Casino customer support metrics: what to track and why
A fast first reply is useful. It tells you very little about whether the agent understood the problem, gave the right answer or followed up when they said they would.
First response time: split the queue before averaging it
Measure the time between a player's first message and the first human response. State whether the clock runs around the clock or only during staffed hours. If a bot replies immediately, report that separately from the human response.
Break the result down by channel and shift. A combined average can hide a slow overnight queue behind a busy daytime team. Live chat and email also have different expectations, so compare each with its own agreed target.
Use the median alongside a slower-end measure such as the 90th percentile: the time within which 90% of first replies were sent. This makes it easier to see whether a minority of players are waiting much longer than everyone else.
Resolution time: show where the time goes
Define the start and end of a case before measuring resolution time. Decide how reopened cases are handled, and keep that definition consistent. Otherwise, closing a conversation and opening a new one can make the report look better without improving the player's experience.
Report elapsed time as well as, where your tools support it, time waiting on a player, an internal team or an external provider. A withdrawal investigation may remain open while payments checks the transaction. Support should still be judged on the accuracy and timing of its updates.
Compare similar case types. Password questions, verification cases and payment disputes do not need the same amount of work. A change in the mix of contacts can alter the average even when agent performance stays the same.
CSAT: include the sample, not just the percentage
For a survey using satisfied and unsatisfied responses, CSAT is satisfied responses divided by all valid responses, multiplied by 100. If you use a five-point scale, document which ratings count as satisfied.
72 satisfied responses out of 90 valid responses gives 80% CSAT. If 600 surveys were delivered, the response rate is 15%. Report all three figures so the percentage has context.
Read the comments and check why players are unhappy. Some ratings reflect the support conversation; others reflect a payment delay or an outcome the player disliked. Both matter, but the fix may sit with different teams.
Be precise when reporting improvement. A rise from 60% to 80% is 20 percentage points, or about a 33% relative increase. Those are different descriptions of the same change. Name the measurement period and use the same survey method.
Backlog: count age as well as volume
Track how many cases remain unresolved and how long they have been open. A queue with 30 old payment cases may need more attention than one with 100 new, straightforward questions.
Useful age bands might be under 24 hours, one to three days and over three days. Those are reporting examples, not universal deadlines. Set your own bands around the service commitments and case types in your operation.
Separate cases waiting on the player from those waiting on your team or a provider. Count overdue follow-ups too. This helps you identify work that can move now and cases where the player still needs an update.
Repeat contact and quality reviews
Define repeat contact around the same issue within a chosen period, rather than treating every new conversation from a player as a failure. Document how you link cases. A player asking about a new bonus after a resolved deposit issue is not necessarily a repeat-contact problem.
Then review a sample of conversations for correctness, investigation, clear wording, escalation and follow-through. Include different shifts and case types, plus some random cases, so the sample is not limited to complaints or your best agents.
Critical mistakes should remain visible even when the overall score looks good. An incorrect account-security action should not be averaged away by friendly wording. Quality assurance should help agents improve and identify procedures that need fixing.
Turn the weekly report into a short action list
Use a weekly report to decide what changes next. For each issue, show the affected queue or shift, the evidence, a named owner and a review date. Avoid adding a metric unless somebody can explain what decision it supports.
- Slow overnight replies: inspect demand and the roster before adding staff.
- Older payment cases: check the escalation route and outstanding follow-ups.
- Repeated bonus questions: review the terms and the agent briefing.
- Low quality scores: separate training gaps from unclear procedures.
Roundtide combines player support with training and quality reviews. If you are still preparing to launch, the 24/7 support checklist covers what to put in place before you start measuring the live queue.