Quality and analytics
Compliance flagging
How conversations are checked against rules your compliance team writes, how required disclosures and approved wording are handled, and how flagged conversations reach a reviewer.
On this page
In a regulated business, the conversations that matter most to your compliance team are the ones where something was said that should not have been, or something required was not said. Telonic checks conversations against rules your compliance team writes, and flags any conversation that breaks one for a person to review. A problem is found while it can still be put right, not months later at an audit. This page covers the rules, the disclosures and approved wording that prevent problems in the first place, and the review queue where flagged conversations are handled.
Prevention first, then detection
Most of what keeps a conversation within your rules happens before anything is flagged. The agent answers only from your approved sources, does not discuss topics you place out of scope, and cannot take actions it has not been granted. Compliance flagging is the second layer: a check on every conversation afterwards, against rules written in your own terms.
| Layer | What it does | Where it is described |
|---|---|---|
| Approved sources only | Answers come from your systems and approved documents, not from general knowledge | Knowledge and documents |
| Topics out of scope | Subjects such as legal or investment advice are declined and routed to a person | Policies as limits |
| Required disclosures | Wording you are required to say is said at the defined point in the conversation | This page |
| Approved responses | Answers your compliance team must control are given word for word | This page |
| Compliance flagging | Every conversation is checked against your rules afterwards, and any breach goes to review | This page |
The rules your compliance team writes
Compliance rules are configured during implementation, with your compliance, legal and operations teams. Each rule describes something that should never happen in a conversation, or something that must always happen. Like quality criteria, a good rule is observable: a reviewer reading the conversation can say whether it was broken.
| Industry | Example rule | Why it matters |
|---|---|---|
| Insurance | Did not give advice or opinion beyond the approved policy wording | Advice outside approved wording can create an obligation the insurer did not intend |
| Insurance | Did not suggest whether a claim will be accepted or how much will be paid | Claims decisions stay with the claims team |
| Property | Did not give or confirm a handover date that is not in the project schedule | A promised date is revisited years later, often in a dispute |
| Property | Did not discuss changes to a payment plan | Payment plan changes stay with the collections team |
| Travel and hospitality | Did not promise compensation or a refund outside policy | Commitments outside policy stay with your team |
| Any | Gave the recording notice before recording began | Recording consent is required before a call is recorded |
Each rule has a severity, so a missed product disclosure and a promise outside policy can be handled by different reviewers on different timescales.
Required disclosures
Some wording must be said, in a set form, at a set point. Recording notices, product disclosures, and statements about how a customer can complain are common examples. Required disclosure playback is configured during implementation for the workflows that need it: the agent says the mandatory wording at the defined point, in the customer's language, and the conversation record shows that it was said.
Where a conversation ends before a disclosure point is reached, for example because the customer hangs up, the record shows that too. Your compliance rules decide whether that conversation is flagged.
Approved responses
For some questions, your compliance team will want the answer word for word, every time. Approved response libraries are configured during implementation for those questions: the agent gives the approved answer exactly as written, with only the variables that must come from your systems filled in, such as a policy number or a date. Changes to an approved response go through the same approval and testing as any other change. See How changes go live.
The review queue
Flagged conversations appear in a review queue in the Quality view of the console, alongside conversations flagged by quality review. Each flag shows the rule broken, its severity, the part of the conversation it relates to and the reason it was flagged. The reviewer opens the full conversation, with the transcript, the recording where the customer consented to one, and the decision log showing what the agent looked up and why.
The reviewer records a finding and what happens next: no issue, a follow-up with the customer, a correction to a source document or rule, or referral inside your organisation. Every flag and every finding is kept in the audit trail. See Audit trail and decision records.
Compliance flagging helps your compliance team find and act on problems. It does not decide whether your organisation meets its regulatory obligations. That judgement stays with your compliance and legal teams.
Conversations your team handles
Where conversations held by your own team are captured into the customer record, which is configured during implementation, the same rules can be applied to them. Your compliance team then sees one set of findings across the agent and your people. See Conversations your team handles.
In practice
A health insurer deploys the agent for network and coverage questions on one product.
- Its compliance team writes nine rules, including "Did not say whether a treatment will be approved" and "Did not recommend a hospital outside the approved network list".
- The question "Is physiotherapy covered?" is given an approved response, word for word, with the plan's session limit filled from the policy system.
- In the first week, a flag appears: a member asked whether an MRI would be approved, and the agent said pre-approval was "usually quick".
- Salma, a compliance reviewer, opens the conversation. The statement did not promise approval, but it gave an expectation the insurer does not want to set. She records the finding and asks for a rule and an approved response for pre-approval timing questions.
- Both are tested and approved before they go live, and the change is recorded against the rule set.
What your team controls
- Every compliance rule, its wording and its severity.
- Which disclosures are required, their wording in each language, and where they are said.
- Which questions have an approved response, and its exact wording.
- Who reviews flags, and who can see findings, by role.
Related
- Quality review on every conversationQuality and analytics
- Audit trail and decision recordsGovernance and control
- Policies as limitsAgents
- Knowledge and documentsAgents
- Call recording and consentLegal and compliance
Product names and logos are trademarks of their owners. Their mention shows systems Telonic connects to and does not imply partnership or endorsement.