TikTok Comment Webhook Not Arriving? Check Polling and Routing
A UnifyPort post.comment.created subscription does not switch on TikTok comment polling. Polling defaults off: configure provider_data.enable_comment as the Boolean true, then apply it through start or reconnect; changes to an online account require reconnect. Check the receiving endpoint separately. A successful start is not proof that polling was applied, and a quiet webhook is not proof that nobody commented.
Key takeaways
- Separate comment discovery, connection configuration, endpoint routing, and application processing.
- Verify environment availability and prepare the receiver before enabling polling.
- Account-specific webhook configuration can override workspace configuration without falling back when inactive or filtered out.
- Notifications are best-effort signals, not complete history or permission to reply automatically.
First identify which webhook you configured
TikTok’s official developer webhook catalogue describes its own event names and request structure. UnifyPort’s post.comment.created reference describes a different contract: a normalized TikTok comment notification after identification and own-post ownership checks.
Do not put a UnifyPort event name into a TikTok Developer Portal configuration and expect that to configure UnifyPort. Keep official TikTok webhook handling separate from the connected-account unofficial interface. This guide concerns the latter, not TikTok Shop customer-service messages or DMs.
If the account is not connected, complete the TikTok QR authorization setup first. Authorization and a running connection are prerequisites to investigate, not evidence that comment polling is enabled.
Trace the four boundaries in order
| Boundary | Evidence to inspect | What that evidence cannot prove |
|---|---|---|
| Discovery | Environment supports the feature; polling deliberately enabled | Every historical comment will be found |
| Connection | Configuration change followed by the required start or reconnect | A successful start applied the polling setting |
| Routing | Correct endpoint scope, active status, matching subscription | The upstream source has produced an event |
| Processing | Authenticated event durably stored and processed | Every comment was delivered or a reply appeared publicly |
This order avoids changing signing secrets to solve disabled polling, or repeatedly reconnecting to solve a filtered-out event.
1. Review the saved configuration, not a UI checkbox
The Update account reference defines provider_data.enable_comment as Boolean-only. true and false are valid; strings, numbers, and null return 400 invalid_request. Record the intended value and the actual API result in your protected configuration workflow.
Treat provider_data carefully: omitting it retains configuration, while {} clears it. Do not assume a partial nested object merges with all existing provider settings. Prepare the intended configuration from your controlled source, preserving required values without logging secrets. A read response that omits sensitive configuration is not a complete configuration backup.
For an online account, arrange one deliberate runtime reconnect after the update. For a stopped, authorized account, use the documented start path when appropriate. Coordinate with any recovery worker so it does not issue competing actions. The public contract does not give a successful start response the meaning “comment polling verified.”
2. Check the endpoint that actually governs this account
Inspect both workspace endpoints and independent account endpoints using List webhook endpoints. Listing without account_id returns workspace endpoints; listing with it returns that account’s independent configurations, not inherited workspace configurations.
The webhook configuration reference states that account configuration overrides workspace configuration. An inactive independent endpoint or a subscription that does not match does not fall back to the workspace endpoint. This explains why a healthy workspace receiver is insufficient evidence for this particular account.
Inspect the governing endpoint’s status and subscription. Include post.comment.created explicitly, or use the documented broad selection such as ["*"]. Preserve any existing message and authorization subscriptions you still need. The event-filter guide explains the filter’s role; filtering selects events but does not start the upstream polling process.
Do not delete an independent endpoint merely as a diagnostic experiment. Deleting the last independent configuration restores workspace inheritance for subsequent initial deliveries, changing where future events go. Review that routing change intentionally.
3. Verify intake without inventing comment fields
Use the UnifyPort delivery contract: with signing enabled, verify X-Device-Signature as HMAC-SHA256 over X-Device-Timestamp, a dot, and the raw body; check timestamp freshness before accepting the event. Persist authenticated deliveries before returning 2xx and deduplicate this event type by event id within your workspace.
The event reference establishes the common envelope—id, type, provider, account_id, occurred_at, and data—but says a detailed field reference and captured sample are still being prepared. Do not build a production parser around guessed data.comment paths or synthesize a sample as if it were a documented payload. Keep the authenticated envelope for controlled inspection and confirm the actual payload contract before enabling field-dependent automation.
Use a controlled check, not a reconnect loop
The following is a proposed diagnostic exercise, not a reported test result:
- Confirm feature availability for the target environment and identify an owned post.
- Keep automatic replies disabled. Prepare signed intake and the correct endpoint subscription.
- Apply the Boolean polling setting and required connection action once.
- With authorized test participants, add a distinguishable comment on that post and retain its native reference securely.
- Follow receiver access, signature validation, durable storage, deduplication, and worker processing separately.
If the signal is missing, retain redacted account and endpoint identifiers, configuration/action results, times, and request IDs where available for investigation. Do not claim a fixed arrival deadline: the documented stream can be delayed, duplicated, reordered, or incomplete. A single successful observation establishes that path worked for that test, not full coverage.
Disabling polling does not recall queued events
If the immediate goal is to stop business actions, pause your application’s comment-action worker first. Then apply enable_comment: false and the required reconnect for an online account. Continue safely accepting any already queued deliveries under your retention policy; the documentation explicitly says disabling does not retract them.
A local worker pause is an application safeguard, not a new API field. Keep unrelated DM processing independent if it should continue. Avoid interpreting an event arriving after disablement as automatic proof that the configuration failed.
If you later enable replies, use the separate top-level comment reply workflow. Missing parent_comment_id does not prove a replyable top-level target, and a notification is not a guaranteed receipt for a previous reply.
FAQ
Does subscribing to post.comment.created enable polling?
No. The webhook filter and provider_data.enable_comment are separate controls. The connection must also apply the configuration.
Does running mean comment monitoring works?
No. It describes connection state. Confirm configuration, routing, and a controlled observation separately.
Does no event mean there are no new comments?
No. Notifications are best-effort and may be omitted. Do not present silence as a complete comment count or archive.
Can I use this event as a DM auto-reply trigger?
No. It is a comment notification, not message.received. Verify the post and comment target before any separate public reply action.
Next step and sources
Start with the post.comment.created reference and record which diagnostic boundary remains unverified before changing production configuration.
Official references checked on 2026-11-14:
Turn messaging integration into a stable product pipeline.
Start by sending through one API, then bring every inbound message back into your business system with standard events.