Definitional · 4 min read

Withdrawal is the only moment that matters: allocating friction where value is

Every operator knows their registration funnel drop-off to the decimal place. Almost none can tell you what their withdrawal review queue costs, and that asymmetry explains where the friction ended up.

Where should an operator put its identity friction?

At withdrawal, not registration. Operators apply their heaviest checks where the player is least committed and most likely to abandon, and their lightest where value actually leaves. That allocation optimises for conversion and against loss, which is precisely backwards.

Key takeaways

Friction is allocated by regulatory checkbox and historical accident rather than by value at risk.
Moving cryptographic friction from registration to withdrawal improves conversion and loss simultaneously, which is unusual and is the whole argument.
The manual review queue is a cost most operators have never attributed, and it frequently pays for the change on its own.

Part of Gaming and consumer marketplace identity

Where the money is, and where the friction is

Map the player lifecycle against two variables: how much value is at risk at each step, and how much friction the operator applies.

Step	Value at risk	Friction applied	Aligned?
Registration	Zero	Heaviest — document upload, selfie, verification wait	No
First deposit	Player's own money in	Light — payment rails handle it	Reasonable
Play	None to the operator	None	Yes
Payment method change	High	Light	No
Withdrawal	Highest	Variable — risk scoring and manual review	No

Registration friction is defensible on regulatory grounds — operators must verify identity before allowing play, and that requirement is not optional. The point is not to remove it but to recognise that it is doing regulatory work rather than fraud-prevention work.

The cost of registration friction

Every operator measures this because it is a marketing number. Document upload and verification steps produce material abandonment, concentrated among players who were least committed — which is to say, the marginal player the acquisition spend just paid for.

That cost is visible, attributed, and argued about in every product meeting.

The cost nobody attributes

The withdrawal review queue. Risk scoring flags a share of withdrawals; humans review them; legitimate players wait.

Three costs, and only the first is usually counted:

Review labour. Analysts, at a rate, for a volume. Countable.
Player churn from delayed withdrawals. Slow payouts are the most cited complaint in the sector, and players who experience one churn at elevated rates. Rarely attributed to the review queue.
Fraud that passes review. An account takeover matching the account's normal pattern is approved by exactly the process designed to catch it.

Add those three and compare against registration abandonment cost. In most operators they are the same order of magnitude and only one of them has an owner.

The reallocation

Move the cryptographic step to where the value is.

Keep regulatory verification at registration — it is required and removing it is not an option.
Enrol a credential during registration, at the moment the player is already completing verification steps. Marginal additional friction: seconds.
Require a credential-bound signature on withdrawal and on payment method change, rendering the amount and destination.
Reduce manual review scope to withdrawals that fail the signature or exceed a value threshold.

The player experience changes from submit withdrawal, wait for review to confirm withdrawal on your device, receive payment. That is a product improvement, not a control.

The arithmetic, stated plainly

For the change to be worth it:

savings = review_labour_reduction
        + churn_reduction_from_faster_payouts
        + ATO_loss_prevented

costs   = enrolment_friction_at_registration
        + support_volume_from_credential_issues
        + implementation

The second term in savings is usually the largest and the least measured. An operator that can pay out in minutes rather than days has changed its most-complained-about metric, and that shows up in retention rather than in the fraud budget.

What to measure before deciding

Withdrawals per month, and the share flagged for manual review.
Median and p90 time from withdrawal request to payment, split by reviewed and not reviewed.
Churn rate among players who experienced a review delay, versus those who did not.
Account takeover losses, and how many were approved through review rather than caught by it.

That fourth number is the uncomfortable one and it is the one that ends the debate. A review process that approves the takeovers is not a control; it is a delay applied to legitimate players.

Why the incentive points the wrong way

What each moment costs the operator
Moment	Cost of friction	Cost of no friction
Registration	Abandonment, measured daily	Some fraudulent signups
Deposit	Abandonment	Chargebacks
Withdrawal	Almost none — the player waits	Direct loss

Abandonment is measured continuously and attributed to the team that owns the funnel. Withdrawal loss is measured monthly and attributed to fraud. The allocation follows the measurement, not the risk.

Objections and honest limits

“Regulators require identity checks at onboarding.” They do, and that is a floor rather than a budget. Meeting the onboarding requirement does not oblige you to spend the rest of your friction budget there too.

“Players expect fast withdrawals.” They expect them to be fast and correct. A signature is seconds; a wrongly paid withdrawal takes weeks to unwind and is frequently unrecoverable.

Re-allocating the friction

Measure abandonment at each step. You already do for registration; do it for withdrawal too.
Move discretionary checks to withdrawal. Keep the regulatory minimum at onboarding.
Gate the payout destination hardest. Not the withdrawal amount.
Report loss and abandonment to the same owner. Split ownership is what produced the current allocation.

Terms used here

Friction budget: The total amount of user effort an operator can spend before conversion suffers materially.
Payout destination: The account a withdrawal pays to — the field that redirects value, as distinct from the amount.
Abandonment: Users who begin a flow and do not complete it, measured continuously at registration and rarely at withdrawal.

Frequently asked questions

Can we reduce registration friction? Not the regulatory verification, which is required. The argument is about where additional friction is added, not about removing what the licence conditions mandate.

What about players who do not enrol at registration? They follow the existing withdrawal review path. Coverage builds over time and the control improves as it does.

Does this reduce manual review entirely? No. Value thresholds and AML obligations will keep some review. It removes review from the population where the signature already answers the question.

How do we measure churn from withdrawal delays? Cohort players by whether they experienced a review delay and compare retention. Most operators have the data and have never run the query.

Why is friction concentrated at registration? Because abandonment is measured daily and owned by the growth team, while withdrawal loss is measured monthly and owned by fraud.

Do regulators require onboarding checks? Yes, and that is a floor. It does not oblige an operator to spend the remaining friction budget at the same moment.

What should be gated hardest at withdrawal? The payout destination rather than the amount.

Where this fits in Manav

Manav proves a specific person authorised a specific action, without a vault, a token or surveillance. The biometric never leaves the device and the platform receives a signature rather than a profile.

What we do not do →

Sources and further reading

Published gambling payment and account takeover fraud analyses.
Operator licence conditions on identity verification before play.
Consumer research on withdrawal speed as a driver of operator choice and complaints.
FinCEN advisories and alerts
FATF — publications