LLMs and the future of mobile advertising

Successful mobile app user acquisition has traditionally relied on a balanced combination of paid and organic channels, with effective strategies integrating both to drive mutually reinforcing growth. The organic side focuses on ASO-powered app store search and keyword and category rankings, and to a lesser extent SEO rankings on web and mobile web, while paid advertising enables marketers to reach vast audiences across channels, platforms, and devices. As users turn to generative artificial intelligence (AI) for direct, personalized answers more and more, large language models (LLMs) like ChatGPT, Google’s Gemini, and Anthropic’s Claude, and more are now key app discovery channels. According to Gartner, traditional search engine usage is expected to decline  25% by 2026, and organic traffic is anticipated to drop up to 50% by 2028.

The appeal of LLMs is their ability to minimize the need to browse search results or click through multiple sites. A user can ask, for example, “What’s the best fitness app for beginners?” and get a suggestion instantly. These tools act as answer engines, analyzing intent and generating real-time responses that link directly to apps, websites, or services.

For mobile marketers, this shift represents a challenge to long-standing practices, but introduces a new opportunity to get in front of users—in terms of both organic and paid. ChatGPT, for example, is already driving measurable referral traffic to websites and app store landing pages, and platforms like Perplexity and Bing chatbot are testing native ad formats within AI responses. Growth teams must now rethink visibility, creative strategy, and performance measurement in an AI-first discovery environment.

When ChatGPT launched in late 2022, it reached 100 million users in two months, becoming the fastest-growing consumer app in history. Microsoft also integrated ChatGPT into Bing Chat in early 2023. Google followed with Gemini, embedding it into search through the Search Generative Experience (SGE). Other platforms quickly entered the space, each offering conversational, citation-rich alternatives to traditional search.

But the move toward AI-driven answers didn’t happen overnight. Search engines had already been moving toward more direct results. Google’s featured snippets, Knowledge Graph panels, and voice assistants trained users to expect instant answers. In 2024, nearly 60% of U.S. searches resulted in no clicks to external sites, a trend known as zero-click search. On mobile, voice search further amplified this,  with 1 in 5 searches in the Google App now being made by voice and typically returning a single spoken response. LLMs now deliver an enhanced version of this experience in text, and more.

According to a recent survey, over 80% of search users rely on AI-generated summaries for at least 40% of their queries. Among generative AI users, 68% use it for research, 48% for real-time updates, and 42% for shopping recommendations and decision support. Sensor Tower reported that nearly one-third (31%) of search engine app users also used ChatGPT in April 2025, up from 13% the previous year. These numbers demonstrate that users are defaulting to LLMs to search and make decisions more and more.

Firstly, LLMs are a super fast-growing organic discovery channel for mobile apps. Users ask questions or provide prompts like “top budgeting apps for couples” or “best fitness trackers under $100,” and are provided with direct recommendations, including app names and links.

LLM discovery prioritizes content that best fits the user’s question and appears the most useful or credible at the moment of inquiry. Unlike traditional search or app store listings, LLMs typically return only a few results. Your app is either included or it isn’t. With such limited and highly competitive visibility, each recommendation slot is increasingly valuable and likely to result in high-value users.

When your app is recommended, the resulting traffic can be tracked. Some platforms, such as ChatGPT, can now append UTM tracking parameters to outbound links, enabling marketers to track traffic and installs from these sources. Many brands are already seeing measurable referral traffic from AI platforms in their analytics. This has also created the emerging generative engine optimization (GEO) concept, where teams are working to strategically improve the likelihood of their sites and apps appearing in LLMs responses.

To stay competitive on the organic side, mobile marketers should focus on:

• Creating content that aligns with natural-language queries
• Understanding how LLMs choose and structure recommendations
• Building attribution models that capture AI-driven engagement.

As LLM discovery becomes a standard channel, marketers who actively optimize for it will be better positioned competitively.

Several LLM platforms are attracting the interest of advertisers and some are already testing or launching native ad formats. Others are scaling fast and are likely to introduce options for advertisers soon. Here’s where things stand:

ChatGPT (OpenAI)

In early 2025, ChatGPT surpassed 800 million weekly active users. Currently ad-free, OpenAI has publicly confirmed ongoing internal tests of conversational native advertising formats, expected to launch within the next year. Given ChatGPT’s scale and direct attribution potential, these developments should be monitored closely.

Gemini (Google)

Gemini, deeply integrated within Google's ecosystem, powers prominent products like Bard and the AI-driven SGE. Gemini already supports native ad placements within SGE, and Google is actively expanding these ad formats. With approximately 400 million monthly users, Gemini represents a scalable advertising opportunity, especially for mobile marketers already familiar with Google’s ad ecosystem.

Claude (Anthropic)

Anthropic’s Claude, known for its emphasis on safety and context awareness, currently attracts approximately 16 million monthly visits. Although Claude has not yet introduced ads, recent strategic investments from Google indicate monetization plans are likely in development. Claude’s technology is also utilized in third-party tools, such as Poe and DuckAssist, suggesting potential for a broader reach and future monetization pathways.

Ernie Bot (Baidu)

Baidu’s Ernie Bot engages approximately 78.6 million monthly active users. Already embedded into Baidu’s core search and advertising ecosystem, Ernie Bot is actively experimenting with conversational advertising enhancements. Mobile marketers operating within the Chinese market should experiment with Ernie Bot’s features, as it offers a mature environment for testing and optimizing ads based on user input.

DeepSeek (DeepSeek AI)

DeepSeek, an open-weight LLM developed in China, has gained traction for its strong multilingual and code-generation capabilities. As of April 2025, DeepSeek reported 96.88 million monthly active users worldwide. While it does not yet support advertising, its growing user base and enterprise traction make monetization features, such as sponsored responses, possible.

Perplexity

Perplexity, currently handling over 100 million monthly user queries, is also actively testing conversational ad formats, including sponsored follow-up prompts and contextually triggered recommendations. Its user base, characterized by research-oriented queries and high intent, offers marketers precise targeting opportunities for specific app categories, making it especially effective for promoting specialized or category-specific mobile apps.

Meta AI (LLaMA)

Meta's AI platform, LLaMA, is already embedded into WhatsApp, Instagram, and Messenger, becoming a core component of Meta users’ daily mobile interactions. While LLaMA does not currently support advertising, industry reports indicate that Meta is actively exploring monetization strategies, including the potential integration of ads into its AI-driven interfaces.

Conversational AI platforms are beginning to integrate ad placements that are limited in volume but highly contextually relevant. Some formats are already live, while others remain in testing or early rollout phases, depending on the platform. Based on what’s already live and what’s currently in testing, advertisers can expect several core formats:

• Native text suggestions: These are ad prompts that appear right after the AI’s response, formatted to resemble organic follow-up questions within the chat interface. Perplexity, for example, is currently in an experimental rollout phase with sponsored follow-up questions and related paid queries. These ad units often appear in the “People also ask” section, where the first follow-up question may be sponsored. The ads are clearly labeled, and the responses are generated by Perplexity’s AI, not written or edited by advertisers. This approach maintains the platform’s tone and structure while allowing brands to appear contextually relevant without disrupting the user experience.
• Sponsored links: These are visually distinct ad units that appear below the AI’s response in the chat interface. In Snapchat’s My AI, for example, sponsored links are triggered based on the user’s query and clearly labeled as “sponsored results.” While not part of the AI’s response itself, these ads are embedded within the conversation flow and designed to feel timely and valuable.
• Interactive product showcases: These are rich, visual cards that often feature product images, brief descriptions, and tap-to-explore actions, appearing in response to a user’s query. Amazon’s Rufus, for example, presents these cards directly below the AI’s answer, showing specific product categories. While not all results are sponsored, the format supports in-flow product discovery, is designed for mobile use, and is well-suited for eventual monetization.

As LLM-based ads continue to develop, organic inclusion in AI-generated responses is already shaping how users discover mobile apps. Here are key strategies mobile growth teams can use to increase visibility today:

Refine user-facing content for LLM prompts

To improve discoverability, app content should reflect how users naturally phrase queries and align with the tone of AI responses. Messaging should clearly convey value using structured, conversational language that LLMs can paraphrase easily and accurately.

Because AI tools often mix your app’s content with external sources, like reviews or editorial summaries, it’s essential that core benefits are clear, consistent, and easy to extract. Teams can test queries in tools like ChatGPT or Gemini to monitor how their app is described, and benchmark competitors to identify gaps or opportunities.

Improve backend inputs LLMs ingest

LLMs personalize responses based on prompt context. The same app might be presented differently depending on a user’s intent or experience level. While marketers can’t control these responses directly, they can shape the inputs that LLMs rely on.

Structured assets, like app store listings, metadata, feature tables, and FAQs, help LLMs extract factual, reusable product information. High-quality formatting makes it easier for models to position your app reliably across a range of prompts.

Strengthen app reputation across sources

LLMs often draw on public sentiment, including app store reviews and third-party testimonials. Strong ratings, helpful feedback, and credible mentions can influence whether your app is included in recommendations, and how it’s framed. Marketers should prioritize reputation management as part of LLM optimization, ensuring that external sources reinforce the same product strengths highlighted in your own content.

Track AI-generated traffic

As mentioned earlier, platforms like ChatGPT are already appending referral identifiers, like utm_source=chatgpt, to outbound links. Marketers should configure analytics tools to segment and measure this traffic separately from other sources.

To capture non-click-based influence, consider including “How did you hear about us?” survey questions post-install, with AI platforms listed as options. This can help identify prompts that drive decisions, even if the AI didn’t generate a direct referral link.

Update content based on model changes

LLMs are constantly evolving. As models are retrained or updated, the way they frame your app may shift. Treat prompt testing as an ongoing input to content strategy. Refresh queries regularly and adjust descriptions, benefits, and structured fields based on how your app appears over time.

Align with broader discovery efforts

LLM visibility should work in tandem with existing discovery strategies. Coordinate your efforts across SEO, ASO, and CRM to ensure that messaging is consistent, whether it’s being pulled into an AI answer, shown in an app store, or used in lifecycle marketing. Cross-functional alignment strengthens credibility and reinforces user trust across channels.

LLMs are quickly becoming a core interface for how users discover and assess mobile apps. As native ad formats begin rolling out on platforms like ChatGPT, Gemini, and Perplexity, this is creating new opportunities for high-intent engagement.

To stay ahead, mobile marketers should focus on how their products are interpreted in AI-driven environments through content that’s structured, helpful, and easy for systems to understand. Advertisers who act early—adapting creative, refining targeting strategies, and preparing for low-volume, high-impact placements—will be better positioned to stand out as these channels mature.

Request a demo and see how Adjust can help you measure, analyze, and optimize user acquisition across both new channels and your existing marketing mix.