# Crypto AI Visibility Index, edition one — findings

480 responses. 40 prompts x 4 surfaces x 3 runs. Collected 2026-09-07/08 
via DataForSEO. Total API spend $4.96.

Gemini was dropped before the run on cost. That matters: in the pilot it 
surfaced stablecoins on both stage-1 prompts, more than any surface kept, 
so its absence probably lowers M1.


## M1/M2/M3 — surfacing, naming, recovery

| Engine | Stage 1 surfaced | Stage 1 named a product | Stage 2 surfaced | Gap |
|---|---|---|---|---|
| chatgpt | 61.1% | 1.1% | 100.0% | +38.9% |
| claude | 28.9% | 0.0% | 100.0% | +71.1% |
| google_aio | 34.4% | 5.6% | 100.0% | +65.6% |
| perplexity | 16.7% | 0.0% | 100.0% | +83.3% |
| **all** | **35.3%** | **1.7%** | **100.0%** | |

Stage 1 names a product in **6 of 360** runs. 
Stage 2 names one in **54 of 120**.


## M4 — what gets recommended instead

| Rank | Product | Responses |
|---|---|---|
| 1 | Wise | 246 |
| 2 | PayPal | 156 |
| 3 | Payoneer | 139 |
| 4 | Stripe | 127 |
| 5 | Revolut | 78 |
| 6 | TransferWise | 47 |
| 7 | Remitly | 46 |
| 8 | Western Union | 43 |
| 9 | Deel | 40 |
| 10 | Airwallex | 35 |
| 11 | Mercury | 35 |
| 12 | M-Pesa | 35 |

## M5 — caution asymmetry, response level

Share of recommendations carrying at least one caution that survived manual 
classification as a genuine warning.

| Engine | Crypto warned | Traditional warned | Ratio |
|---|---|---|---|
| chatgpt | 59/85 = 69% | 19/91 = 21% | 3.3x |
| claude | 17/56 = 30% | 12/82 = 15% | 2.1x |
| google_aio | 15/61 = 25% | 8/94 = 9% | 2.9x |
| perplexity | 22/45 = 49% | 16/68 = 24% | 2.1x |
| **all** | **113/247 = 46%** | **55/335 = 16%** | **2.8x** |

### Are those warnings any good?

| Kind | Accurate | Outdated | Generic fear | Not a warning | % accurate |
|---|---|---|---|---|---|
| crypto | 210 | 1 | 46 | 46 | 82% |
| tradfi | 43 | 0 | 16 | 14 | 73% |

Crypto cautions are **more** accurate than traditional ones. The dominant 
crypto warnings are depeg and reserve risk, custody and exchange failure, 
off-ramp friction, KYC/AML, tax treatment and absence of deposit insurance 
-- all correct. Engines also repeatedly distinguish stablecoins from BTC/ETH 
rather than applying blanket volatility fear.


## M6 — source basis

| Domain | Citations | Type |
|---|---|---|
| reddit.com | 293 | forum_or_social |
| wise.com | 185 | vendor_or_other |
| stripe.com | 88 | vendor_or_other |
| youtube.com | 81 | forum_or_social |
| linkedin.com | 48 | forum_or_social |
| worldfirst.com | 47 | vendor_or_other |
| facebook.com | 46 | forum_or_social |
| remitly.com | 44 | vendor_or_other |
| westernunion.com | 39 | vendor_or_other |
| deel.com | 36 | vendor_or_other |

By type: vendor_or_other 3409, forum_or_social 493, aggregator 50, editorial 26, docs 9


## Search rate

`web_search: true` only permits retrieval; it does not force it.

| Engine | Searched |
|---|---|
| chatgpt | 65/120 = 54% |
| claude | 78/120 = 65% |
| perplexity | 120/120 = 100% |