All notes
    AEONovember 18, 20267 min read

    ChatGPT Isn't Your Buyer's Search Engine Yet. Here's Where AI Visibility Pays for Crypto and Fintech

    ChatGPT Isn't Your Buyer's Search Engine Yet. Here's Where AI Visibility Pays for Crypto and Fintech

    We'll run your category's buyer prompts across ChatGPT, Perplexity, Gemini, Claude and Google's AI surfaces and show you who owns the answer today.

    Get an AEO Growth Plan

    Share

    AI summary

    Short on time? Generate a TL;DR with key takeaways.

    Purchase research is a small slice of what people do in ChatGPT. That's the headline from SparkToro's new analysis of 118,163 ChatGPT prompts, published on 17 November 2026. For crypto and fintech brands it doesn't mean AI visibility is overhyped. It means most teams are pointing their effort at the wrong surface.

    Here's what the study found, and how we'd turn it into a plan.

    What did the SparkToro study measure?

    It measured what real US users actually type into ChatGPT, not what they say they do. Rand Fishkin's team analysed 118,163 prompts across 41,961 conversations from 12,321 people on Verso, an opt-in, anonymised panel of ChatGPT users, between 1 January and 29 October 2026.

    A few method choices make the numbers more trustworthy than most AI usage stats doing the rounds:

    • Figures are per user, so a handful of power users can't skew them. The team also removed 131 heavy users who made up about 1% of the panel but 21% of all prompts.

    • Every conversation was labelled from a written codebook into 12 categories. A second, independent labeller re-coded a sample and agreed 90% of the time on categories and 97 to 99% on brand tags.

    • The panel was weighted to 2024 US Census targets. No category moved by more than 1.4 percentage points.

    How do people actually use ChatGPT in 2026?

    Mostly as a personal assistant, not a search engine. Images and visual design was the largest category at 16.1% of conversations, followed by information and entertainment (14.0%), everyday life and admin (12.2%) and writing (11.9%).

    The categories closest to buying sit further down. Money, work and business came in at 8.5%. Shopping, brands and local services was 5.7%. Health and wellness, another high-trust category, took 10.3%.

    Fishkin's read is that the ChatGPT versus Google contest is overstated. People treat ChatGPT as a creative tool, a confidant and a helper far more than a place to research what to buy.

    How much ChatGPT use is purchase research?

    About 6.9% of prompts involved researching a brand to buy from. In SparkToro's earlier study of 332 million Google queries, more than 15% met the same definition.

    Fishkin then scales that by volume. He estimates Google handles roughly 5.5 times as many searches as ChatGPT gets prompts, so he puts purchase-related queries on Google at 16 times or more the rate on ChatGPT. That's his estimate, not a measured figure, but the direction is clear. He also puts the buyer-journey share of ChatGPT use at around 12%, which makes it closer in size to Bing than to Google.

    Does ChatGPT recommend brands nobody asked about?

    Yes, slightly more often than users ask for them. 12.1% of ChatGPT responses named at least one brand, product or company, and responses recommended brands more often than users explicitly requested them.

    That detail matters more than the headline. A brand mention in ChatGPT isn't triggered by a keyword the way a search ad is. The model decides when a shortlist is useful and who goes on it. If your category is one where it names brands, you're either in that shortlist or invisible, with no position 7 to fall back on.

    What does this mean for crypto and fintech brands?

    Google's AI surfaces deserve first claim on your budget, and ChatGPT deserves targeted, measured effort rather than panic. Here's how we'd apply the study.

    1. Lead with Google's AI Overviews and AI Mode. Fishkin's own conclusion is that AI visibility matters mainly because of Google's AI features. In regulated finance, comparison still happens on Google, and an AI Overview now sits above the results your SEO team fought for. If you're only tracking ChatGPT, you're watching the smaller screen.

    2. Target the trust questions, not only "best X" prompts. Most crypto pre-purchase work doesn't look like shopping. It looks like "is this exchange safe", "how do stablecoin reserves work" or "what happens to my funds under MiCA". In SparkToro's categories, those land in information, everyday admin and money, which together are far bigger than shopping. That's where an answer engine decides which brand sounds credible before any shortlist appears.

    3. Read the panel against your audience. Verso's panel skews younger, female and lower-income, and is mostly mobile. That's closer to retail crypto users than to a bank's treasury team. API traffic, desktop and work accounts are under-represented, so for enterprise fintech buyers this study likely understates ChatGPT's role in research.

    4. Measure engines side by side. Fishkin says ChatGPT tracking is only worth it if you track Bing too. We'd go further: run the same buyer prompts across ChatGPT, Perplexity, Gemini, Claude and Google's AI surfaces, and classify each category by who owns the answer. That's the approach in our methodology.

    Where should the effort go in 2027?

    We'd split AI visibility work for a crypto or fintech brand in roughly this order:

    1. Entity foundations. Consistent company names, licences, regulators and leadership across your site, registries, app stores and press. Every engine leans on this, Google included. Our SEO and AEO audit guide walks through the checks.

    2. Google organic plus AI Overviews and AI Mode. Answer-first pages, clear comparisons and FAQ content that Google can lift into its AI answers.

    3. Citable third-party sources. Earned coverage and data on sites AI engines already trust, because those sources feed every model at once.

    4. ChatGPT and Bing tracking for a short list of priority prompts. Not hundreds. The 20 to 40 prompts closest to revenue, checked monthly. Regulation can move these fast, as our MiCA recheck found.

    None of this is a separate "AI channel". It's the same discovery problem SEO always solved, with more places the answer can come from. If you're new to the term, start with What Is AEO?

    What are the study's limits?

    It's one opt-in US panel, so it shows what people ask better than how much they use ChatGPT. API usage is missing, which probably under-counts work and technical prompts. The second labeller flagged brand questions more often than the first, so the brand figures are likely conservative. And it's US only, so patterns in other crypto markets may differ.

    FAQ

    What share of ChatGPT prompts are about buying something? About 6.9% of prompts in SparkToro's 2026 study involved researching a brand to buy from, compared with more than 15% of Google queries in an earlier SparkToro study.

    Is ChatGPT replacing Google for product research? Not yet. SparkToro estimates purchase-related queries happen on Google at 16 times or more the rate on ChatGPT, and puts ChatGPT's buyer-journey traffic closer in size to Bing.

    Does ChatGPT recommend brands? 12.1% of responses in the study named at least one brand, product or company. ChatGPT recommended brands more often than users asked for them.

    Should crypto brands still invest in AI visibility? Yes, but in order. Google's AI Overviews and AI Mode come first, then the entity and citation work that feeds every engine, then targeted tracking in ChatGPT and Bing for the prompts closest to revenue.

    Want to see where your brand lands?

    We'll run your category's buyer prompts across ChatGPT, Perplexity, Gemini, Claude and Google's AI surfaces and show you who owns the answer today. Start with a growth plan conversation.

    Free · 5 business days

    Want this for your team?

    Start with twenty prompts against your brand. Free Growth Plan, no slide deck.

    Request a Growth Plan