Decide how recording is permitted before it starts.

Tell the reviewer where the operator and called person may be located, what the campaign concerns and why the business wants a recording. Do not assume that only the called person's state controls every situation, or that an area code establishes the person's current location.

Identify the channels and data involved: a live conversation, available audio, a transcript, an AI score and a summary. The derived text can repeat sensitive details even when the original goal was simply to help an operator remember a callback.

State any required retention, access or industry-specific process before configuring the account. A plan that includes recordings does not supply a legal opinion, a fixed storage period or an approved policy.

The reviewer determines the process that the campaign must follow. The product team establishes whether the actual configuration can support that process. Both answers are necessary.

Do not assume a per-call recording toggle, automatic jurisdiction detection or a campaign-specific consent prompt exists because such a control would be convenient. Ask to inspect the supported behavior, including when capture starts and what happens when it must not continue.

If the required process cannot be demonstrated, keep recording or the affected workflow out of use until the issue is resolved. A disclosure script cannot compensate for a capture mechanism that starts at an inappropriate time under the approved policy.

Have the responsible reviewer approve the introduction or permission request and the point at which it is delivered. This guide does not offer one universal sentence that makes every recording permissible.

Continued conversation should not be treated as a universal consent mechanism. Follow the actual policy for the relevant circumstances, including what evidence of consent must be retained and whether recording can start before that evidence is obtained.

Train operators on the meaning of the process, not only the words. They should know whether a refusal, uncertainty, a transfer or a newly joined person changes what they may do. The escalation path should be clear before the first real call.

1. Use synthetic records and the supported sandbox fixtures during the trial. No real person should receive a trial call.
2. Inspect when the recording mechanism begins and ends, to the extent the test environment exposes it.
3. Confirm the supported response to the policy's consent, refusal and uncertainty cases.
4. Inspect the resulting record and any derived AI output separately.
5. Document what the test establishes and what still requires a separate, approved production acceptance check.

Do not create an unlawful recording merely to prove that a consent process works. The responsible reviewer should approve the test and the evidence to retain. The existence of an audio file is not the only possible form of consent evidence, and this page does not prescribe one.

A person objects: follow the approved response and supported controls; do not continue recording on the assumption that finishing the sales pitch is more important. The applicable policy is unclear: pause or avoid the recording workflow and escalate rather than improvise a legal conclusion.

Expected audio is missing: investigate the record and capture path without inventing a cause. Missing audio does not prove that no recording was attempted or that the policy was followed. The transcript includes unnecessary sensitive information: handle the derived record under the agreed access and data-minimization process.

Keep a concise incident reference and avoid copying complete recordings or personal details into broad support channels. Arrange an appropriate sharing method when source material is needed to investigate.

Confirm who may listen, download, share or use derived output in the actual account. Do not assume that operator and manager labels enforce the desired boundaries without testing them.

Ask the setup team to confirm the current storage, retention, export and deletion arrangements before uploading production data. The privacy notice describes categories and unresolved procurement questions; it does not invent a data-processing agreement or automatic deletion schedule.

For coaching, use only permitted recordings and share the minimum context needed. An AI summary does not establish that a call was lawfully recorded or that a particular statement was captured accurately.

The ready state includes an accountable reviewer, an approved policy, a supported configuration, trained operators and a documented response to refusal or uncertainty. It also includes a clear record of any requirement the product setup has not established.

Review the process when the campaign, location, data use or configuration changes. The recording feature guide explains source review; the permission decision belongs to the qualified reviewer for the actual campaign.