LLM Product Discovery Benchmarks
Introduction
For the past year, we've been tracking arguably the most interesting trend in our attribution data: consumers increasingly discovering products through large language models (LLMs) like ChatGPT. This isn't surprising if you've been following the space—LLMs are becoming the new search engine for many users. But the pace at which this is happening is worth paying attention to.
We've analyzed data from more than 2,500 digital brands using Fairing's attribution surveys. Here's what we're seeing.
Key Findings
ChatGPT dominates the LLM discovery landscape, accounting for 64.6% of all LLM-attributed transactions
11.8% of Fairing customers now report at least one customer mentioning an LLM in their attribution surveys, up from less than 1% this time last year
Grok is second most popular LLM used by consumers, after ChatGPT
The growth trend is consistent and accelerating, with April 2025 showing the highest rates to date
Most brands see between 0-1% of total responses mentioning LLMs, with a long tail distribution.
This chart shows the weekly growth in "How Did You Hear About Us" responses mentioning LLM platforms, normalized to the April 27th total. The trend is undeniable—LLM attribution is up by over 5.4x from January 2025 to the end of April 2025.
ChatGPT has maintained its dominant position throughout, but competitors like Grok, and Gemini are starting to gain market share. The collective impact of these platforms is creating a new acquisition channel that marketers need to understand and optimize for.
Not all product categories are experiencing LLM attribution equally. Automotive stands out with 100% of LLM mentions going to ChatGPT – although this sample size may not be as large as other categories. Consumer Electronics (77.7%), Food & Drug (73.4%), and Office (72.7%) categories also show strong ChatGPT dominance.
What's interesting is where we see competitors making inroads. In Specialty categories, Claude has a 15% share. Grok is performing relatively well across categories like Sporting Goods (21.3%) and Specialty (19.4%). This category breakdown gives marketers a clearer picture of which LLMs to monitor based on their industry.
A small portion of Fairing customers have an AI specific option on their attribution survey, this is codified as the AI Tool (Misc) category.
Looking at the overall share of transactions attributed to LLMs, ChatGPT takes the lion's share at 64.6%. Grok follows at 12.6%, then "Unspecified" LLM mentions at 11.9%, Perplexity at 6.1%, Gemini at 3.1%, and Claude at 1.8%.
As mentioned, the "Unspecified" or "AI Tool" category is worth noting—these are responses where customers mention using an AI or LLM, but don't specify which one. This is most likely due to survey design options versus consumer behavior.
This is where things get really interesting for marketers. The percentage of Fairing customers who've had at least one customer mention an LLM has grown from near-zero in May 2024 to 11.8% in April 2025.
The growth hasn't been linear. We see significant acceleration starting in November 2024 (holiday), a brief plateau in January-February 2025, followed by another acceleration in March-April 2025.
For context, this means that roughly 1 in 8 brands using Fairing now see LLMs appearing in their attribution data. If you're not looking for these signals in your customer acquisition data, you're missing a growing channel.
When we look at how LLM mentions are distributed across our customer base, we see a typical power law distribution. The vast majority of brands see between 0-0.5% of their total responses mentioning LLMs, with a quick dropoff after that.
However, the tail is what makes this interesting. Some brands are seeing up to 6% of their responses mentioning LLMs as a discovery channel. These outliers are worth studying—what are they doing differently that's driving higher LLM discovery rates?
Finally, we looked at the average percentage of responses mentioning an LLM platform per Fairing customer over time. After normalizing for growth in our customer base, we see that the rate has remained relatively consistent between 0.2-0.3% throughout the year, with a slight upward trend in Q1 2025.
The stability of this metric suggests that while more brands are seeing LLM mentions (as shown in the previous chart), the intensity of LLM attribution remains fairly consistent on a per-brand basis. That said, a portion of brands are seeing an incredible amount of conversions via LLMs, way above the mean.
What This Means For Marketers
If you're not thinking about LLMs as a discovery channel yet, now's the time to start. This isn't just another social platform—it's a fundamentally different way consumers are finding products. A few thoughts on what to do with this data:
Start measuring: If you're not already using attribution surveys like Fairing's, you need to start capturing how customers are finding you. Traditional click-based analytics are missing the majority of these discovery paths.
Monitor platform-specific trends: Different LLMs have different strengths in different categories. Know which ones matter most for your industry.
Optimize for LLM discovery: This is early days, but forward-thinking brands are already figuring out how to make their products more discoverable through LLMs. This includes structured product data, clear value propositions, and building brand recognition that transcends specific platforms.
Watch for rapid changes: The growth curve suggests we're still in the early adoption phase. What's 0.3% today could be 5%, maybe as much as 10%, in a year.
Given the rapidly changing nature of this performance, we'll be tracking these metrics monthly to provide you with the most current insights on LLM product discovery. If you're seeing interesting patterns in your own data, we'd love to hear about it.
Want to start measuring how customers are discovering your products, including through LLMs? Get started with Fairing today.
