Chatbots are just one step in the journey. Consumers gather information, then act. RealityMine^® tracked post-LLM app transitions across ChatGPT, Claude, and Gemini to see where users go next. Shopping and personal finance destinations are growing fast, while traditional search is losing post-LLM share. Here's what the data shows, and what it means for brands.

Every brand with a digital presence is wrestling with the same question: does LLM usage drive real-world commercial outcomes? And if so, how do you track it? The answer matters for how brands think about content, attribution, and where in the purchase funnel LLMs sit. This kind of insight is exactly what app competitive intelligence can reveal.

We used RealityMine behavioral data to track where consumers go next after an LLM session – and how these destinations may be evolving.

Where do people go after a chatbot session?  

We looked at next app visited after chatbot sessions across app and web — focusing on ChatGPT, Claude, and Google Gemini — and compared results for two time periods: summer 2025 vs. spring 2026 (referred to as “early” and “late” periods here).

At an overall level, the types of apps gaining traction as post-LLM destinations could be classed as action-oriented rather than passive scrolling sites: from the early to late periods, Shopping apps (e.g., Amazon, Walmart, Temu) rose from 15.7% to 17.2% of transitions, Personal Finance (e.g., Venmo, Capital One, TurboTax) from 13.1% to 16.7%, Food & Drink (e.g. DoorDash, Instacart) from 5.1% to 6.7%, and Education (e.g., ClassDojo, Duolingo, Coursera) from 3.6% to 6.5%.

What does this signal suggest? More and more, people are leaving LLM sessions and taking action. Social Networking and Photography (e.g. Google Photos, Vimeo) categories – apps more associated with passive scrolling - remain among the most common post-LLM destinations, but their share is declining. Instead, post-LLM behaviour is becoming more diverse and increasingly action-oriented. Perhaps LLMs are evolving from tools for information gathering into tools that actively shape consumer intent and decision-making.

Shopping effect

Amazon is the standout. Its share of post-LLM traffic grew roughly 33% across the period. This becomes more interesting when you investigate data from Amazon itself: Sessions that transition from ChatGPT directly to Amazon have the highest purchase rate of any traffic source we measured (around 11%), beating out both search and social.

These ChatGPT to Amazon sessions are also the longest, running about 50% longer than Google or Facebook-transitioned sessions, with more product page views and more searches per visit. What’s more – this is consistent across early and late time periods and is consistent with previous analysis we’ve run on special events such as Black Friday.

While this doesn’t take into consideration what happens inside the LLM session, the behavior on Amazon afterwards looks different from other transition sources: more purposeful, more focused. This depth of engagement is consistent with stronger consumer intent — someone looking to complete a goal rather than explore and discover.

This pattern suggests the LLM session may be doing a specific kind of work: narrowing the consideration set before the shopper arrives at retail. They still browse on Amazon, but they're operating within a tighter frame. The result is more purposeful searching/comparing ultimately a higher likelihood of buying.

While purchase rate is consistently the highest of any source, average order value is more mixed. In the current analysis periods, ChatGPT-referred sessions have a lower AOV ($23) than Google ($27), Pinterest ($30) or Snapchat ($27). Previous analysis has found this varies. The picture developing is that LLM-influenced shoppers are more likely to buy, but not necessarily on higher-priced items

What about traditional search?  

Zooming in on the Google eco-system, post-LLM traffic is a mixed picture. Gmail, Google Maps and Google Docs are up as next destinations; Search, Play Store, and Chrome are down. This tracks with our previous findings: the apps that involve completing a task or action are gaining share after chatbots, but traditional starting-point search-based apps are losing post-LLM momentum.

The upshot: Among people who use LLMs, search is less likely to be the next stop: the top-of-the-funnel function that search engines provide may already have happened inside the chatbot.

What this means for app competitive intelligence

LLM usage is growing fast, but how best to determine its downstream effects on other platforms is still an unanswered question. Brands that win in this space will be the ones paying attention to the whole picture on online behavior and tracking subtle shifts over time. This means:

1. Understanding where LLMs sit in your funnel: The post-LLM behavior we're seeing on Amazon (higher purchase rates, longer sessions, more product views) suggests LLMs are helping focus intent before purchasing. Knowing whether that's true in your category requires tracking the full journey and these transitions.
2. Scrutinize share shifts, not just volume: Social Networking and Photography are still high-volume post-LLM destinations, but their share is eroding while other categories like Shopping and Personal Finance grow. Looking only at volume metrics means you may miss these important signals. Habits change slowly, not overnight, but when they do, some categories will move much faster than others. Early detection is key.
3. Benchmark your own metrics: Amazon sessions referred from ChatGPT convert at around 11% — but that figure is only meaningful if you know how it compares to other sources, and how it's trending.

The brands best positioned to act on LLM-driven behavior change are the ones with the app competitive intelligence to see it in the first place.