Quality and practice

Seeing what works in your conversations, spotting what goes wrong, and letting a team rehearse on cases that look like the real ones.

7 days, 20 answered calls, no card

The task it takes over

A receptionist handling hundreds of calls a month builds habits — good ones and others. With nobody looking, one clumsy phrasing repeats for months and nobody knows what it cost.

This add-on looks at explicitly chosen conversations, scores them against a versioned grid, and shows the excerpts behind a score beside the business outcome reached. You can annotate, correct and contest an evaluation: it is your business, and an automatic score does not get the last word.

It also serves practice. Voice or text simulations replay difficult customer profiles — somebody wanting a price immediately, somebody telling everything except the essential, somebody changing their mind three times — with feedback and progression. It is an exercise, rather than an evaluation of a person.

Step by step

  1. Choose

    An explicit sample

    Analysed conversations are selected deliberately. Nothing gets scored in bulk behind your back, and the usage of an analysis is estimated before it starts.

  2. Score

    A grid, a version, excerpts

    A score comes from a versioned grid. The detail shows the excerpts explaining it and the business outcome of the conversation, rather than a number floating alone.

  3. Compare

    Two versions on the same cases

    Two configuration versions can be compared on one set of conversations. That is what lets somebody say a change improved something, instead of assuming it.

  4. Practise

    Simulations with feedback

    Voice or text exercises replay customer profiles. Feedback is immediate, progression is tracked, and none of it reaches a real customer.

An add-on, above an already tested base

Checks on your configuration and a first simulation are included in your base.

Analysis is a tool; the conclusion is yours.

What this changes, in practice

Clinics & home care

What matters is administrative accuracy and a limit held: one misstated instruction repeats hundreds of times. Evaluation covers the conversation, not the patient.

Contractors & trades

Calls ending in a visit and calls fading out look alike at first glance. Comparison shows what the first ones asked and the others left out.

Law & accounting firms

What matters is whatever went unsaid: an eligibility question left open rather than settled. A grid written for that trade detects exactly that.

Salons, gyms, classes & shops

Conversations resemble each other week to week: price, availability, duration. That is the ideal corpus for seeing which phrasing books an appointment and which ends a call.

What it receives

  • Conversations you choose to analyse
  • The scoring grid, with its version
  • Customer profiles used for exercises
  • Your annotations and contestations

What it produces

  • A detailed evaluation, with excerpts and business outcome
  • A comparison of two versions on one set of cases
  • Exercises with feedback and progression
  • A record of corrections you asked for

You choose what gets analysed and may contest every score. An evaluation triggers no automatic consequence.

What is counted, and how

The add-on carries its own monthly volume of analysis and practice, which you follow in your app, and it can be tried during the free trial.

  • An evaluated audio conversation is counted by its length
  • Text alone is counted by its length
  • Where audio and transcript are analysed together, the maximum is kept rather than the sum
  • An analysis rerun on the same content, with the same grid, does not count twice

Analysing a past call and running a simulation do not count as an extra call: the existing transcript is reused.

Questions about this add-on

Is every conversation scored?

No. Analysed conversations are selected deliberately: nothing gets scored in bulk behind your back, and the usage of an analysis is estimated before it starts.

Can I contest a score?

Yes. You can annotate, correct and contest an evaluation: an automatic score does not get the last word, and an evaluation triggers no automatic consequence.

Do the simulations speak to real customers?

No. Voice or text exercises replay difficult customer profiles, with immediate feedback and tracked progression, and none of it reaches a real customer. It is an exercise, rather than an evaluation of a person.

How is analysis counted?

An evaluated audio conversation is counted by its length, and text alone by its length. Where audio and transcript are analysed together the maximum is kept rather than the sum, and an analysis rerun on the same content, with the same grid, does not count twice. You follow that volume in your app.

Where to next

Revisit your instructions, or our measures and their definitions.

Quality and practice

Seeing what works in your conversations, spotting what goes wrong, and letting a team rehearse on cases that look like the real ones.

Start your free trial7 days, 20 answered calls, no card