AI search citations keep shifting from month to month. A source that AI engines lean on heavily one month can lose ground the next.
It’s a real problem when you’re a marketing leader with a fixed budget. You can’t spend everywhere, and you can’t rewrite your strategy every month.
Meltwater’s AI Search Visibility Report shows how quickly the AI citation sources change. It tracked about 7.3 million citations across eight AI platforms, and its summary highlights three shifts from August 2026.
For teams running Generative Engine Optimization (GEO), each shift changes where effort should go.
In this article, we break down each shift, flag the data’s limits, and share four moves for your GEO plan. That way, you’ll know where to put your next dollar of content and PR effort.
| Shift | The numbers (Meltwater AI Search Visibility Report, August 2026) | What it means |
|---|---|---|
| LinkedIn enters the top three | 100.2K citations, up 24.8% from July. Moved from fifth to third. Perplexity generated 69,302 of them. ChatGPT and Gemini recorded zero LinkedIn citations in the tracked data. | LinkedIn content now feeds AI answers, mostly through one engine. |
| Earned media drops below 30% | 29.4% of citation domain count, down from 33.1% in July. About 2.1M citation domain counts, second only to the “Other” category. | News coverage still feeds AI answers. Measure PR by citations as well as coverage. |
| Platforms split | YouTube (212.5K) and Reddit (127.0K) stayed the two most-cited domains. Facebook fell 29.6%. Instagram fell 43.1%. The Social category’s share slipped from 7.8% to 7.2%. | Report each platform on its own line. |
What the Meltwater AI Search Visibility Report Tracks
Meltwater publishes a regular AI Search Visibility Report, with recent editions released monthly. It uses the company’s GenAI Lens tool to record which websites AI platforms cite in their answers.
The August 2026 edition was published on September 11, 2026. It analyzed about 7.3 million citations from thousands of prompts. It tracked these eight AI platforms.
- Anthropic Claude
- ChatGPT
- Microsoft Copilot
- Google AI Mode
- Google AI Overviews
- Google Gemini
- Perplexity
- xAI Grok
The report compares August with July and places both within a six-month tracking window.
The prompts covered high-intent consumer categories, including AI, retail banking, automotive, fast food, streaming services, apparel, and fashion. None named a brand. The report lists no B2B categories, so B2B teams should treat the findings as signals.
The report counts citations, which Meltwater’s developer documentation describes as sources an AI model references to support its answer.
It also reports a visibility score and each category’s share of “citation domain count.” The report doesn’t define these two metrics, so we use Meltwater’s labels as written.
This is one vendor’s tracking method. Use it as a directional signal for planning, and track your own category directly.
Shift #01: LinkedIn AI Citations Climb Into the Top Three
Perplexity drove most of LinkedIn’s AI citations
LinkedIn citations by AI platform, August 2026 (100.2K total)
Perplexity
Google AI Mode
Google AI Overviews
Microsoft Copilot
xAI Grok
Anthropic Claude
ChatGPT
Google Gemini
Source. Meltwater AI Search Visibility Report for August 2026, published September 11, 2026. Per-engine counts as published.
LinkedIn was the only top-three source to grow in August 2026. Its citations rose 24.8% to 100.2K, lifting it from fifth to third place in Meltwater’s tracking.
Its citation rate rose 11.1%, and its visibility score rose 14.3%. Meltwater notes that all three metrics climbed together.
The climb started after June. LinkedIn ranked eighth in June, fifth in July, and third in August. In July alone, its citations grew 67.8% to 80.3K.
Perplexity Drives Most of the Growth
Perplexity generated 69,302 of LinkedIn’s August citations, roughly 69% of the total. Meltwater calls it the largest model-source pairing in its six months of tracking.
The other engines cited LinkedIn far less. ChatGPT and Gemini recorded zero LinkedIn citations in the tracked data.
| AI platform | LinkedIn citations (Meltwater AI Search Visibility Report, August 2026) |
|---|---|
| Perplexity | 69,302 |
| Google AI Mode | 13,855 |
| Google AI Overviews | 8,232 |
| Microsoft Copilot | 3,941 |
| xAI Grok | 3,540 |
| Anthropic Claude | 699 |
| ChatGPT | 0 |
| Google Gemini | 0 |
Per-engine counts are as published. They add up to about 99,600, slightly below the 100.2K total Meltwater reports.
The engine mix also narrowed. In Meltwater’s July report, Google AI Mode cited LinkedIn 21,202 times. In August, that fell to 13,855, while Perplexity’s count more than doubled from 31,311.
Individual profiles earned most LinkedIn AI citations
Share of LinkedIn citations by source type, B2B prompts (about 75% individual overall)
Individual member profilesCompany Pages
Google AI Overviews
Google AI Mode
Claude Sonnet 4
GPT-5
Microsoft Copilot
Source. Meltwater and LinkedIn, “How LinkedIn Content Wins in AI Search,” May 2026. Perplexity was not tracked in this study.
That concentration is a risk. Among the most-cited domains, Perplexity’s top source was YouTube in May, Facebook in July, and LinkedIn in August. One model update could shrink LinkedIn’s numbers as fast as they grew.
So check where your buyers research. If they use Perplexity, LinkedIn content deserves a deliberate plan. Keep working on the other engines too. Our guide to improving brand visibility in AI search engines covers Perplexity, ChatGPT, and Google’s AI tools.
What a Separate LinkedIn Study Adds
Meltwater and LinkedIn published a separate B2B study in May 2026, called “How LinkedIn Content Wins in AI Search“; it analyzed 9.5 million citations across 16 B2B categories.
About 75% of LinkedIn citations in that study came from individual member profiles. The other 25% came from Company Pages.
Perplexity had a different top source in May, July, and August
Leading domain on Perplexity among Meltwater’s most-cited domains
May 2026
YouTube
Top source among leading domains
July 2026
45,049 citations
August 2026
69,302 citations
June 2026 data not stated in Meltwater’s reports.
Source. Meltwater AI Search Visibility Reports for June 2026 (May data), July 2026, and August 2026.
Individuals led on all five engines in the study’s per-engine chart. The share ranged from 83.1% on Google AI Overviews to 53.9% on Microsoft Copilot.
The May study didn’t track Perplexity. So its 75% figure can’t explain August’s Perplexity-driven growth.
Shift #02: Earned Media Loses Share of AI Citations
Earned media’s share of AI citations fell below 30%
Earned/News share of citation domain count, by data month, 2026
Source. Meltwater AI Search Visibility Reports for June 2026 (May data), July 2026, and August 2026. Shown by data month.
Earned/News, the category Meltwater uses for earned media, fell to 29.4% of citation domain count in August 2026. That’s down from 33.1% in July.
Meltwater calls it the third consecutive monthly decline, counting from May. Its published figures show the share falling every month since April. It was also the first reading below 30% in Meltwater’s six months of data.
| Data month | Earned/News share of citation domain count (Meltwater AI Search Visibility Reports, 2026) |
|---|---|
| April | 39.5% |
| May | 37.6% |
| June | 35.0% |
| July | 33.1% |
| August | 29.4% |
Shown by data month. Meltwater’s report titled “June 2026” covers May data.
Earned/News recorded about 2.7 million citation domain counts in July and about 2.1 million in August.
Meltwater’s broad “Other” category held at about 4.5 million in both months. Those rounded counts point to fewer earned citations as the main driver of the share drop.
Even after the drop, Earned/News outranked every category except “Other.” Social stood at 522.0K citation domain counts, and Reviews at 91.6K.
What This Means for Your PR Program
Some will read this as the end of PR. However, the data supports a narrower conclusion. Earned coverage still reaches AI answers, and its share has fallen every month since April, on Meltwater’s published figures.
Two changes follow.
- Measure coverage by AI citations: Track whether AI engines cite your coverage in your category. A clip count can’t show that.
- Look past the biggest outlets: CRN’s Mark Haranas topped Meltwater’s list of most-cited journalists and creators for a fourth straight month. Six new names joined the top 10, including writers at Xtalks and Athletech News.
Meltwater notes that one month of turnover doesn’t make a trend. Still, its data suggests widening your media list to niche reporters who cover the exact questions your buyers ask AI.
Shift #03: Social Media AI Citations Split by Platform
Social platforms moved in opposite directions
Citations in August 2026 and change from July
Down 7.34%
212.5K
YouTube
Down 0.95%
127.0K
Up 24.8%
100.2K
Down 29.6%
62.4K
Down 43.1%
30.3K
Source. Meltwater AI Search Visibility Report for August 2026, published September 11, 2026.
The Social category’s share of citation domain count slipped from 7.8% in July to 7.2% in August 2026, according to Meltwater’s August report. That small dip hides big moves on individual platforms.
| Platform | Citations (Meltwater AI Search Visibility Report, August 2026) | Change from July |
|---|---|---|
| YouTube | 212.5K | Down 7.34% |
| 127.0K | Down 0.95% | |
| 100.2K | Up 24.8% | |
| 62.4K | Down 29.6% | |
| 30.3K | Down 43.1% |
Facebook and Instagram Fell Sharply
Facebook dropped 29.6% to 62.4K citations, giving back its July gain. On Perplexity alone, Facebook citations fell from 45,049 in July to 8,994 in August.
Instagram dropped 43.1% to 30.3K. Its citation rate and visibility score fell too. Meltwater notes that the June surge on both platforms hasn’t turned into a lasting climb.
YouTube and Reddit Held the Top Two Spots
YouTube stayed the most-cited domain at 212.5K, down from a record 229.3K in July. Its citation rate and visibility score held steady.
Most of YouTube’s citations came from Google’s engines. Gemini cited it 68,073 times, Google AI Mode 68,011 times, and Google AI Overviews 38,217 times.
Reddit held almost flat at 127.0K. Its citation rate rose 16.7%, and its visibility score rose 5.56%. Gemini, Perplexity, Google AI Mode, and Google AI Overviews each cited it more than 16,000 times.
ChatGPT is the gap. In Meltwater’s August data, ChatGPT recorded zero citations for YouTube, Reddit, LinkedIn, Facebook, and Instagram.
Here’s what this means for your social media budget.
- Track each platform separately: The Social category’s share of citations dipped only 0.6 points in August. Yet LinkedIn grew 24.8%, and Instagram dropped 43.1%.
- Post where your buyers’ AI tools look: Among top domains, Google’s AI tools cite YouTube most, and Perplexity cites LinkedIn most. ChatGPT cited none of these five platforms in Meltwater’s August data.
- Wait before cutting Facebook or Instagram: Although both platforms fell hard in August, Facebook grew in July, so one month isn’t enough to move budget.
How to Get More AI Search Citations
Here are four GEO moves that fit a fixed budget. Each one starts small, so you can scale what earns citations in your category.
1. Benchmark Your Citations Engine by Engine
AI engines cite different sources, and their choices change. Among the most-cited domains, Perplexity’s top source moved from YouTube in May to Facebook in July to LinkedIn in August.
So start with your own baseline. In our free audits, we run 20 buyer-intent prompts across ChatGPT, Perplexity, Gemini, and Google AI Overviews, for both the client and its competitors.
- Write 20 questions your buyers ask before they buy.
- Run each one in every engine your buyers use, for you and your top competitors.
- Run each prompt more than once, since answers can change from run to run.
- Note which sources each engine cites, which brands it recommends, and whether you appear.
- Rerun the same prompts every month to see what changed.
Meltwater’s August report also recommends repeated measurement, since results shift with prompt wording, timing, and model version.
2. Have Your Experts Answer Buyer Questions on LinkedIn
In the May study, individual profiles earned about 75% of LinkedIn citations. In its follower breakdown, 51% of citations came from members with under 10,000 followers.
The type of post mattered too. Among the 24 most-cited LinkedIn URLs, 54% were “best X” lists and 50% were side-by-side comparisons. Only 8% were data-backed thought leadership.
That’s a small sample of 24 URLs, and the shares add up to more than 100%. Some posts fit several types.
- Pick two or three internal experts.
- Have them publish comparisons, “how to choose” guides, and ranked lists that answer real buyer questions.
- Publish on a steady schedule. In the study, 48% of cited content was under three months old.
Meltwater recommends two to three posts per week and three to four articles per month for each expert. Its May study didn’t track Perplexity, so results there may differ.
3. Turn Each Earned Story Into Several Citable Pages
Press releases earned just 0.2% of citation domain count in August. Earned/News coverage earned about 2.1 million citation domain counts.
So get more from each story. Meltwater suggests carrying strong coverage into an executive LinkedIn post, a YouTube explainer, a research page, or a customer-facing FAQ.
Then judge PR by citations. Add your coverage URLs to your monthly benchmark, and check whether AI engines cite them.
4. Add YouTube and Reddit Where Your Engines Cite Them
If your buyers use Google’s AI tools, YouTube deserves attention. Among the most-cited domains, it led on Gemini, Google AI Mode, and Google AI Overviews in August.
Meltwater recommends clear titles, detailed descriptions, accurate transcripts, and chapters for each video.
Reddit needs a lighter touch. Answer questions in relevant communities from real accounts, without pitching. ChatGPT is less clear. Meltwater’s tracked sample recorded zero ChatGPT citations to Reddit from May through August.
The same report points to a separate dataset where ChatGPT did cite Reddit. Test it in your own benchmark.
Review all four moves against your benchmark each month. Then shift budget toward what earns citations in your category.
How Reliable Is AI Citation Data?
AI citation data is reliable enough to show direction. Exact numbers need care, for three reasons below:
- Results depend on the setup: Meltwater notes that different prompts, timing, and methods can produce different baselines. GPT-5 cited LinkedIn in Meltwater and LinkedIn’s May B2B study. ChatGPT recorded zero LinkedIn citations in Meltwater’s August consumer tracking.
- Look for patterns across months: Meltwater recommends monthly reviews to separate a temporary spike from a sustained movement. Earned/News share has fallen every month since April. Facebook rose in July and fell in August.
- Citations aren’t traffic or revenue: Meltwater notes that its platform-level patterns describe citation behavior and don’t measure answer quality. A citation also can’t show whether anyone clicked or bought. Track referral traffic and pipeline alongside citations.
Find Out Which Sources AI Engines Cite in Your Category
Meltwater’s data shows the market average for AI search citations, but your category has its own mix.
Start by benchmarking your own category. If you’d rather not build that yourself, our SEO and AI search team can run it for you.
Our free AI visibility audit covers ChatGPT, Perplexity, Gemini, and Google AI Overviews for your brand and your competitors. You’ll see where you’re cited, where competitors show up, and where you’re missing.
If a competitor’s experts keep showing up in AI answers and your press coverage doesn’t, this is where you’ll see it.
Get your free AI visibility audit. Free, no credit card, delivered in 24 hours.
Rather talk it through first? Book a strategy call.
AI Search Citations FAQs
What Is the Difference Between an AI Citation and a Mention?
An AI citation links to a source, while a mention names a brand in the answer text. A brand can be mentioned without being cited, so it’s best to track both.
What Sources Does ChatGPT Cite Most?
ChatGPT favors institutional and data sources in Meltwater’s tracking. In August 2026, NIH led ChatGPT’s citations among the most-cited domains, with 10,638. In May, its top sources were Wikipedia and NIH. ChatGPT recorded zero August citations for YouTube, Reddit, LinkedIn, Facebook, and Instagram.
Do AI Engines Cite Review Sites Like G2?
AI engines cite review sites in B2B categories, though less often than LinkedIn, Reddit, and YouTube combined. In Meltwater and LinkedIn’s May 2026 B2B study, peer review sites such as G2 and Capterra made up 15% of AI citations.
Do AI Engines Cite Company Websites?
AI engines do cite company websites. In Meltwater and LinkedIn’s May 2026 B2B study, company websites across 69 vendors made up 18.7% of AI citations.
Sources
- Nathaniel Pallen. “AI Search Visibility Report for August 2026: LinkedIn enters the top three.” Meltwater, September 11, 2026. Accessed October 12, 2026.
- Nathaniel Pallen. “AI Search Visibility Report for July 2026: The latest developments in generative search.” Meltwater, August 14, 2026. Accessed October 12, 2026.
- Nathaniel Pallen. “AI Search Visibility Report – June 2026: How Generative Search Changed This Month.” Meltwater, July 3, 2026. Accessed October 12, 2026.
- Meltwater and LinkedIn. “How LinkedIn Content Wins in AI Search.” Meltwater, May 2026. Accessed October 12, 2026.
- Meltwater. “LinkedIn is the #2 most-cited source in AI answers, new Meltwater report finds.” Meltwater, May 12, 2026. Accessed October 12, 2026.
- Meltwater. “AI Visibility Overview.” Meltwater Developer Documentation, n.d. Accessed October 12, 2026.