New ICODA research, conducted in partnership with Semrush, their official partners, tested how ChatGPT and Perplexity cite crypto sources across four query categories. Data aggregators never appeared once. Four other source types consistently did — and only one of them matches what most GEO advice currently recommends.
Bellevue, 10.08.2026. ICODA, a growth marketing agency for Web3 and iGaming brands, published original research quantifying a gap between the GEO advice most crypto marketers receive and what AI engines actually do.
The research directly tests the most repeated recommendation in crypto AI visibility strategy: get listed on CoinGecko, CoinMarketCap, and DeFiLlama, and AI engines will pick you up as a verified source. That assumption treats data aggregators as the default verification layer for AI, the same role they play for human researchers checking a token’s price, market cap, and contract address in one place. The audit was designed to test whether that transfer holds.
The Finding
ICODA, working with Semrush and the Semrush One platform, ran over 100 controlled query-runs across ChatGPT and Perplexity. Four query categories covered how real users ask about crypto projects: recommendation-style (“best X in 2026”), breaking news, safety and legitimacy, and community-framed prompts explicitly asking what people are saying. Each category ran in clean sessions with the full answer and every citation captured.
The top-line result: across every category and both engines, CoinGecko, CoinMarketCap, and DeFiLlama did not appear once. No citation, no passing mention, nothing.
What appeared instead splits cleanly into four source types:
Purpose-built SEO roundups — Coin Bureau, Strategy Arena, and Datawallet won recommendation-style queries outright. The same query independently cited Coin Bureau in both engines — the only source type to appear in both without coordination. These sites out-compete data tables not on authority but on structural match: a page built to answer “best DeFi protocols in 2026” beats a price feed for that query every time.Financial media and trade press — Safety queries surfaced Investopedia, Reuters, Yahoo Finance, and Business Insider. Regulatory depth pulled The Block, paulhastings, and Investors.com — a separate outlet set with zero overlap. paulhastings is a law firm publishing legal analysis; when a query requires regulatory depth, the engine reached for a source built to provide it.The company itself — Circle and Tether were cited directly on a stablecoin safety query — named as primary sources, not summarized by a third party. The specific condition: a standing record of recurring, audited reserve disclosures. Not size. Not brand recognition. The audit record.Community platforms — Reddit surfaced in one query that explicitly named it, and even then appeared alongside a blog that had already summarized the same sentiment secondhand. Community reach is secondhand by default.
Every citation in the dataset followed the same sequence:
Query intent → source structure → relevance → citation
The engine is not looking for the most authoritative site in the category. It is matching content structure to query intent — and that is a different optimization problem than the one most crypto marketers are currently solving.
Cross-Engine Overlap
Citation overlap between ChatGPT and Perplexity is not a single number. On roundup queries, it was high: both engines independently cited Coin Bureau. On breaking-news queries, it dropped to zero — the same event surfaced entirely different outlets in each engine. On the stablecoin safety query, overlap was zero again, but for a different reason: ChatGPT went to primary sources, Perplexity to comparison blogs.
Semrush’s own AI Visibility Index, measured across 126 million prompts, found the same structural pattern at industry scale: source overlap between which brands get mentioned and which sources actually get cited can drop to 30% on a single platform like Gemini. Averaging citation data across query types into one aggregate score hides exactly the structural differences that determine what gets cited.
Methodology
ICODA, working with Semrush and the Semrush One platform, ran over 100 query-runs across ChatGPT and Perplexity in clean sessions. Four query categories — recommendation-style, breaking news, safety and legitimacy, and community-framed — were each replicated across both engines, with full answers and all citations captured. Nine representative runs are published as walkthroughs in the full report; the complete dataset produced the four source-type patterns described above. Queries ran in August 2026. No data aggregator citation was observed at any point during data collection.
About ICODA
ICODA is a global agency specializing in full-cycle digital marketing, public relations, and growth strategies for Web3, DeFi, GameFi, iGaming, and crypto exchanges. Core services include Crypto PR, influencer and KOL campaigns, community management, token sale execution, and AI Overviews SEO and GEO.
