Skip to main content

Playbook: Increase Authorization Rates

Last updated . Copy page as text

TL;DR
  • Week 1: Retry logic experiments. Soft declines (51, 61) retry timing, technical (91, 96) immediate retry, unless Mastercard's advice code says wait or stop
  • Week 2: Card updater enrollment (VAU, ABU), proactive expiration emails
  • Week 3: 3DS on recurring setup, frictionless tuning (more data = higher frictionless rate)
  • Week 4: Network token conversion for card-on-file; compare matched payments with your baseline
  • Measure each experiment individually: Count additional captured payments, fees and fraud losses

Systematic approach to improving payment authorization rates.

Auth rate optimization is a series of experiments, not a checklist. Each change is a hypothesis about what's causing declines.

Workflow Overview​

PhaseKey Tasks
Retry LogicBaseline assessment, soft decline retry timing, technical retry experiments
Card UpdateAccount updater enrollment (VAU, ABU), proactive expiration outreach, measure ROI
3DS Tuning3DS on recurring setup, frictionless tuning (more data), measure impact
Network TokensConvert cards to network tokens, compare matched payments, calculate ROI

When to Use This Playbook​

  • Auth rates falling below your own baseline on matched issuer, country and transaction-type segments
  • Significant auth rate decline
  • High rate of soft declines
  • Customer complaints about payment failures

First Experiment to Run This Week​

Hypothesis: We don't know which decline codes are actually addressable.

Experiment:

  1. Pull your top 5 decline codes from last 30 days
  2. Categorize each as: retry-able, card-updatable, fraud-related, or hard decline
  3. Calculate what % of declines are actually addressable

Expected outcome: You'll know where to focus. Count retry-eligible declines and cards that need updated details, then measure how many become captured payments.

Baseline Assessment​

Current State:
□ Overall auth rate: _______%
□ Auth rate by card brand:
   - Visa: _______%
   - Mastercard: _______%
   - Amex: _______%
□ Auth rate by transaction type:
   - First purchase: _______%
   - Recurring: _______%
   - Card on file: _______%

Decline Code Categorization​

CategoryExample CodesAddressable?How to Test
Insufficient funds51, 61, 65YesRetry with timing experiment
Card errors14, 54, 41, 43YesAccount updater
Do Not Honor05MaybeMultiple approaches
Technical91, 96YesImmediate retry
Fraud blocks57, 59MaybeReview your fraud rules

Week 1: Retry Logic Experiments​

Experiment: Soft Decline Retry Timing​

Hypothesis: Retrying insufficient funds declines at end of month will recover X% of transactions.

Test:

  • Segment A: Retry code 51/61 after 24 hours
  • Segment B: Retry code 51/61 on the 1st and 15th of month
  • Segment C: No retry (control)

Metrics: Recovery rate, customer complaints

Run length: 30 days (need full month for payday cycles)

Suggested starting guardrail: Pause if duplicate-charge complaints exceed 0.1% of retried payments. Tune this to your traffic and investigate every reported duplicate.

Experiment: Technical Retry​

Hypothesis: Retrying code 91/96 immediately will recover X%. Skip the retry when Mastercard sends MAC 02 (wait 24 hours) or MAC 03/21 (stop).

Test:

  • Retry technical errors once immediately
  • If fail, retry again after 5 seconds
  • Max 3 attempts

Metrics: Recovery rate per retry

Expected: Compare technical-error recovery with a matched control. Count successful captured payments, repeat failures and duplicate-charge complaints.

Week 2: Card Update Experiments​

Experiment: Account Updater ROI​

Hypothesis: Account updater will improve auth rate by X% and is worth the cost.

Test:

  1. Enroll in Visa Account Updater (VAU) and Mastercard ABU
  2. Measure: cards updated, auth rate lift on updated cards, cost per update

Decision rule: If auth rate lift × transaction value > cost per update, keep it

Results to check: Compare approvals and captured payments on matched customers with and without updates. Keep it only when added value covers update fees and losses.

Experiment: Proactive Expiration Updates​

Hypothesis: Reaching out to customers before card expiration reduces involuntary churn.

Test:

  • Segment A: Email customers 30 days before expiration with easy update link
  • Segment B: No proactive outreach (control)

Metrics: Card update rate, churn rate, auth rate

Week 3: 3DS Optimization​

Experiment: 3DS on Recurring Setup​

Hypothesis: Using 3DS on initial recurring transaction will improve subsequent auth rates.

Test:

  • Segment A: Require 3DS on initial subscription transaction
  • Segment B: No 3DS on initial transaction

Metrics: Initial conversion (expect drop), subsequent recurring auth rate (expect lift)

Decision: If recurring auth improvement × LTV > initial conversion drop × AOV, keep it

Experiment: Frictionless 3DS Tuning​

Hypothesis: Sending more data to 3DS will increase frictionless rate without hurting fraud protection.

Test:

  1. Baseline your current frictionless rate
  2. Add additional data fields (device info, customer history, etc.)
  3. Measure frictionless rate change

Expected: Compare frictionless completion before and after on matched issuers, countries and risk segments. Check fraud losses and completed purchases too.

Week 4: Network Tokens​

Experiment: Network Token Conversion​

Hypothesis: Network tokens will improve auth rates on card-on-file transactions.

Test:

  • Convert existing saved cards to network tokens (Visa, Mastercard)
  • Compare auth rates: tokenized vs. PAN

Results to check: Compare token and PAN approvals on matched payments. Visa publishes a scoped comparison; measure your own captured revenue and costs.

Cost consideration: Some processors charge per token. Calculate ROI.

Measurement Framework​

For each experiment:

ExperimentHypothesisMetricControlResultKeep/Kill
Retry timingRecover eligible declinesRecovery rateNo retry___%
Account updaterReduce failures from stale detailsAuth rateNo updater___%
3DS on recurringImprove recurring approvalsRecurring authNo 3DS___%
Network tokensImprove COF approvalsCOF authPAN___%

Expected Improvements (Cumulative)​

OptimizationWhat to MeasureCompare With
Smart retryCaptured payments after eligible declinesMatched no-retry control
Account updaterApprovals and captured payments after updatesMatched customers without updates
Network tokensApprovals, captured revenue and feesMatched PAN payments
3DS optimizationChallenge completion, approvals and fraud lossesMatched eligible payments under the prior setup
Data quality fixesApprovals and completed purchasesMatched payments before the fix

Cumulative potential: Measure the combined result after testing each change. Don't add separate lifts: they can affect the same payments.

More approvals aren't the whole result. Count the extra payments you captured, then subtract fees and fraud losses.

Where Experiments Lie to You
  • Seasonality: auth rates move with spending patterns. Compare to the same period last year, because last month isn't the control.
  • Card mix changes: If you're acquiring different customer types, your baseline is shifting.
  • Processor changes: if you switched processors, your auth rates moved for that reason and you can't read your optimizations through it.
  • Selection bias: If you only retry "good" looking declines, you'll overestimate retry effectiveness.

Anti-Pattern: What Not to Do​

  • Don't implement all optimizations at once (can't measure what worked)
  • Don't retry hard declines (05 codes) aggressively (can get you blocked)
  • Don't assume vendor claims (test their claimed lift on your own matched payments)
  • Don't optimize auth rate at the expense of fraud rate (measure both)

Next Steps​

After running optimization experiments:

  1. Keep winners, kill losers: Document which experiments improved auth rates and make them permanent
  2. Set baselines: Lock in new auth rate benchmarks for future comparison
  3. Monitor for regression: Set alerts if auth rate drops below new baseline
  4. Expand to other segments: Apply winning strategies to international or different card types
  5. Balance with fraud: Verify fraud rate didn't increase alongside auth improvements

Something wrong on this page? Email matt at paymentsandrisk.com and say what needs fixing.