The Changing Landscape of Web3 Project Discovery
How do Web3 users actually find and evaluate new projects? The answer matters enormously for protocol growth teams, and it is changing fast. This study presents survey data from 2,400 active Web3 users across DeFi, NFT, gaming, and infrastructure categories, mapping discovery channels and measuring how AI assistants are reshaping the project evaluation funnel. The survey was conducted in January-February 2026 and supplemented by behavioral data from Presenc AI's monitoring platform.
The headline finding: AI assistants have grown from a niche discovery channel to the third most important source of Web3 project discovery in under two years, and they are on track to reach second place by late 2026.
Web3 Discovery Channel Rankings
We asked respondents to identify all channels through which they discover new Web3 projects, then rank their top three in order of influence on their evaluation decisions.
| Discovery Channel | % Who Use This Channel | % Who Rank It Top 3 | Avg Influence Score (1-10) | YoY Change |
|---|---|---|---|---|
| Crypto Twitter / X | 84% | 72% | 8.1 | -4% |
| Friends / Community (Discord, Telegram) | 71% | 58% | 7.6 | -2% |
| AI Assistants (ChatGPT, Claude, etc.) | 52% | 41% | 7.2 | +19% |
| Crypto Media (CoinDesk, The Block, etc.) | 58% | 38% | 6.8 | -6% |
| YouTube / Video Content | 49% | 32% | 6.4 | -8% |
| DeFi Aggregators (DeFiLlama, etc.) | 44% | 28% | 6.9 | +3% |
| Google Search | 41% | 22% | 5.8 | -12% |
| Podcasts | 32% | 18% | 6.1 | -3% |
| 28% | 14% | 5.4 | -7% | |
| Newsletter / Email | 36% | 16% | 5.9 | +1% |
Twitter/X remains the dominant discovery channel, but its influence is declining (-4% YoY) while AI assistants are surging (+19% YoY). The most striking data point is that 52% of Web3 users now use AI assistants for project discovery — more than double the 24% measured in our 2025 survey. Google search has seen the largest decline (-12% YoY), with AI assistants appearing to cannibalize traditional search for crypto research queries specifically.
The influence score tells a nuanced story: while AI assistants rank third in usage, their average influence score (7.2) is already close to community channels (7.6) and approaching Twitter (8.1). Users report that AI provides more balanced and comprehensive assessments than social media, which they view as increasingly promotional and unreliable.
AI Discovery Behavior by User Segment
AI assistant usage for project discovery varies significantly by user experience level and primary crypto activity.
| User Segment | % Using AI for Discovery | Avg AI Queries per Week | Top AI Use Case | Trust Level in AI Recs (1-10) |
|---|---|---|---|---|
| DeFi Power Users (3+ years) | 48% | 6.2 | Protocol comparison | 5.8 |
| DeFi Newcomers (<1 year) | 68% | 8.4 | Safety assessment | 7.4 |
| NFT Collectors | 34% | 3.1 | Project background research | 5.2 |
| Crypto Traders | 54% | 7.8 | Exchange/tool comparison | 6.1 |
| Developers / Builders | 62% | 9.1 | Technical documentation | 6.7 |
| Institutional / Professional | 58% | 5.4 | Due diligence | 5.9 |
DeFi newcomers are the heaviest AI users for discovery (68% adoption, 8.4 queries per week), and they also trust AI recommendations the most (7.4/10). This has profound implications for protocol user acquisition: the newest and most impressionable users are the ones most likely to be shaped by AI recommendations. Developers are the second-heaviest users (62% adoption, 9.1 queries per week), using AI primarily for technical documentation and tooling comparisons — reinforcing why developer documentation quality is such a strong predictor of AI visibility.
NFT collectors are the least likely to use AI for discovery (34%), preferring Twitter and community channels. This reflects the social, trend-driven nature of NFT markets where AI models struggle to keep pace with rapidly shifting cultural signals.
How AI Discovery Affects User Behavior
We asked respondents who use AI for project discovery to describe how AI recommendations influence their subsequent actions.
- 71% said they chose a product or protocol that appeared in an AI recommendation. This is the single most important statistic in this study: AI recommendations convert at an extraordinarily high rate when users are actively researching.
- 44% said they avoided a project because of an AI safety concern. The flipside of recommendation power is deterrence power — nearly half of AI-using respondents reported being steered away from a project by AI.
- 62% said they use AI to validate information found on social media. AI is increasingly used as a fact-checking layer on top of Twitter/Discord discovery, rather than as a standalone discovery channel.
- 28% said they have changed their primary wallet or exchange based on AI advice. AI is not just influencing new project discovery — it is reshaping existing tool preferences.
- 83% said they would use AI more for crypto research in the next year. The growth trajectory shows no signs of slowing.
The pattern that emerges is a two-stage discovery funnel: users first encounter projects through social channels (Twitter, Discord, YouTube), then turn to AI assistants for validation, comparison, and safety assessment. This means AI visibility is critical even for projects that primarily acquire users through community channels — if a user hears about your protocol on Twitter and then asks ChatGPT about it, the AI response can either validate or undermine the social signal.
Methodology and Demographics
This survey was distributed to 2,400 active Web3 users in January-February 2026 through crypto community channels, newsletter partnerships, and the Presenc AI network. Respondents self-identified their experience level and primary activity. The sample skews toward English-speaking markets (78% US/EU/UK), active DeFi users (62%), and users aged 22-40 (71%). Influence scores are self-reported on a 1-10 scale. Year-over-year comparisons reference our identical 2025 survey (n=1,800). Behavioral data points (AI query volumes, conversion rates) are supplemented by Presenc AI platform analytics where noted. The full methodology document and raw data summaries are available from the Presenc AI research team upon request.