Voice campaign audit records: Why attempted calls are not completed conversations

An audit-ready campaign record can show that a call was attempted, which events occurred, and whether the recipient opted out. It can only call that attempt a completed conversation when the recorded evidence meets a clear, pre-agreed definition of a conversation.

In 1999, NASA’s Mars Climate Orbiter approached Mars for orbit insertion and then stopped communicating with ground controllers. The investigation found that a navigation error arose because one part of the system used pound-seconds while another expected newton-seconds. NASA’s Mars Climate Orbiter Mishap Investigation Board documented the loss later that year.

The issue was not a lack of numbers. There was telemetry, calculations, software, and a defined mission sequence. The failure came from treating measurements as compatible when they represented different things.

A campaign dashboard can make the same mistake in miniature. An attempted dial, a provider event, an answered call, speech detected, a completed assistant turn, and a recipient opt-out are all meaningful records. They do not prove the same outcome.

Separate call evidence from campaign outcomes

A campaign manager reviewing activity should be able to point to a call record and answer basic questions without filling gaps from memory:

• Was a call attempt created?
• Did the telephony provider report a lifecycle event?
• Did the recipient answer?
• Was speech exchanged with the agent?
• Did the call end?
• Did the recipient opt out?

Those records support an audit trail. They also make it possible to identify what remains unknown.

“Completed” is where teams often compress too much. A provider may report a call ended. That tells you the call lifecycle reached an end state. It does not establish that the intended person heard the message, understood it, spoke with the agent, or completed the campaign’s intended task.

The practical discipline is simple: preserve the underlying events, then define outcome labels separately. A campaign can report “attempted,” “answered,” “assistant interaction detected,” “opt-out recorded,” or another state that reflects evidence actually captured. The label should never outrun the record behind it.

For a closer look at the definition problem, see What Does “Completed” Actually Mean in a Voice Campaign Dashboard?.

Build the review around traceable events

Asenda Talk’s telephony lifecycle webhook pipeline is designed to retain call-truth events rather than replace them with a single flattering status. Its consent, opt-out, and audit trail records give an operator material to review for every call.

That matters most when a recipient challenges the campaign. A manager should be able to retrieve the call record, inspect the event sequence, confirm the consent and opt-out status, and say exactly what the platform recorded. If the available evidence shows an attempted call followed by a terminal provider event, that is the honest description.

This is useful even when the result feels less impressive in a weekly report. A lower completed-conversation count can be more valuable than a larger number built from ambiguous call endings. The first can be defended in an audit. The second creates work for the person asked to explain it.

The same principle applies to language. A Twi or English interaction may require a definition beyond “audio was present.” For an outreach campaign, a completed conversation might require an answered call, an agent disclosure, and a recipient response. For support, it might require an identified issue and a recorded handoff or resolution state. Those definitions belong in the campaign plan before calls begin.

Keep opt-outs visible, even when they change the report

An opt-out is not a failed data point to hide. It is a recipient instruction and a compliance event.

When an individual asks to stop receiving calls, the campaign record should retain that request and prevent it from becoming an unexplained gap in the next contact list. The audit review should show when it was captured, which call it relates to, and how it affects future calling.

This is especially important for outbound campaigns, where a manager may be under pressure to report reach and completion. A dashboard that counts every attempted call as a conversation can conceal both poor contact rates and opt-outs that should change the campaign’s next step.

Asenda Talk is in active early access. Its outbound calling remains gated behind an explicit telephony-provider decision that has not been made live. That limitation should stay visible in planning and reporting. A campaign should not claim operational outbound results before the relevant provider path, controls, and evidence are live.

Define “completed” before the first call

Before a campaign runs, write one sentence that states what counts as a completed conversation. Then list the event evidence required to assign that label.

For example, a team might require:

• A provider record that the recipient answered.
• An assistant interaction event showing a response after the opening message.
• No conflicting event that indicates an immediate disconnect or opt-out.

The precise rule will vary by use case. What matters is that it can be applied consistently and examined later.

NASA’s Mars Climate Orbiter did not fail because its teams had no data. The mission failed because critical data was interpreted through incompatible units. In a campaign review, an attempted call and a completed conversation are also different units. Keep them separate, and the audit record can carry the weight you place on it.