The Human/AI Collections Matrix


Generative AI risk assessment to help you evaluate vendors.

The Human/AI Collections Matrix


What to Automate, What to Escalate, and What Happens to the Humans Left Behind

P.S. It’s free, just click the button below 🙂

Here’s a sneak peek…

AI is beginning to take a larger role in collections and financial services conversations. 

But the most important question isn’t whether AI can make a call.

It’s which calls it should make/take and what happens to your agents when it does.

But if AI takes the easiest calls, the average human call gets harder.

The Human/AI Collections Matrix

There probably shouldn’t be one simple rule that says “AI calls go here” and “human calls go there.”

A better decision model looks at several characteristics at once:

Conversation typeOwnerComplexityCompliance riskEmpathy requiredJudgment required
Balance inquiryAI1211
Due-date reminderAI1211
Standard account FAQAI1211
Simple paymentAI / Hybrid2312
Promise-to-pay arrangementHybrid3323
Negotiated payment planHybrid4434
Consumer disputeHuman + AI assist4534
Financial hardshipHuman + AI assist4455
ComplaintHuman4555
Identity uncertaintyHuman4525
Legal threat / attorneyHuman5535
Bankruptcy / deceasedHuman5555
Vulnerable consumerHuman5555

*These ratings are an Abstrakt operating framework rather than regulatory classifications. Companies should adapt the matrix to their business, clients, policies, and risk tolerance.

What Happens When AI Takes 44% of the Conversations?

This is where workforce planning gets interesting.

Consider an operation with 100,000 completed monthly conversations.

For simplicity, we’ll divide them into three groups:

Call categoryMonthly conversationsShareModeled AHT
Routine40,00040%3 min
Moderate35,00035%6 min
Complex25,00025%10 min
Total100,000100%5.8 min weighted AHT

Download the guide to keep reading!