Last Updated: December 1, 2025
- AI search and chat tools still send most of their referral traffic from desktop, but mobile is quietly catching up and you cannot ignore it for 2026 planning.
- Google Search on mobile, now blended with AI Overviews, behaves very differently from pure chatbot traffic and still drives huge volumes of clicks.
- Your brand is more likely to earn AI citations when your content is clear, structured, fast on mobile, and backed by strong author and entity signals.
- Marketers who track AI referrers, separate mobile vs desktop behavior, and shape content for AI summaries will be in a much safer spot next year.
If you want a blunt answer to where AI search referrals really come from right now, here it is: desktop still dominates for pure AI tools, while Google stays heavily mobile, especially where Safari and Chrome rule.
The twist is that the gap is shrinking, mobile AI UX is far better than a year ago, and AI Overviews sit in the middle of both worlds, quietly changing how people click results.
AI Search Referral Traffic Has A Desktop Bias, But Mobile Is Catching Up
Most AI-native tools still send the bulk of their clicks from desktop, though not always at the extreme 90-plus percent levels you might have seen in older studies.
Fresh datasets from enterprise platforms like BrightEdge and seoClarity through mid‑2025 show the same broad pattern: people ask AI questions everywhere, but they click through far more often when sitting at a laptop or monitor.
| Platform / Experience | Desktop share of AI-driven sessions | Mobile share of AI-driven sessions | Notes (mid‑2025 directional ranges) |
|---|---|---|---|
| ChatGPT (web + apps) | 75-85% | 15-25% | Desktop still stronger for research-heavy tasks |
| Perplexity | 70-80% | 20-30% | Mobile rising fast where the app is popular |
| Microsoft Copilot (Bing, Edge, app) | 65-75% | 25-35% | Blended into Bing SERPs and Edge UI |
| Google Gemini app / Gemini web | 60-70% | 30-40% | Heavily cross-linked with Google Search |
| Google Search AI Overviews | 40-50% | 50-60% | Still more mobile than desktop, like classic search |
Desktop still carries most AI-driven referral sessions for pure chat tools, while Google Search with AI Overviews remains the big mobile traffic engine.
These ranges are directional, not hard absolutes, and they skew toward larger US and European sites that share anonymized data with SEO platforms.
If your audience is younger, more global, or skewed to Android, you may see a higher mobile share than the table suggests, which is why you must cross-check this against your own analytics before you bet strategy on it.
How Different AI Platforms Actually Send Traffic Right Now
The easy mistake is to throw all AI tools into one bucket and treat them like they behave the same, but they do not.
Each has its own way of showing links, opening pages, and reporting referrers, especially on mobile, and those details drive your traffic more than big headlines about AI ever will.
ChatGPT: From Answer Box To Actual Click
On desktop, ChatGPT with browsing enabled typically shows a narrative answer with a column or row of cited sources underneath, and each citation opens in your normal browser tab with a clean referrer header.
On mobile, both the iOS and Android apps have improved a lot: links now open in an in‑app WebView first, with an option to open in your default browser, which is convenient for users but makes attribution messy.
- One tap to expand the answer and see sources.
- One more tap to open the cited page in the in‑app webview.
- Option to send that page to Safari or Chrome, sometimes dropping referrer data.
So your analytics might classify some of this traffic as direct or app-webview instead of clearly labeling it as chat.openai.com.
This is why some brands think ChatGPT sends almost no traffic, when in reality the clicks are just hiding under ugly referrer names or none at all.
Perplexity: Strong Citations, Research-Heavy Use
Perplexity is probably the closest thing to an AI search engine that really wants you to click out, not just stay in its interface.
On both desktop and mobile you see a stacked layout of sources under each answer, with focus modes like Web, Academic, YouTube, and others that shift which domains get exposure.
- Citations show favicon, title, and snippet.
- Clicking a source on mobile opens in a webview that preserves the Perplexity referrer more reliably than most.
- Scrolling deeper often reveals more long tail sources beyond the usual big sites.
Perplexity is one of the friendliest AI tools for site owners, because its design pushes users toward reading the sources, not just the AI summary.
Still, the heaviest Perplexity usage tends to be on desktop for knowledge work, developer research, and academic tasks, so the traffic skew remains desktop-heavy for now.
Microsoft Copilot: Three Different Entry Points
Copilot referrals are harder to read at a glance, because you are actually dealing with at least three main experiences that behave differently.
- Copilot inside Bing SERPs.
- Copilot inside the Edge sidebar.
- Copilot as a standalone mobile app.
In Bing SERPs, AI answers usually sit above or alongside classic organic results, with inline links that go to the same pages you might see in normal blue links, so traffic often looks like regular bing.com referrals in GA4.
In the Edge sidebar, users can get quick answers and then open cited pages into the main tab or a new tab, which looks exactly like a normal browser visit in your logs.
The Copilot mobile app, which some people ignore, opens results in an in‑app browser first, and only then hands visitors to Chrome or Edge, again muddying the referrer for you.
Google Gemini: Split Between App And Search
Gemini has two lives: the standalone Gemini app, and the pieces blended into Google Search as AI Overviews.
In the app, users can chat, see cards of links, and tap into the browser, and that often shows up as traffic from Gemini.google.com or related endpoints.
In classic Google Search, AI Overviews answer many informational queries with an AI block that contains multiple cards linking out to sources, and these clicks normally show up just like normal Google organic traffic.
You cannot really talk about Google Search anymore without including AI Overviews in the picture, especially on mobile where they sit in prime real estate.
This is one reason brands sometimes underestimate AI influence: the traffic is still labeled as Google organic, but the user saw an AI paragraph first, not your meta description.
Desktop Vs Mobile: The Real UX Differences That Drive Clicks
The older theory was that people only used AI chat seriously on desktop and treated mobile as quick snack time, but that is only half true now.
The real difference today comes from how many steps it takes to reach your site and how clean the experience feels on a small screen.
How Many Taps To Reach Your Page?
If you sit down and count taps, you get a more honest story than any opinion piece.
Here is a rough comparison of a typical journey from AI answer to your page.
| Platform | Desktop path to click | Mobile path to click | Friction notes |
|---|---|---|---|
| ChatGPT | See answer → Click citation → Land on site | See answer → Expand sources → Tap link → In‑app view → Optional open in browser | Extra step and double loading on mobile |
| Perplexity | Sources visible → Click → Land on site | Sources visible → Tap → In‑app view (usually OK) | Pretty low friction on both |
| Copilot (Bing SERP) | AI answer + links → Click → Land on site | AI answer + links → Tap → Land on site | Behaves similar to normal search |
| Google AI Overviews | Overview → Card click → Land on site | Overview → Card tap → Land on site | Standard web flow; strong mobile experience |
Every extra step drops clicks a bit, especially on mobile where people are more impatient and less willing to wait for multiple loads inside app shells.
This is why desktop still wins for many AI tools, even though smartphones are everywhere and people search from them constantly.
In‑App Browsers And The Analytics Headache
There is another quiet problem: in‑app webviews do not always pass helpful referrer data to your analytics platform.
When someone clicks your link from ChatGPT or the Gemini app and stays inside that app’s browser, you might see:
- Referrer as a generic app domain or none at all.
- Source classified as direct or unassigned in GA4.
- Sessions that look like mobile organic or direct but are really AI-driven.
A slice of AI traffic on mobile is already here, but it arrives wearing a “direct” mask in your analytics reports.
This is why you should not assume AI sends no mobile traffic simply because ChatGPT is not listed as a clear source.
You need to look at patterns by landing page, by device, and by time window after major AI product launches to get closer to the truth.
Why Google Still Owns Mobile Search Behavior
Google did not lose its mobile habit just because AI got trendy.
On iOS, Safari still defaults to Google in many regions, and on Android the Google app and Chrome search box are the front door for billions of daily searches.
- People are used to typing or talking to the same bar for years.
- Google’s mobile pages load fast and feel predictable.
- AI Overviews are baked into an experience users already trust.
So even when AI handles the answer, the click data shows up as classic Google organic, which keeps Google’s mobile share very strong on paper.
This is why your mobile strategy cannot ignore Google, even while you experiment with AI-native tools like Perplexity or ChatGPT.
Apple, Safari, And The New Search Power Structure
A few years ago, people kept guessing about what Apple might do with its default search deal, but by now the picture is much clearer.
The key takeaway is that Apple leaned hard into on-device intelligence and subtle answer layers, while keeping a close commercial relationship with Google in most markets.
Where Apple Stands Right Now
In major Western markets, Safari still sends a large chunk of its web queries to Google, and they keep working together on commercial search despite regulatory pressure.
At the same time, Apple has pushed more answers into Spotlight, Siri, and new on-device features that look a lot like AI search for basic questions.
- Spotlight and Siri can answer factual queries without always bouncing to a browser.
- Safari suggestions often show quick answers or rich snippets above full search results.
- Apple Intelligence features blend local context and online facts where possible.
This mix means part of your potential mobile search audience never hits a standard SERP at all, which dilutes click volume for basic questions.
But for anything transactional, complex, or commercial, users still end up on a classic search results page in most cases, which keeps referrals flowing.
Regulation, Defaults, And What Actually Changed
Regulators in the US, EU, and UK have pushed hard on exclusive default search deals, especially the Google-Apple agreement.
The practical outcome so far looks more like added choice screens and subtle nudges than a total breakup.
- Some users now see prompts to pick a search provider when setting up or updating devices.
- Alternative search engines got a small boost, but not a tidal wave.
- AI players gained more attention as possible partners, without fully replacing Google.
This might feel slower than many experts predicted, but for marketers it means you still live in a world where Google, on Apple devices, is central to your mobile search traffic.
That may change over time, but today you plan around reality, not wishful thinking.
What Apple’s On-Device Answers Mean For Your Site
Apple’s on-device features rarely show explicit web citations the way Perplexity or AI Overviews do.
So they reduce some shallow search traffic without giving you a clear new referrer to study, and that can feel frustrating.
When on-device assistants answer basic questions, you lose some cheap, top-of-funnel clicks, but visitors who still come through are more intent-driven.
This is not good or bad by default, it just changes the mix: fewer visits from people asking “what time is it in London” and more from people asking “best email marketing software for mid-size SaaS.”
Your content should reflect that shift by going deeper on decision support queries and being less obsessed with micro trivia that assistants now handle on their own.
AI Overviews And How They Reshape Mobile CTR
Google’s AI Overviews sit on top of many informational queries now, especially in English, and they matter more on mobile where the screen is small and attention is tighter.
For many users, the AI paragraph is the search result, and links are just backup.
What Early Studies Show About CTR
Studies from SEO tools and clickstream providers through 2025 show some common patterns, even though the exact numbers vary by sample.
- For simple informational queries, total organic clicks per search often drop, because users read the overview and stop.
- For complex or commercial-intent queries, click behavior is more mixed, with many users still scrolling down to check multiple sources.
- On mobile, fewer visible blue links above the fold mean the top cited cards in AI Overviews get a stronger share of attention.
So yes, AI Overviews can cannibalize part of organic traffic, especially for questions with clear, short answers, but they also introduce a new way to get exposure if your content is cited near the top.
And unlike many chatbot experiences, AI Overviews track roughly like normal Google organic, which makes them easier to measure even if you cannot split them out perfectly.
How Users Click Inside AI Overviews
On mobile, people scan the AI paragraph first, then either tap one of the top cards or flick straight past the whole block to the classic results.
This creates three main patterns you should care about.
- Your site appears in a card and wins a click almost immediately.
- Your site appears below the fold in classic results and still gets a share of clicks.
- Your site is missing from both, and you lose visibility on that query.
In my view, the worst case is not AI stealing the click.
The bigger risk is AI answering without mentioning you at all, while still drawing from the same knowledge your content covers.
How To Win More AI Referrals In 2026
If you only remember one thing from this article, make it this: you cannot control AI interfaces, but you can control how cleanly your content can be quoted, cited, and loaded on mobile.
That is where you win or lose traffic as AI takes a bigger slice of search behavior.
Step 1: Fix Measurement For AI Sources
Before you talk strategy, you need a clear picture of what you get today, even if it is messy.
In GA4, start by building better grouping for AI traffic.
- Create custom channel groupings or reports that catch hostnames like chat.openai.com, perplexity.ai, bing.com, copilot.microsoft.com, and Gemini.google.com.
- Look at mobile vs desktop splits for those sources and track the trend over several months, not just one snapshot.
- Watch landing pages that suddenly gain direct mobile traffic after AI feature launches; some of that is likely in‑app AI visits.
I know this will not give you a wildly precise view, but you only need it to be consistent and directional to guide content decisions.
Step 2: Shape Content For AI Citation, Not Just Classic SERPs
AI systems favor content that is clear, scannable, and easy to quote, which lines up with good SEO, but you do need to be more deliberate now.
Think about how your page looks to a model that needs a one or two sentence answer plus a short justification.
- Lead with a direct answer in the first paragraph, then expand in sections below.
- Use subheadings that line up with common questions, not just clever titles.
- Add FAQ, HowTo, Product, and Article schema where relevant so your structure is explicit.
- Keep author bios, source references, and brand information visible to support authority and trust signals.
AI tools are more likely to cite you when the answer is clear, self-contained, and sitting near the top of the page in simple language.
Long, rambling intros and buried answers do not just hurt human readers, they make it harder for AI to pull a clean quote with your brand attached.
Step 3: Design Pages For Mobile AI Visitors
Many AI-sourced users on mobile arrive inside an in‑app browser, which is slower and more fragile than a full browser, so your page needs to work even under that constraint.
If your site chokes inside those shells, you lose the click you just fought to win.
- Test key landing pages inside ChatGPT, Perplexity, and Gemini mobile apps and click through to see what happens.
- Reduce heavy interstitials, popups, and complex scripts that slow first paint or jump the layout.
- Put a short summary or answer box near the top, since many users will not scroll far on that first visit.
- Use clear in-page navigation and simple CTAs that are easy to tap with a thumb.
This is not just UX work, it directly affects your ability to convert the narrower slice of AI traffic that does click through.
Step 4: Watch Vertical-Specific Impacts
AI does not treat every topic the same way, and your strategy should respect that.
From what we see across client data and public studies, a few trends stand out.
- News and fresh content: AI often lags on ultra-fresh stories, so classic search and news surfaces still matter more here.
- Health and finance: AI tends to surface large, trusted entities first, tightening competition for smaller brands.
- Ecommerce: AI answers often list a handful of products or stores, but people still click through to compare details and prices.
- B2B and technical content: AI is strong at summarizing docs but often pushes users to original sources when details matter.
If you operate in a sensitive or regulated space, you should spend extra time checking how AI tools describe you and your competitors right now.
There is no guarantee they are accurate, and cleaning that up can have real revenue impact.
Step 5: Prepare For Different 2026 Scenarios
No one knows the exact mix of AI vs classic search traffic you will see next year, but you can at least prepare playbooks for a few simple cases.
Think about your plan in three buckets.
- Scenario A: AI referrals grow fast
Focus more resources on answer-first content, structured data, and brand presence in AI tools; prioritize tracking AI referrers and building relationships where possible. - Scenario B: AI cannibalizes clicks without sending many referrals
Lean harder into brand building, email, and community so you are less exposed to SERP volatility, and watch conversion rate on remaining organic traffic. - Scenario C: AI stabilizes as just another layer
Keep balancing traditional SEO with AI-focused formatting, and test small experiments across new AI search products without overcommitting.
The worst move is to treat AI as a passing trend or as an unstoppable black box; both mindsets keep you from testing, measuring, and adjusting with intent.
Your actual outcome will likely sit somewhere between these scenarios, but having responses mapped out will keep you calm when headlines spike again.
Global And Device Differences You Cannot Ignore
Everything so far leans toward US and European patterns, and that is fine if those are your main markets, but it is a bad shortcut if they are not.
AI usage and mobile behavior look very different once you step into other regions.
Android-Heavy Markets And Local Search Engines
In many countries, Android dominates far more than iOS, and the default Google experience on the home screen shapes how AI shows up.
Think of the long-press search features, Lens integrations, and Google Discover surfaces that now show AI summaries or quick answers before the web.
- Circle-to-search type gestures can answer questions over screenshots without a classic search.
- News and evergreen content can show up in Discover with AI summaries sprinkled in.
- Local search engines like Baidu, Yandex, or Naver have their own AI layers and referrer patterns.
If you have real revenue in these regions, you need localized tracking and content, not a copy-paste of your US strategy.
I would not rely on one global dashboard to tell that full story; local teams or partners can spot shifts faster.
Social Search And AI: TikTok, YouTube, And Others
There is another shift you should not ignore: younger users searching inside TikTok, YouTube, Instagram, and other apps, often with AI-assisted results.
These platforms are building their own recommendation and answer tools that can bypass web search entirely.
- Short videos that answer specific questions can rank in in-app search and feed for months.
- AI-assisted captions, summaries, and highlights help users skim without clicking out.
- Links in profiles and descriptions still drive traffic, but often in bursts instead of steady flow.
If your audience is under 30, a chunk of your “search” exposure may come from these platforms, and your website is just one hop in a longer attention path.
So when you think about desktop vs mobile AI referrals, be honest about this: for some people, mobile social is the default, and web search, AI or not, is the alternative.
Key Moves For Your 2026 Search Plan
AI search is not replacing classic search overnight, but it is changing who gets cited, how people click, and where those clicks show up in your reports.
Desktop still drives most referrals from pure AI tools, Google still owns a huge slice of mobile traffic with AI Overviews blended in, and Apple’s on-device answers quietly shave off shallow queries.
- Tighten your analytics to spot AI-originated traffic, especially messy mobile referrers.
- Rewrite key pages so the main answer is clear, concise, and near the top, with structure that models can parse easily.
- Improve mobile loading and in‑app browser behavior so AI clicks do not bounce away in frustration.
- Pay close attention to how your brand is cited inside AI Overviews, ChatGPT, Perplexity, and Copilot, not just how you rank in blue links.
You do not need to chase every AI launch or throw out your SEO playbook, but you also cannot keep pretending that all traffic behaves like 2019 Google Search on desktop.
If you keep your measurement honest, your content sharp, and your mobile experience fast and readable, you will be in a much stronger place, no matter how much AI search shifts in the next year.
Need a quick summary of this article? Choose your favorite AI tool below:
