Free 2-minute test
Pending isn't paid.
Five transfer, payout and balance statuses that fintech AI agents get wrong. Tap the reply that sounds like yours.
Your results
What your answers tell me
Your fixes, ready to paste.
Free
Unlock your fixes.
Enter your work email to see them now. Add your company website and I'll also build your company's version of this test from your help center, within 2 business days.
Unlocked
Add each one to your agent's guidance or procedures in Fin, Decagon, Lorikeet or Zendesk. They're written for a typical status model, so yours should use your own status names and policies.
What this costs at your volume
Assumes 1 in 10 conversations is about one of these five statuses, your agent misses them at the rate you picked, and each wrong answer brings the customer back to a human at $8 a ticket (the usual range is $6 to $13.50). The hours are my estimate at about 20 minutes per test. I deliver the full set in 10 business days.
Where these come from
I tested an AI support agent on 25 transfer scenarios. 8 failed at Critical. One page of written fixes cleared every Critical failure on retest.
Who's reading your answers
Allana Jackson. I write the instructions and tests AI support agents run on.
Before that: Google's Search Ads 360 billing and payments help center, plus knowledge systems for DoorDash and Domino's. See the full test