A forum reply suggests deleting the connection and starting again. You have already tried two other fixes, and nobody has explained what deleting it would remove.
That is a good point to stop. Ask for the supported procedure before losing the settings—or the evidence of what went wrong. The same applies when the next step needs authority you do not have or information only the provider can see.
Three reasons to stop
You are not authorized to make the next change. Being able to inspect a dashboard does not necessarily mean you can change billing, security or customer permissions.
If a proposed fix requires a role you do not have, do not create another account, borrow credentials or promote your own access. Record the specific missing authority and route the question through the person responsible for that system.
For a fictional team, an editor can document a private preview problem. The editor cannot assume that changing organization-wide sharing is an acceptable workaround. A narrow symptom does not authorize a broad correction.
You cannot explain how to undo it
A suggestion to delete and reconnect an integration may sound routine. Its effects could include lost settings, resent events or altered access. Before trying it, identify what would be removed and how the original state would be restored.
If you cannot establish that scope, ask the vendor for the supported procedure. Preserve the current configuration and relevant redacted evidence. Do not destroy the only available example of the failure merely to produce a cleaner dashboard.
The same principle applies to repeating operations. A timeout does not prove that an action failed. Repeating a payment, message send or access grant can have consequences even when the screen appears unchanged.
The answer is inside the provider’s system
You may reach a boundary between two systems where your account shows receipt but not processing. Another local setting change will not necessarily reveal what happened inside the provider’s service.
Identify the exact missing fact: whether a request was accepted, which plan capability applies, what a control affects or whether a known incident explains the symptom. Ask one focused question rather than requesting a general account reset.
A current official status notice can be relevant, but it does not automatically prove the cause of your particular incident. Keep your own observed evidence separate from the provider’s general information.
Decide who needs to take the next step
| Situation | Appropriate next move |
|---|---|
| Safe, reversible check within your authority | Continue with one defined change |
| Scope or rollback uncertain | Hold the change and request a supported procedure |
| Payment, security or customer entitlement affected | Escalate before testing |
| Provider-side processing evidence unavailable | Ask the vendor about the exact event boundary |
| Suggested fix already failed once | Preserve the result; do not repeat without a new reason |
| Suspected compromise or active data exposure | Follow the organization’s security response route |
This card does not replace a vendor’s incident procedure. It helps you recognize when ordinary self-help has reached its limit.
Ask one question support can answer
Record the symptom, account context, nonsecret event reference, timestamp, safe tests already performed and the next question that needs an answer. Include the current state: held, restored or still affected.
For the connection example, ask: “What does deleting this connection remove, could reconnecting repeat earlier actions, and what supported recovery method preserves the existing settings?” Add the safe event reference and observed symptom through the verified private support route. Do not change that into “your system lost the event” unless the evidence establishes it.
Write the stopping reason and the exact question that would let you continue. Keep the current state clear: still affected, safely restored or waiting for instructions. You have finished the part you can responsibly do. When support replies, check whether the answer actually resolves that question before trying another change.