❯ git show --details cases/trust-science.md
AI/ML

Pricing and discovery for a credit model that lenders had to trust

Trust Science scores borrowers that traditional bureaus cannot, using a patented AI/ML model. The buyers are lenders, banks and credit unions. I worked on how the product was priced, how its ROI was explained, and what the next version should do.

role
Consultant, Strategy & Product Development
timeline
2019 – 2020, Edmonton
users
Lenders, banks, credit unions and their underbanked borrowers
work
Pricing, unit economics, ROI models, discovery
outcome
23% user growth; +10 NPS

The problem

A lender buying an alternative credit score is buying a promise: approve more of the right people without raising losses. The model can be excellent and still not sell if the lender cannot see the ROI in their own numbers, or if the price does not fit how they already pay for bureau data.

What I did

  • Pricing and unit economics. Rebuilt the pricing around the lender's decision volume and the value of an incremental approval, with a model that showed each lender's break-even in their own terms.
  • ROI models for go-to-market. A standard ROI case that sales could tailor in a meeting, so the conversation moved from the model's accuracy to the lender's loss curve.
  • Discovery. Structured interviews with existing lenders and borrowers that reshaped the roadmap toward integration and explainability, which turned out to be what drove both growth and satisfaction.

Results

+23%user growth
+10NPS
Topranked Canadian fintech at the time

What stayed with me

This was my first product work on AI in a regulated setting, and it set a pattern I have followed since: the model is the entry ticket, and the product is whether the buyer can see, in their own numbers, that it is working.

06.

connect()

Building something where a wrong number costs money?
Fintech, accounting, operations, applied AI. I am always open to talking with people who ship.

$ echo "Let's build something that survives an audit"