Support quality tools

Support quality tools help a team review customer conversations, give specific feedback and fix recurring problems. Choose a setup by deciding what good support looks like, which conversations to inspect, who will act on findings and how customer results will be checked.

Define what reviewers assess

Use a short scorecard for decisions an agent could reasonably make. Criteria might cover whether the answer addressed the question, preserved an important condition, took an authorised action and gave a clear next step. Define when a criterion does not apply. Keep speed separate: a prompt reply can still be incomplete.

Read the relevant exchange before attributing a result to one agent. A bot, another agent or a channel change may have shaped the case. Record whose contribution is under review and what information they had at the time.

Zendesk QA scorecards are evaluation forms built from rating categories. Admins and workspace managers can view and manage them; the scorecard list shows status, workspaces, categories and last update time. Filters cover workspace, category and status: active, draft or inactive. Settings can use emojis instead of numbers and allow reviewers to skip categories, adding an N/A option to rating scales.

Separate routine sampling from urgent inspection

Select a baseline sample from a defined set of eligible conversations. Keep complaints, low ratings and suspected policy errors in a separately labelled queue for prompt review. Findings from that targeted queue cannot describe all support work.

Before choosing software, settle the period, channels and review unit. Decide whether a shared conversation receives one review or separate reviews for individual contributions. Have reviewers score some of the same cases and discuss disagreements before interpreting a score trend.

NIST guidance defines a simple random sample as one in which every response in the population has an equal chance of being observed. Random sampling does not guarantee each sample is representative, but on average it ensures data collection does not leave behind biases. Probability theory can then quantify how often particular modelling procedures are affected by extreme variation.

Setting up a defensible review sample

Align reviewers with calibration

Calibration has all reviewers rate the same batch of conversations and compare their scores and comments. It aligns reviewers so agents receive consistent feedback regardless of who conducts the review. Admins, workspace leads and managers can perform calibrations within an existing workspace.

In Zendesk QA, calibration scores do not affect the existing Internal Quality Score and do not trigger notifications. Agents do not see conversation-related information in calibration. As a practical responsibility, workspace managers or leads should inform reviewers about the calibration session and its due date. Settings control whether previous reviews are visible and whether reviewers see only their own or peers' evaluations after submitting.

Turn findings into action

A useful review identifies the relevant message or decision, explains the concern and names the next action. Coach an agent on a response they could improve. Send an outdated article, unclear permission or broken procedure to its owner. Keep any customer correction on the active customer case; an internal review does not deliver it.

Store feedback internally with a reliable case reference. Check who can see it, how the agent can respond and how an inaccurate assessment can be corrected.

The TIO Quality Framework sets five pillars: stakeholder and user focus, capability building, quality management, organisational support and ownership, and continuous improvement. It commits to structured quality performance monitoring aimed at building capability and performance and providing clear expectations. It also calls for clear organisational support for quality commitments, including articulated roles and accountabilities.

TIO Quality Framework pillars

Choose the tool for the work

A dedicated QA product can supply scorecards and coaching records. An inbox with internal notes may suffice for occasional case discussions, but notes alone do not establish a formal review workflow. Confirm the proposed subscription, permissions and supported channels.

Intercom's Pro add-on includes Monitors to select conversations for routine sampling, high-risk review or a specific theme, and custom Scorecards to assess criteria such as tone, accuracy or policy adherence. It is priced on conversation volume rather than seats. Operator is included with the add-on and comes with 2,000 free credits per month; usage above that is charged at $0.01 per credit.

Matching support quality needs to capabilities

Intercom Pro add-on facts

Check whether customers benefit

Follow a finding through the action taken and then examine a related customer result. For an answer correction, inspect later answers and repeat contact about the same issue. For a handover correction, examine whether customers had to restate information.

Keep the case population and observation window clear. A higher internal quality score alone does not establish a better customer experience.

Intercom's CX Score uses AI to rate every conversation and explain why it was positive or negative. Intercom states this provides significantly more coverage than traditional Customer Satisfaction (CSAT) surveys. Reporting metrics that use CX Score require the Pro add-on.