Call QA Audit — Human or AI Receptionist
Whether calls are answered by your team or an AI voice agent, this audit scores every pasted transcript against a 6-point rubric, quotes the exact moments bookings were lost, prices the missed revenue, and delivers a rewritten script — or a paste-ready prompt patch for your voice agent.
A caller rings, gets answered, and doesn't book. Somewhere in that conversation is the reason, and it is usually still sitting in the transcript. The Call QA Audit is A$99 one-off: paste in your call transcripts and each one comes back scored against a six-point rubric, with the exact lines quoted where the booking slipped away, a dollar figure set against those missed leads, and a rewritten script — or, if a voice agent took the call, a paste-ready prompt patch. Turnaround is around 15 minutes, delivered on-page with a Markdown download that prints to PDF.
Scored call by call, then read across the batch
Every transcript you paste gets its own scorecard against the same six dimensions, which is what makes one rough Tuesday distinguishable from a habit. Sitting above that per-call detail are three to five systemic pattern findings — the things going wrong repeatedly rather than once, which are the only ones worth rewriting a script over. A score here is an AI judgement of a conversation rather than a measurement, so the quoted evidence underneath a low score is the part actually worth reading.
That missed-lead analysis is where it gets specific. Rather than noting that calls could be handled better, it quotes the moment — the qualifying question nobody asked, the objection that ended the conversation, whatever your transcripts genuinely show. It works from words and nothing else: it reads the text you paste, so it never connects to your phone system, dials anything, or listens to recordings, and tone, pace, interruptions and dead air stay outside what it can see. A garbled auto-transcript produces a shakier read for the same reason.
A rewritten script, or a patch for your voice agent
Who takes these calls is one of the things the intake asks, and the answer changes the shape of the final deliverable. A human team gets a rewritten script. An AI voice agent gets a prompt patch instead, written as plain text to be pasted straight into wherever your agent's instructions live. The audit does not log into your voice platform or reconfigure anything, so you stay the one deciding what goes live — which also means reading the patch before you paste it.
The other thing the intake wants is what a booked call is worth to you, and that figure does real work. It turns "the caller hung up" into a number you can rank fixes against, and it is why the booking-rate improvement plan comes back ordered by revenue rather than by how easy each fix looks. Treat the value-lost totals as your own estimate applied to the leads the audit flags, not measured lost revenue, and treat a rewritten script as a better starting position rather than a guaranteed lift in bookings. Three free AI revisions are included, so if the rewrite does not sound like your business, or the patch needs tightening, send it back.
What you get
- Per-call scorecard across 6 dimensions
- Missed-lead analysis with quoted evidence + value lost
- 3–5 systemic pattern findings
- Rewritten script or paste-ready voice-agent prompt patch
- Booking-rate improvement plan ranked by revenue
- 3 free AI revisions included
Turnaround: ~15 minutes. Delivery: On-page + Markdown download (print to PDF).
Frequently asked questions
I have call recordings, not transcripts. Can I send audio?
No — the intake asks for pasted text, so recordings need transcribing first. Many phone systems and AI receptionists already produce transcripts, and any transcription tool covers the gap otherwise. The audit works from the text it is given and nothing else.
How many calls make this worth running?
One transcript gets you one scorecard, so even a single call returns something. The systemic pattern findings need a spread to work with, though — a handful of calls, ideally mixing ones that booked with ones that didn't, gives the analysis something to compare. The booked calls matter as much as the lost ones, because they show what your good version already sounds like.
Will the prompt patch fit the AI receptionist platform we use?
It arrives as plain prompt text meant to be pasted into your agent's instructions, so it travels to any platform that lets you edit the system prompt. It does not integrate with, log into, or test against your platform, and it is worth reading before it goes live — you know your booking rules better than a transcript does.
Can I see the six dimensions before I buy?
They aren't published on this page. What matters for the output is that the same six are applied to every transcript in the batch, which is what keeps the scores comparable from call to call rather than each one being judged on its own terms.