Skip to main content
OneView automatically handles Conversion API errors and retries failed requests. No action required.

What is Conversion Replay?

Conversion Replay combines contextual error handling with automatic retries to ensure conversions always flow correctly to their destinations. OneView’s stateful architecture guarantees that once an event is accepted, it will eventually reach its destination—even during extended outages.
Understanding the problem:Traditional tracking solutions lose data permanently when server timeouts occur, temporary connectivity issues drop events, or endpoint unavailability blocks conversion data. Proxy solutions like GTM Server-Side are stateless by design—the browser cannot guarantee it will always retry a failed event, leading to irrevocably inconsistent data across your marketing platforms.Each advertising platform also has unique API error codes, configuration requirements, and attribution model compatibility rules. Without contextual error handling, you must manually interpret and resolve each platform’s specific errors.

How Conversion Replay works

OneView operates a high-throughput global queue system that classifies errors and applies platform-specific resolution strategies:
Error TypeClassificationAction
Transient errorsNetwork issues, temporary outagesAutomatic retries with exponential backoff
Rate limit errorsAPI rate limits exceededRetries with backoff to respect limits
Fatal errorsConfiguration issues, incompatible settingsRetries stopped, requires fix
Skipped eventsIntentional skips (consent, lookback windows)No retry, recorded for observability
Once an event is accepted by OneView, you’re guaranteed it will eventually reach its destination, even if downstream services experience extended outages (up to 7 days).

Error classification

OneView classifies integration responses into normalized outcomes.

Transient errors

Temporary failures that OneView automatically retries:
CauseExampleResolution
Network issuesConnection timeoutAutomatic retry with backoff
API outagesPlatform temporarily unavailableRetries until service recovers
Rate limitsToo many requestsRetries with exponential backoff
No action is typically required. If transient errors persist for prolonged periods, review rate limits and batch sizes for the specific integration.

Fatal errors

Non-retryable problems that require configuration changes:
CauseExampleResolution
MisconfigurationWebhook misconfiguredFix configuration, reprocess
Incompatible settingsAttribution model conflictAlign settings, reprocess
Validation errorsInvalid data formatFix data format, reprocess
Fatal errors are not retried. Fix the configuration and reprocess affected conversions if necessary.

Skipped events

Intentional skips to prevent errors or policy violations:
ReasonExampleBehavior
Consent not grantedad_storage deniedEvent skipped, no retry
Lookback window exceededConversion outside click-through windowEvent skipped, no retry
Skipped events are acknowledged by the pipeline (no retry). They are recorded for observability and do not count as failures.

Platform-specific handling

Each Conversion API integration is context-aware and handles platform-specific requirements:
PlatformAutomatic Handling
Google AdsAttribution model compatibility, Enhanced Conversions requirements
MetaConsent mode enforcement, event matching rules
Microsoft AdsConversion goal configuration, lookback window validation
TikTokEvent schema validation, attribution settings
OneView automatically adapts error handling to each platform’s API specifications, ensuring conversions flow correctly regardless of platform-specific quirks.

Benefits

No data loss Unlike stateless proxy solutions, OneView’s stateful architecture ensures events are never lost due to temporary network issues or endpoint failures. Automatic recovery OneView automatically retries transient errors and resolves configuration conflicts without manual intervention. Platform-specific intelligence Each integration understands its platform’s error codes, requirements, and behaviors, applying appropriate resolution strategies. Consistent reporting By guaranteeing delivery, OneView ensures your marketing platforms receive all conversions, preventing discrepancies between your dashboards and platform reports.

Common questions

OneView retries failed requests for up to 7 days. This covers most temporary outages while preventing indefinite retry loops for permanent failures. After this period, permanent failures are marked as fatal errors and require configuration fixes.
OneView stops retrying and marks the event as a fatal error. The pipeline may pause the affected webhook to prevent repeated failures. Once you fix the configuration and resume the webhook, you can reprocess affected conversions.
After 7 days, OneView stops retrying and drops the event. Some Conversion APIs (such as Meta) do not accept data past 7 days prior.
Browser-based tracking cannot guarantee retries because users may close the browser before retry completes, network conditions may prevent retry attempts, and no persistent queue exists to ensure eventual delivery. OneView’s server-side queue system persists events independently of user actions, guaranteeing delivery.
Yes. OneView logs show error classifications, reasons, and resolution actions for each integration, allowing you to monitor and troubleshoot issues. Retry attempts are visible while ongoing, and the final result appears in your reports once delivered.
If transient errors persist for prolonged periods, review rate limits and batch sizes for the integration, network connectivity and endpoint availability, and platform-specific requirements and recent API changes.

Next steps