, LinkedIn may support executive authority and business context, Facebook may carry high-volume model visibility, and Instagram may support consumer and visual categories.
Wikipedia, NIH and Statista lost ground together
July brought a clear reversal for several reference and institutional sources.
Wikipedia fell 14.5% month over month to 66.6K citations, its second consecutive decline after the March to May ascent. Statista fell 7.32% to 50.5K citations, also marking its second consecutive decline. NIH declined 9.65% to 81.1K citations, its first negative month after the major gains seen earlier in the series.
These sources still carry authority. NIH remains #4 overall, Wikipedia remains #6 and Statista remains in the top 10. The issue is growth priority. Brands should maintain accurate, neutral and well-structured presence across reference and data sources, but July shows that the strongest current growth sits elsewhere.
The right response is rebalancing. Keep Wikipedia accurate. Keep data partnerships and authoritative references current. Then direct incremental GEO effort toward LinkedIn, YouTube, social content and strong owned content.
Perplexity now needs a social and professional strategy
Perplexity changed more than any other model across June and July.
In July, Facebook led Perplexity’s citations with 45,049. YouTube followed closely at 44,103, LinkedIn reached 31,311, Instagram reached 25,527 and Reddit reached 24,010. NIH fell to 1,793 for Perplexity, down sharply from earlier months.
That source mix gives Perplexity a new profile, so we can no longer think of it as a purely research-led model, since it now pulls heavily from social, professional and video sources.
The practical consequences of this are that a Perplexity visibility strategy should include LinkedIn posts, public social content, YouTube transcripts and topic-specific content designed for quick extraction. Brands that only optimize for institutional citations may miss Perplexity’s current source behavior.
Earned media needs closer monitoring
Earned/News fell to 33.1% in July, down from 35.0% in June and below the 37% to 40% band seen from March through May. This marks the second consecutive decline.
PR still plays a major role in AI visibility because earned coverage provides third-party validation, category authority and source diversity. But the July data shows that PR needs to watch whether earned media is settling into a lower share of citation activity.
This creates a measurement challenge. PR professionals should continue pursuing high-quality editorial coverage, but they should also connect that coverage to other AI-citable surfaces. Strong media stories can feed LinkedIn posts, YouTube explainers, data pages, executive commentary and owned content. That content system gives AI models more ways to find and cite the same brand authority.
Press Release remained flat at 0.2% for the third consecutive month, suggesting that the category has stabilized at a low base. Press releases can support disclosure and media operations, but the data does not support treating them as a primary AI visibility lever.
What marketers should do now
1. Make LinkedIn content more structured
LinkedIn now deserves a place in the core GEO plan. Treat company pages, executive profiles and employee thought leadership as public citation surfaces.
Lead with the main point. Use specific claims. Add data when available. Keep posts focused on one idea. Create content that answers real category questions instead of posting generic brand updates.
2. Keep YouTube at the center of AI visibility planning
YouTube’s lead continues to grow. Brands should prioritize transcripts, descriptions, metadata, chaptering and topic coverage.
For GEO, the transcript often carries the most value. Videos should answer questions directly, use clear terminology and cover topics with enough depth for AI systems to extract reliable information.
3. Rebalance reference-source work
Wikipedia, NIH and Statista still support authority, but July shows declining momentum for all three.
Maintain accuracy and completeness across these sources. Then place new growth effort into LinkedIn, YouTube, social platforms and owned content that can answer high-intent questions clearly.
4. Treat Perplexity separately
Perplexity now relies heavily on Facebook, YouTube, LinkedIn, Instagram and Reddit. Brands should measure Perplexity visibility on its own rather than folding it into a general AI visibility score.
A blended view can hide large model-level differences. Track source pathways by model so you know whether visibility comes from social, professional, video, earned or institutional sources.
Conclusion: GEO now depends on source mix
July’s data shows a more divided AI citation environment.
YouTube remains the strongest source by a wide margin. LinkedIn has become a major source for professional visibility. Facebook, Reddit and Instagram continue to support social discovery. Owned and other web sources keep growing in aggregate.
At the same time, Wikipedia, NIH and Statista lost ground together, while Earned/News moved below its earlier band.
Brands should respond with a broader, model-specific GEO strategy. Build video content that AI can parse. Make LinkedIn content more structured. Keep reference sources accurate. Strengthen owned content. Track Perplexity, Gemini, ChatGPT, Claude, Copilot, Grok and Google’s AI experiences separately.
The brands that win in AI search will earn visibility across the sources each model already uses.
FAQ
1. What did the July 2026 AI citation data show?
The July data showed YouTube reaching an all-time high at 229.3K citations, LinkedIn growing 67.8% and Social reaching its highest share across the research. Wikipedia, NIH and Statista all declined in the same month.
2. Why is LinkedIn more important for GEO now?
LinkedIn moved to #5 overall in July and gained strongly across citation volume, citation rate and visibility score. Perplexity and Google AI Mode drove much of that growth, which means LinkedIn content now supports visibility across high-reach AI models.
3. Should brands stop investing in Wikipedia, NIH or Statista?
Brands should maintain those sources as authority assets. July shows they should not receive the same level of growth focus they had earlier in the year. Incremental effort should move toward LinkedIn, YouTube, social content and well-structured owned content.
4. How should brands measure AI visibility after July’s data?
Brands should measure AI visibility by model, source type and citation pathway. Track where each model pulls information from, because ChatGPT, Claude, Gemini, Perplexity, Copilot, Grok and Google’s AI experiences now show very different citation patterns.
