Seven questions your regulator will ask your collections team first.

The problem. When a conduct review lands, it does not open with your policy manual. It opens with a complaint, a name, and a request: show us everything that happened on this account. Every review that goes badly goes badly at that exact moment, because the answer lives across an agency's dialler logs, three spreadsheets, a CRM and someone's memory.

Why policy documents do not save you. A beautifully written contact policy proves intent, not conduct. The reviewer knows the difference. If your enforcement mechanism is training and hoping, the honest answer to "how do you know the policy was followed on this account" is that you do not. Vendors promising a compliance module bolted onto a dialler have the same gap: the log records what happened, nothing prevented what should not have.

The seven questions, in the order they come:

• 1. Show me every contact attempt on this account. Date, time, channel, outcome, in one timeline. Assembling this from multiple systems in a conference room is where credibility dies.
• 2. How was identity verified before details were disclosed? Per conversation, with evidence, including calls made by AI or by your agency.
• 3. Were required disclosures delivered? Prove it. A script that contains the disclosure is not proof it was spoken.
• 4. Why was this account contacted at this time, this often? Contact windows and frequency caps, and the mechanism that enforces them. "The system cannot place a call outside the window" is a complete answer. "Agents are trained not to" is not.
• 5. What happened when the customer disputed? The expected answer: contact froze immediately, the dispute entered a managed queue, and here is the resolution trail.
• 6. How was this offer approved? Eligibility rules, checked at decision time, logged with the decision.
• 7. Who had access to this customer's data? Role-based access with an audit log, including every third party on your panel.

The structural answer. Teams that pass these reviews share one property: the evidence is generated by the system as a by-product of operating, not reconstructed afterwards. Conduct rules live in software, so violations are prevented rather than detected. Every decision carries its reason. One query produces the account timeline the reviewer asked for, in minutes, including everything your agencies did.

If reconstructing one account takes your team more than an hour today, that is the finding before the review even starts.