AEO Test deep dive: the 47-prompt benchmark for crypto AI visibility
How the AEO Test scores AI engine visibility across 47 vertical-specific prompts. Reading the report, what the median 11 out of 100 means and a 60-day playbook to move from 11 to 40+.
AEO Test vs AI Citation Checker: when to use which
Crawlux ships two AEO measurement tools. The AI Citation Checker runs 12 calibrated prompts and returns a citation rate in 47 seconds median. The AEO Test runs 47 vertical-specific prompts and returns a detailed 0-100 score in 4 minutes median.
Use the Citation Checker for quick health checks, regression testing after fixes and continuous monitoring. Use the AEO Test for strategic planning, competitive analysis and identifying which crypto verticals you have the strongest right-to-win in. The two tools share the underlying calibration methodology covered in the AI Citation Checker press release.
The 47-prompt set: how it is organized
The prompt set covers 9 crypto verticals at varying depth: DeFi (12 prompts across lending, perpetuals, spot DEX, options), staking (6 prompts across liquid, native, restaking), NFT (5 prompts across marketplaces, collections, GameFi), wallets (6 prompts across custodial, self-custody, smart account, hardware), infrastructure (6 prompts across oracles, bridges, indexers, RPC providers), stablecoins (4 prompts across collateralized, algorithmic, payment-focused), centralized exchanges (3 prompts), layer-1 chains (3 prompts) and layer-2 chains (2 prompts).
Within each vertical, prompts span the three intent buckets. Head intent ("best DeFi lending protocol"), body intent ("compare Aave and Compound for stablecoins") and tail intent ("how does the GHO peg mechanism work"). Tail prompts carry the highest weight in scoring because they correlate best with downstream business outcomes for crypto sites.
The 0-100 scoring math
The AEO score weights citation outcomes across the 47 prompts by three factors. Engine coverage (cited on 1, 2 or 3 of ChatGPT, Perplexity and Claude), citation position (first cited source vs second vs third) and prompt intent weight (tail intent weighted 2.5x more than head intent).
A page that wins a tail-intent prompt as the first cited source on all three engines scores roughly 6.2 points. A page that wins a head-intent prompt as the third citation on one engine scores roughly 0.4 points. The maximum theoretical score across 47 prompts is 100 (achievable in practice only by category-dominant protocols like Uniswap for spot DEX queries).
The score is not bell-curved or normalized. A site scoring 40 actually does about 4x as well as a site scoring 10. This makes the metric usable for goal-setting and progress tracking without curve adjustments.
Reading your AEO Test report
The report has three sections. The headline score across all 47 prompts. The per-vertical breakdown showing which verticals you score highest and lowest in. The per-prompt drill-down listing the specific prompts where you were cited (with position and engine), where you were close (mentioned but not as the citation) and where you were absent.
The strategic value sits in section 3. Pages where you were close are the cheapest wins. The content is already known to AI engines for the relevant query; small fixes (schema, citation pattern, content depth) often flip the page from close to cited. Pages where you are absent require new content or topical authority development.
The competitor column shows which domains got the citation instead. This is the single most actionable piece of the report. Reverse-engineering what the cited domain does differently provides a concrete fix list.
What the median 11 out of 100 means
Beta data: 156 crypto sites tested. Median AEO score 11/100. 78% scored below 20. 17% scored 20-50. Only 5% scored above 50. Top performer at 73/100 was a DeFi lending protocol with strict FinancialProduct schema discipline and 47 audit firm citations.
Two ways to read 11/100. Pessimistic: AI engines do not cite crypto sites well. The optimistic read: the market for AI visibility is wide open because almost no one is competing in it yet. The optimistic read is the right one. AEO is at the stage SEO was at in 2003. The teams that move first capture compounding citation authority that gets harder to displace later.
For protocols, the 11 median means even small AEO improvements show up disproportionately. Moving from 11 to 25 puts you above 86% of competitors. Moving from 25 to 40 puts you in the top 5%. The market structure rewards early movement.
A 60-day playbook to move from 11 to 40+
Week 1-2: Mechanical fixes. Allow GPTBot, ClaudeBot and PerplexityBot in robots.txt (see the robots.txt guide). Migrate token schema from Product to FinancialProduct (see the schema guide). Add audit firm citations with linked reports. These three fixes typically move score from 11 to 18-22 within 10 days.
Week 3-4: Content depth on top product pages. Expand each major product page to 1,500+ words covering mechanism, edge cases, comparisons and data tables. Move FAQ answers from the bottom to the lead paragraph. Tie all factual claims to verifiable sources. Most teams see another 8-12 point lift in this window.
Week 5-6: Tier 1 source presence. Submit your protocol to DefiLlama, CoinGecko, Etherscan token lists and DappRadar with complete metadata. Maintain accurate TVL and volume data. This week typically adds 3-5 points to the score as AI engines start incorporating the structured data.
Week 7-8: Iteration. Run the AEO Test weekly. Identify the 5-7 prompts where you score "close but not cited". Fix the specific gaps for each (often a schema field, a missing citation or a content-depth gap on the linked page). The compound effect adds another 4-8 points. By end of week 8, most beta participants moved from 11-15 to 30-45.
Vertical-specific patterns from beta data
AEO scoring patterns vary by vertical. DeFi lending sites benefit most from FinancialProduct schema discipline (median 14 point lift). Wallet sites benefit most from audit firm citations and feature comparisons (median 11 point lift). NFT marketplaces benefit most from collection-specific page depth (median 9 point lift). Layer-1 chains benefit most from developer ecosystem citations on GitHub (median 12 point lift).
The Crawlux Pro audit returns vertical-tuned recommendations based on the prompt set most relevant to the audited domain. Generic AEO advice misses the variation. The Crawlux AI Visibility Audit module covers the vertical-specific patterns automatically.
Take
Median AEO score across 156 crypto sites: 11 out of 100. The market for AI visibility is wide open because almost no one is competing in it yet.
// Related
Crawlux is the world's first automated SEO audit tool built for Web3, DeFi and blockchain. The platform runs 23 analyzers across 6 check groups including AI visibility testing across ChatGPT, Perplexity and Claude. Free tier available. Paid tiers from $25 per audit. More at crawlux.com.
Frequently asked questions
How often should I run the AEO Test?
Weekly during active optimization sprints. Monthly during steady state. Crawlux Pro runs it weekly automatically and alerts on score moves greater than 5 points.
Why does the AEO Test take 4 minutes when the Citation Checker takes 47 seconds?
The AEO Test runs 47 prompts across 3 engines (141 total queries) compared to the Citation Checker's 12 prompts across 3 engines (36 queries). Parallel execution plus rate limits on the AI engines push runtime to roughly 4 minutes median.
Is the score comparable across different audit dates?
Yes within the same quarterly prompt version. The prompt set updates quarterly; scores from before a quarterly update are not directly comparable to scores after. Crawlux Pro tracks prompt-set version with each score.
Can I see prompts where my competitor wins but I am close?
Yes. The free tool shows this in section 3 of the report. Crawlux Pro adds competitor-specific drill-down with up to 5 named competitors per audit.
RUN YOUR FIRST AUDIT FREE
See Crawlux on your own crypto site.
No signup, no credit card. Full Web3-tuned audit report in 60 seconds.
Free first audit · No signup · 60 seconds · Full PDF report