Onboarding Conversion vs. Default Rate: The Real Tradeoff

The default debate is binary: growth wants higher conversion, risk wants lower default. Both are right, both are wrong, and the framing is what kills the platform.

The correct objective

The objective is not "maximise conversion" or "minimise default". It is maximise risk-adjusted activation, the volume of merchants whose 12-month expected contribution net of expected loss is positive.

That single reframing changes everything. Suddenly:

• A friction step that drops conversion 5% but cuts default by 30% is a clear win
• A risk hurdle that adds 3 days of activation but adds no measurable default reduction is a clear loss
• A tier change that moves 10% of merchants from T1 to T2 with the same default rate is the highest-leverage move available

The hard part is measuring it. The good news is you only need three numbers per acquisition cohort.

The three numbers

For each weekly cohort of merchants who applied:

1. % activated within 30 days (conversion)
2. Mean revenue per activated merchant at 12 months (contribution)
3. Mean loss per activated merchant at 12 months (chargebacks + fraud + write-offs)

The product is cohort_size × activation_rate × (revenue_per_merchant - loss_per_merchant). That is the number every onboarding change moves.

You will not have 12-month data for new cohorts. Use 90-day proxies and back-test them against historic 12-month outcomes once a quarter.

Where the tradeoff actually lives

Most "conversion vs default" arguments are about steps that affect neither in practice. The places where the tradeoff is real and large:

• UBO declaration depth, asking for 2 vs 4 layers of beneficial ownership
• Document re-upload friction, automatic OCR vs manual reupload on failure
• Sanctions hit handling, auto-clear with low-confidence vs manual review
• Bank account verification, micro-deposit vs instant Open Banking
• Vertical declaration, free text vs structured taxonomy

For each of these, run a controlled experiment. Most teams have never done this and are guessing.

The asymmetric cost

The cost of a missed good merchant is one missed merchant's revenue. The cost of an accepted bad merchant can be 10x to 100x that, especially in high-ticket verticals or with brand-damage chargebacks. Optimise asymmetrically:

• For low-ticket, low-risk verticals: bias to conversion
• For high-ticket, high-risk verticals: bias to default reduction
• Never use the same threshold across verticals

A single sensitivity threshold across all verticals is the fastest way to misallocate review capacity.

Time, not just decisions

Conversion is not only "did they pass?", it is "did they pass before they gave up?". Every 24 hours of activation delay typically costs 5–10% of cohort conversion. The cheapest conversion intervention is usually:

• Auto-OCR the document the first time, not on retry
• Surface what is needed in real time, not by email
• Show progress, never a blank waiting screen
• Promise a decision window and meet it

These cost very little and move conversion materially without touching risk thresholds.

What to instrument

• Activation rate, per acquisition channel, per week
• Time to activation, p50/p90
• Step-level drop-off, per step
• Default rate by cohort, by tier
• Risk-adjusted contribution per cohort
• Manual review queue ageing

Operator lens

The leaders who win this conversation are the ones who can show, with numbers, that a proposed friction change moves risk-adjusted activation up. Without that math, every meeting is a religious war between growth and risk. With it, the math wins.

Related: KYC and Conversion Designed Together · Risk Tiering Merchants as a Product Decision