Skip to content
VoIPThug — homeBuild My Stack

Call evidence

Every call is evidence. Most floors just throw it away.

Recording a call is trivial. Being able to find the eleven seconds that matter, six weeks later, in a language you do not speak, and prove what was said — that is the part that decides disputes, coaching and whether your compliance story survives contact with a regulator or a bank.

  • Waveform recordings with synchronised transcripts
  • Translation and search by spoken words
  • Ask-a-Call answers that link back to the timestamp
  • Editable QA rubrics with evidence-backed scores

Hear

Conversation memory, not a folder of audio files

A recording you cannot navigate is an archive. A recording with a transcript running beside the waveform is a document.

  • Waveform and transcript in step

    Scrub the waveform and the transcript follows; click a line of transcript and the audio jumps there. Dead air and crosstalk are visible before you press play.

  • Translation

    Multilingual floors get transcripts in the language spoken and a translation for the reviewer who does not speak it, so QA does not stop at the language barrier.

  • Search by spoken words

    Find the calls where a phrase was actually said — a promise, a price, a disclosure, a competitor's name — across the whole account rather than in the one call somebody remembered.

  • Summaries, outcomes and tags

    Each call carries its summary, disposition, tags, talk ratio and dead-air measure, so a list of two thousand calls is something you can filter instead of something you dread.

Recordings and transcripts
RecordingLengthTranscriptReview
2026-08-15 14:0206:41ReadyPassed
2026-08-15 13:4702:18ReadyFlagged
2026-08-15 13:3111:05ReadyPassed
2026-08-15 12:5804:52ReadyNot reviewed
Illustrative view with sample data.

Find

Ask a call what happened

Ask-a-Call answers questions about a specific conversation and links every answer back to the timestamp it came from. The citation is the point. An answer you cannot check is a rumour with better formatting.

Ask whether the disclosure was read, what objection the customer actually raised, what the agent promised about timing, or whether a price was quoted — and get an answer with the moment attached, so a supervisor can press play and see for themselves in four seconds instead of forty minutes.

Prove

QA rubrics, disclosures and prohibited phrases

Compliance on a sales floor fails in one of two ways: nobody checks, or somebody checks and cannot show their working. Both are solved by the same thing — a score attached to the evidence it came from.

  • Editable rubrics

    Build the scorecard your sector and your risk appetite actually need, rather than living with somebody else's idea of a good call. Rubrics are yours to edit as the rules change.

  • Customer scripts

    Hold the approved script alongside the call, so 'did they follow it' is a question with an answer instead of an argument.

  • Required disclosures and prohibited phrases

    Flag calls where a required disclosure is missing or a phrase you have banned appears. Automated review proposes; an authorised manager confirms anything with a consequence.

  • QA coverage

    Track what proportion of calls, agents and campaigns have actually been reviewed. Coverage is the number that tells you whether your QA programme is real.

Improve

From evidence to the number the owner cares about

Productivity, conversion and revenue analytics run off the same call records as the QA scores, which means the coaching conversation and the revenue conversation are finally about the same calls.

  • Productivity

    Talk time, connect behaviour, wrap, dead air and outcome mix per agent, team and campaign — the operational picture behind the leaderboard.

  • Conversion and revenue

    Outcomes through to first-time deposit and redeposit confirmation, with revenue attribution by campaign, team and agent. Confirmation of a consequential outcome stays human-authorised.

  • Heatmaps and goals

    When your market answers, when it converts, and how the floor is tracking against goals — so shift planning stops being folklore.

  • Coaching that closes the loop

    AI text coaching drafts the feedback from the actual call; the coaching queue makes it a task; the next review shows whether it worked.

Questions

Straight answers

Are the AI outputs trusted automatically?
No. Automated review, drafted scores and suggested coaching are proposals. Anything consequential — a score that goes on an agent's record, a confirmed deposit outcome — is confirmed by an authorised manager. That boundary is deliberate and it is not configurable away.
Which languages are supported?
Transcription and translation cover the languages our multilingual floors actually run in, and the set is not identical for every language pair. Tell us the languages on your floor and we will confirm exactly what transcription and translation quality looks like for those before you commit to a QA programme around them.
Can I export recordings and transcripts?
Yes. Recordings and call records are available through the portal and the API, including cost and recording endpoints, so evidence can live in your own systems as well as ours.
How long are recordings kept?
Retention is set per account, because a floor with a two-year dispute window and a floor with a data-minimisation policy want opposite things. Agree it during setup rather than discovering the default when you need a call from last March.

Prove it

Your recordings are already evidence. Start using them like it.

Tell us the languages on your floor, the rubric you score against and the retention your legal position needs, and we will tell you exactly what review and coverage look like on your calls.