Ontopix Audit
Know what was resolved, what wasn't, and why
Monitor, evaluate, learn, improve, and decide
Audit evaluates your conversations against your criteria and links every rating to the original conversation. The same whether a person or an AI agent answered.
The problem
What is getting resolved?
Today quality is decided from a few monitored calls a month and the complaints that come in. Of the rest, you know they happened, not how they ended.
What it costs
What you don't see keeps happening
- Cases closed without a next step that come back as new contacts.
- Mandatory disclosures that are given some of the time.
- Training and script changes decided without data.
Our solution
From a few spot checks to knowing what was resolved
What it reviews
The conversations you want to audit, not just the ones there is time to listen to. You share recordings or transcripts, or Audit connects to your phone system or your customer support software and collects only those.
How it evaluates them
We agree on the criteria with you. Every evaluation shows up on a dashboard with its evidence, linked to the original conversation, and you receive reports. The same whether a person or an AI agent answered.
Example
Audit › Reports › YOUR OPERATION
Mockup — example operation
214 interactions audited
6.84/10
- Email · “still no answer about my bill”
- Chat · issue closed · the customer writes again
- Phone · account holder change
- Email · issue closed · the customer writes again
- Web · rate inquiry
- Chat · meter reading
- Phone · issue closed · the customer calls again
- Email · new sign-up
- Chat · power capacity change
3 closures with no next step explained. “The next step was explained”: 0.4/2 in the conversations that come back. Open each one and decide what to change.
Audit › Reports › YOUR OPERATION
Your operation
Energy · 214 interactions audited
- Score
- 6.84/10
- Category
- 7.40
- Overall
- 7.97
- 7.50
- 55 tests
- CHAT
- 3.53
- 12 tests
- PHONE
- 7.58
- 135 tests
- WEB
- 9.88
- 6 tests
SUMMARY BY BLOCK · WEIGHTED COMPLIANCE
- Availability and Accessibility Management54.27%
- Relational Flow of the Conversation85.63%
- Resolution and Customer Experience69.49%
| ID | CRITERION | TESTS | POINTS | PENALT. | % ERROR |
|---|---|---|---|---|---|
| C1 | Response Time | 55 | 0.87/2 | 42 | 76.36% |
| C2 | Acknowledgment of Receipt | 49 | 1.00/2 | 49 | 100.00% |
| C4 | Sender | 49 | 2.00/2 | 0 | 0.00% |
| C5 | Subject | 49 | 0.53/2 | 36 | 73.47% |
Mockup — this is the real Ontopix Audit screen, with an example operation.
Why Ontopix
An evaluation you can argue with
Your criteria
You choose what is evaluated and how much it weighs. You change it when your operation changes.
Verbatim evidence
Every rating quotes the original conversation. If you disagree, you open it.
One yardstick
People and AI agents are measured against the same criteria.
Configurable per company
You set the criteria
What counts as an excellent call for a clinic is a compliance breach for a lender. That's why we don't ship ready-made criteria: you decide what gets evaluated and how much each thing weighs.
| Criterion | Weight |
|---|---|
| Resolved on first contact | 25% |
| Legally required disclosures were given | 20% |
| Listening and courtesy | 15% |
| Wait time | 15% |
| The next step was explained | 15% |
| Identity was verified | 10% |
You can add criteria specific to your operation and change them whenever you want.
Mockup — example criteria and weights.
Audit works with the recordings or transcripts you share, or connects to your phone system or your customer support software to collect only the conversations being audited. What gets connected and how is tailored to you.
How to start