What is Retry Amplification?
Up: Retry Amplification See also:
Definition
Retry amplification occurs when a payment failure causes multiple follow-on attempts that multiply system exposure rather than resolving the original error. It is a state of exponential growth in transaction volume caused by decentralized or incorrectly configured retry logic across clients and servers.
Why it matters
The Multiplier Effect. Each retry represents a new transaction attempt that must be processed, risk-scored, and potentially disputed. If retries cluster in a narrow time window, they resemble "Card Testing" or a Brute-Force attack, triggering automatic upstream blocks that can shut down processing for a whole Merchant ID (MID).
Signals to monitor
- Unique-to-Total Attempt Ratio: Calculating how many attempts are generated per unique customer order.
- Retry Clustering: Identifying high-density authorization attempts appearing in sub-second time windows.
- Duplicate Authorization IDs: Tracking whether multiple approvals are being generated for the same intent.
- Dispute Exposure Expansion: Measuring the potential financial loss if every retry in a cluster were to be disputed.
Breakdown modes
- Distributed Amplification: Emergent behavior where the Client App, the Server SDK, and the Database Worker all retry a failure independently, turning 1 error into 3, 9, or 27 attempts.
- Latent Failure Retrying: Retrying transactions that have already failed due to a fundamental, non-recoverable error (e.g., Account Closed).
- Infinite Loop SDKs: Third-party libraries that continue to retry "Network Errors" without an exponential back-off or a maximum attempt count.
Where observability fits
Observability provides visibility into the "Lifecycle of an Intent." By linking multiple transaction attempts to a single original User Intent, the system can identify exactly where in the stack amplification is occurring and flag specific code modules for emergency remediation.