How to check compliance and process adherence in support conversations

• Write each rule as a check with a plain-language definition, a clear fail and worked examples.
• Keep compliance checks (identity verification, privacy handling, required disclosures) separate from process checks (escalations, policy steps).
• Mark checks as not applicable when they don’t fit a conversation, so they don’t distort the score.
• Grade the conversations you connect, then review each miss with its evidence.
• Break misses down by contact reason, BPO vendor, in-house team, manager and agent to find the pattern.

Compliance checks cover the rules your business must follow, such as verifying identity before discussing an account, handling personal data properly or giving a required disclosure. Process adherence covers the steps your own playbook requires, such as escalating when a case meets your criteria or following your refund policy. You grade both the same way: as pass/fail checks against a definition your team writes.

If you’re new to support QA as a discipline, start with what support QA is and how it works.

Sample-based review misses compliance gaps because a rule that’s broken in a small share of conversations rarely shows up in a small, hand-picked review queue. A skipped disclosure can repeat many times before anyone samples a conversation where it happened.

Regulators expect evidence, too. In its July 2026 review of outcomes monitoring under the Consumer Duty, the UK Financial Conduct Authority (FCA) wrote: “Collecting data, listing metrics or reporting management information (MI) will not, by itself, show whether customers are receiving good outcomes” ². The same review adds: “Firms using third parties to deliver part of their services are still responsible for their customers’ outcomes” ². That applies to UK financial services firms, and it’s a useful standard for any team that outsources support to a BPO.

Start with the policy or SOP itself and rewrite each requirement as a check with three parts: what it means, what counts as a fail, and worked examples of the edge cases. A check that two reviewers would grade differently isn’t ready yet.

Here’s what a finished check looks like:

• Check: Identity verified before account details are shared.
• Fail: The agent shares a balance, order details or an address before the customer completes verification.
• Edge case: The customer volunteers their account details unprompted. The agent still completes verification before confirming anything.

In unitQ Support QA, you define each quality check in plain language: what it means, which answers count as a fail, and worked examples that show the edge cases. Your team owns and edits the rubric in Tag Studio, the rubric editor in unitQ Support QA. You can attach your policy or SOP document as reference material and refine a check with AI.

Not sure where to start? unitQ Support QA can suggest draft checks from a sample of your own conversations, and your existing QA guidelines if you add them. Your team reviews, edits and approves every check before it’s used. For more on structuring the full rubric, see our guide to support QA scorecards.

Mark a check as not applicable when it doesn’t fit a conversation. A disclosure that’s required only for cancellations shouldn’t count against a password-reset chat, and a scorecard that fails it there will hide the real misses in noise.

In unitQ Support QA, checks that don’t apply to a conversation are left out of the score. You can also weight rubric groups to match how you score quality, so a missed identity check can carry more weight than a missed sign-off.

Grade the conversations you connect against your checks. That way a rare miss has the same chance to surface as a common one.

unitQ Support QA supports all your support channels, including chat, email, messaging, voice and IVR (automated phone menus), through integrations with platforms such as Zendesk, Salesforce Service Cloud, Intercom, Freshdesk, Kustomer, Amazon Connect and Five9. It grades the compliance and process checks you define, such as identity verification, privacy handling or a required disclosure, as pass/fail on the conversations you connect. It flags missed steps that show up in the conversation, such as a skipped identity check, a required disclosure that wasn’t given, or an escalation that should have happened, checked against the process you define ¹.

Before a new check counts toward anyone’s score, run it on past conversations and read the results. You can try a check on real past conversations before it goes live, using Testbench in unitQ Support QA. Edit and refine checks as often as you need before the changes go live.

A compliance result is only useful when a reviewer can open the conversation behind each fail and see why it failed. Without that link, a low score is hard to act on and easy to argue with.

In unitQ Support QA, every scored miss links back to the conversation. unitQ Support QA gives you one quality score built from your rubric, with competency, outcome, operations and compliance views beside it, in the Quality report.

Keep a record of what you decided about each pattern and what you changed. The FCA’s review notes that stronger examples included “a clear audit trail showing how MI translated into decisions, actions and improved outcomes” ².

Break misses down by contact reason, team, vendor and agent to see whether a gap belongs to one person, one queue or one process. When a disclosure is missed across a whole contact reason, look at training, macros and the process itself before you coach individual agents.

unitQ Support QA breaks scores down by contact reason, line of business, BPO vendor vs in-house team, region, manager and agent, all on the same scorecard. Small samples are flagged, so you don’t act on a pattern built from a handful of conversations. AI-generated findings come with recommended next actions, each tagged with a suggested owner such as operations, enablement, QA or conversation design.

If an AI agent talks to your customers, audit its conversations too, using criteria built for how bots fail: loops, wrong information and handoffs at the wrong moment.

unitQ Support QA grades human and AI agents side by side in one product, on as many rubrics and scorecards as your policies require, with one shared standard or bot-specific checks, independent of the vendor that built the bot. AI agents get an independent audit either way: is containment real, was the information correct, did the bot loop or dead-end, escalate when it shouldn’t, misunderstand the issue, or resolve it inaccurately, and how often conversations recover after reaching a person ¹. The bot audit is configured for your workspace during setup. For a deeper walkthrough, read what AI agent QA is.

Grade the refund steps that show up in the conversation, then confirm the back-office action in the system that records it. In the conversation, you can check whether the agent confirmed eligibility, explained the policy correctly, gave the timeline your policy states and avoided promising anything outside it. Whether the refund actually posted lives in your billing or order system’s own records, so pair your QA results with those reports.

The right mix depends on your volume, your vendors and how regulated your lines are, and the options can be combined.

• Manual spot checks fit small teams, low volumes and the first weeks of a brand-new policy, when you’re still learning what a fail looks like.
• Automated grading across the conversations you connect fits teams with higher volumes, several BPO vendors or regulated lines where a rare miss still matters.
• A hybrid fits teams that want both: automated grading surfaces the misses, and experienced reviewers handle sensitive cases, edge cases and changes to the checks.