When a rule goes quiet.

Checks

Work down the list before you rewrite the message.

• The Instagram account still shows as connected. Reconnect if Meta expired the login.
• The comment uses the keyword, not only an emoji.
• A next-post rule stays quiet until you open Automations after you publish, or until a later comment attaches it to the earliest new post.
• The inbox does not show WINDOW_LAPSED for that person.
• The free plan has not passed 1,000 DMs this month.
• Another messaging app is not the thread owner. See error 2534037.

Find the first missing step in the chain

Treat a quiet rule as a sequence to inspect: the account receives an event, the rule matches it, a reply is prepared, and the platform accepts or rejects the send. These are different stages. If no event matches, changing the DM text will not help. If a send is rejected, adding more keywords will not help. Record the account, post or story, keyword, test time, and whether a public reply appeared. That small record makes the next check much more precise.

A workflow to try

Test one new comment from an independent account on the intended post. Verify the caption word against the saved keyword and ensure the automation is enabled on the correct connected account. If Watch is Next post or Reel, open Automations after you publish. The rule stays quiet until DripDM attaches it to the earliest new post. Look for the public reply and then the private message, including message requests. Next, send a direct keyword DM as a separate test. If inbound DMs work but comment-led replies do not, investigate comment eligibility or rule scope rather than immediately reconnecting every account in the workspace.

Checks before you change the setup

A failure limited to one person can reflect recipient settings or a spent private reply. A failure limited to later messages can reflect a closed window. A failure across all fresh events can point to connection, permission, quota, routing, or platform problems. Preserve the error code and subcode when available. Avoid repeated retries against the same comment, and do not delete a working rule before you have saved its settings and isolated the cause.

Use an incident note instead of a cascade of edits

Write one short incident record: what used to work, the last observed success, the first observed failure, and changes made between them. Include account reconnections, caption edits, new integrations, plan state, expired creative, and the Watch choice. A next-post rule stays quiet until you open Automations after you publish, or until a later comment attaches it. Keep a screenshot or exact text of the visible error without exposing credentials. Then test one entry path at a time and record the result. If the first public response still works, that evidence narrows the investigation. If a different fresh recipient works, that narrows it again. Rewriting every rule at once removes the comparison you need to identify the first broken step.