Progressive Trust Underwriting
Start with limited exposure and expand access as evidence builds trust.
- Difficulty
- Advanced
- Time to result
- ~months to results
- Steps
- 5
- Confidence
- 98%
Progressive Trust Underwriting replaces a large upfront approval process with a sequence of small, evidence-based risk decisions. A new customer receives enough access to begin operating, but the provider initially limits transaction size or total exposure. Each completed transaction supplies new evidence: successful settlement raises confidence, while fraud, chargebacks, or abnormal behavior lowers it. Limits then expand or contract as the observed risk changes. The mechanism simultaneously reduces onboarding friction and caps losses from unproven participants. Square used this approach to admit merchants whom traditional financial institutions rejected, while continuously re-underwriting them after transactions. The method works because trust is treated as a dynamic quantity earned through behavior rather than a permanent status awarded before use.
Origin
Square used transaction data to admit previously excluded merchants while controlling fraud risk, challenging incumbents that relied on extensive upfront underwriting.
Core principles
- 01Replace burdensome upfront screening with controlled initial exposure.
- 02Treat successful transactions as evidence that updates trust.
- 03Increase limits gradually rather than making a single binary decision.
- 04Cap early losses so experimentation remains economically survivable.
How to run it
- 1
Set the initial risk boundary
Choose a deliberately small transaction or exposure limit that allows genuine use while containing the loss from a bad actor.
Pro tip Express the limit in terms of maximum tolerable loss, not desired customer volume.
Watch out An initial allowance that is too large defeats the purpose of progressive underwriting.
- 2
Let the customer begin quickly
Remove unnecessary approval hoops and grant limited access as soon as basic identity and eligibility checks pass.
Pro tip Explain that greater access becomes available through successful use.
Watch out Lower friction should not mean abandoning legally required verification.
- 3
Collect behavioral evidence
Observe successful settlements, transaction patterns, chargebacks, fraud indicators, and other relevant outcomes after every use.
Pro tip Weight recent and independently verified behavior more heavily.
Watch out Do not interpret transaction volume alone as evidence of legitimacy.
- 4
Re-underwrite continuously
Update the risk estimate whenever meaningful new evidence arrives rather than waiting for a periodic manual review.
Pro tip Automate routine updates but preserve human escalation for unusual patterns.
Watch out A static model will miss changes in customer behavior.
- 5
Expand or contract access
Raise limits for customers with reliable histories and reduce or suspend access when adverse signals appear.
Pro tip Use gradual limit changes to avoid sudden increases in exposure.
Watch out Never make limit growth automatic without checking for correlated fraud signals.
In the wild
A small merchant unable to obtain a traditional merchant account joins Square without navigating a long bank application. Square permits a limited amount of card volume, monitors the resulting transactions, and increases the merchant's usable capacity as legitimate payments settle without fraud or chargebacks.
→ The merchant gains fast access to card acceptance while Square contains its early risk.
A marketplace allows a new seller to list products but initially delays or caps payouts. As orders arrive on time and buyers report no problems, the platform shortens the payout delay and increases the seller's limit.
→ Legitimate sellers earn better terms while fraudulent accounts remain economically constrained.
Common mistakes
Granting full access immediately
A frictionless signup does not require unlimited exposure. Giving an unproven participant full capacity concentrates risk before behavioral evidence exists.
Using a one-time approval decision
Risk changes over time, so the original decision must be updated as transactions generate new evidence.
Growing limits on volume alone
High activity can indicate adoption or fraud. Limit increases should depend on clean outcomes and settlement quality, not merely throughput.
Is it for you?
Best for
This is best for platforms that can observe repeated transactions and continuously update customer risk.
Not ideal for
It is not ideal for irreversible, one-time transactions where no behavioral history can accumulate.
From the transcript
“We're not gonna put you through a bunch of hoops, we're just gonna give you a little bit, you know, that basically a little bit…”
“And with every time that you completed a transaction and it wasn't a chargeback and it wasn't fraud and it wasn't all these things, they…”
“And so that way they could actually get you started quickly and trust you more over time, which was just nothing, it was completely innovative…”
From the episode
The Square IPO