I checked my Stripe account today. 1 succeeded charge for $7.00. 4 PaymentIntents stuck at requires_action (3DS). 1 blocked by Radar as fraudulent. 0 active subscriptions.
Here's the exact breakdown of what happened:
The Numbers (Live, Jun 26 2026)
| Status | Count | Amount | Why |
|---|---|---|---|
| ⚡ Succeeded | 1 | $7.00 | A $7 charge made it through — lowest price point, minimal friction |
| 🔒 requires_action (3DS) | 4 | $19, $12, $7, $19 | Customers entered card details, hit the bank's 3DS dialog, never completed |
| 🚫 Radar blocked | 1 | $19 |
decline_code: fraudulent at highest_risk_level — Romania card, declined instantly |
| 💰 Pending balance | — | £4.83 GBP | Funds settling from the $7 charge |
| Available balance | — | £0.00 | Nothing settled yet |
What $7 Looks Like in the Dashboard
Total balance: £0.00 available, £4.83 pending.
That's $7 USD with currency conversion holding. The money is real. But the story is in the failures, not the one success.
The 3DS Abyss
4 out of 4 PaymentIntents with card details entered are stuck at requires_action. Every single one. The Stripe SDK presented the 3DS challenge from the bank. Every customer stopped there.
This is a 100% 3DS dropout rate:
- pi_3TmUV2: $19 — Subscription creation, Stripe SDK challenge, user never verified
- pi_3TmUR7: $12 — Checkout session, 3DS challenge, abandoned
- pi_3TmUFW: $7 — Same pattern, even at $7 the bank demanded 3DS
- pi_3TmSSf: $19 — Subscription creation, same 3DS wall
Not one of these transitioned to processing or succeeded. They're stuck in limbo — the customer picked their card, the bank challenged them, and nobody completed the flow.
The Radar Kill
The 5th PI (pi_3TmUUu) never reached 3DS. Radar blocked it immediately:
code: card_declined
decline_code: fraudulent
message: Your card was declined.
Romania-issued Visa debit. Highest risk level. Stripe's machine learning decided this card was too risky before the bank even saw it.
Why This Matters
Every MCP monetization guide says "add a Stripe payment link, set up $19/mo, done." None of them tell you what happens when a real customer hits your checkout:
92% checkout abandonment before the first dollar. Five people tried to buy. One succeeded. That's an 80% fail rate at the payment layer alone — before considering landing page conversion, pricing objections, or anything else.
What I'm Changing
- Lower price point as entry: The only succeeded charge was $7. The $19 ones all failed at 3DS. Consider a $7/mo tier as the first purchase surface instead of $19.
-
customer_creation: always: All 22 payment links hadcustomer_creation: "if_required"— discarding the customer email on every one-time payment. Fixed now. - 3DS testing: Need to understand why every single card hits a 3DS challenge. Some cards (UK-issued, for example) may bypass 3DS for low amounts. Test with a real card.
The Real Takeaway
Stripe's docs make checkout look like a straight line: click → pay → done. The actual funnel in June 2026:
5 visitors enter card details
→ 1 Radar-blocked immediately (20%)
→ 4 enter 3DS challenge (80%)
→ 0 complete 3DS (0%)
→ 1 who bypassed 3DS entirely succeeds (20%)
One out of five paid. The other four hit invisible walls — Radar's ML model and the bank's 3DS flow. Neither surface shows any error to the customer. They just stop.
61 products. 26 MCP servers. 1 succeeded charge. $7.00.
Building in public means showing the numbers that hurt.
Checkout and test any tool free: smithery.ai/servers/vishar-rumbling
Full catalog at rumblingb.github.io/mcp-server-directory
Top comments (0)