The KPIs that matter most for casino and sportsbook support teams fall into five groups: satisfaction, response and resolution speed, productivity, cost and economics, and compliance or safer-gambling indicators. Operational speed and safer-gambling detection deserve top priority, because Gambling Commission guidance now treats timely intervention as a licence condition, not a nice-to-have. AI deflection and compliance tracking increasingly sit at the centre of support economics rather than at the edges.
TL;DR:
- Set channel specific first response targets under 1 minute for live chat and messaging, and under 1 hour for email; never blend SLA results.
- Treat CSAT below 80% and intervention delays beyond 24 hours as immediate review triggers; act within an hour on strong risk indicators.
- Segment CSAT by new, recreational, VIP, and risk flagged cohorts, and read it alongside first contact resolution rather than relying on a blended average.
- Track AI deflection with reopen and escalation rates, because premature closures can inflate savings; log every safer gambling action with a timestamp and reviewer.
Table of Contents
- Key takeaways: what to start tracking this week
- Customer satisfaction metrics: CSAT, NPS and CES explained
- Response and resolution speed: FRT, ART and FCR by channel
- Productivity and efficiency: AHT, occupancy and backlog
- Cost and economics: cost per ticket and AI deflection
- Compliance and risk KPIs mapped to LCCP 3.4.3
- Using benchmarks and percentiles to set realistic targets
- Building the KPI dashboard: data, formulas and governance
- How Pitbot supports these KPIs in practice
- Why single-metric thinking fails casino support teams
- Measure these KPIs with AI and human review built in
- FAQ
- Sources
Key takeaways: what to start tracking this week
Before building a full dashboard, a few moves give you usable signal within days rather than months.
- Publish one canonical formula per metric so every team reports the same number the same way.
- Stand up a live SLA view for first response time across chat, email and messaging channels.
- Start weekly CSAT reporting segmented by player cohort, not just a blended average.
- Begin logging AI deflection rate alongside reopen and escalation rates from day one.
- Read CSAT together with first contact resolution, and read cost per ticket together with AI deflection rate: each pair tells you whether speed or savings are coming at the expense of experience.
For gaming operations specifically, treat any CSAT below 80% or any flag-to-intervention time beyond 24 hours as an immediate review trigger, not a quarterly footnote.
Customer satisfaction metrics: CSAT, NPS and CES explained
Satisfaction metrics tell you whether speed and compliance are translating into a player experience worth keeping. Three measures cover most of what casino support teams need.
CSAT (Customer Satisfaction Score) is the share of respondents who rate an interaction positively, calculated as (number of positive responses ÷ total responses) × 100. It is best asked immediately after a chat or ticket closes, while the interaction is still fresh.
NPS (Net Promoter Score) asks how likely a player is to recommend the operator, on a 0 to 10 scale, and is calculated as (% promoters, scoring 9 to 10) minus (% detractors, scoring 0 to 6). It works better as a quarterly relationship measure than a per-ticket one, since it reflects the overall player relationship rather than a single support exchange.
CES (Customer Effort Score) asks how easy a specific task was, typically on a 1 to 5 or 1 to 7 scale, and is most useful immediately after account, payment or verification queries where friction is common.
Segmentation matters more in gaming than in most sectors. A blended CSAT hides real differences between a new player struggling with KYC, a recreational player asking about a promotion, a VIP expecting white-glove handling, and a risk-flagged player whose support experience intersects with responsible gambling obligations. Split reporting into these cohorts before presenting trend lines to stakeholders.
- Survey immediately post-interaction for CSAT and CES; reserve NPS for periodic relationship pulses.
- Segment every satisfaction metric by player cohort: new, recreational, VIP, risk-flagged.
- Feed CSAT and CES into the weekly operational dashboard; feed NPS into the monthly executive view.
- Present movement as a trend line with cohort breakdowns, not a single blended figure.
Top-performing support operations report high CSAT scores on conversation channels, according to benchmark data from Freshworks, which gives gaming operators a concrete ceiling to aim towards on live chat and messaging.
Response and resolution speed: FRT, ART and FCR by channel
Speed metrics are where regulatory expectation and player experience overlap most directly, since a slow response on a risk-related query carries both a satisfaction cost and a compliance cost.
First Response Time (FRT) is the time between a player’s first message and the first human or AI-assisted reply, calculated as the timestamp of first response minus the timestamp of ticket creation. Average Resolution Time (ART) is the average time from ticket creation to closure across all resolved tickets in a period. First Contact Resolution (FCR) is the share of tickets closed after a single interaction, calculated as (tickets resolved on first contact ÷ total tickets) × 100.
Multi-threaded tickets, where a player reopens a conversation across channels or sessions, need a counting rule decided in advance: either treat each reopen as a new ticket for FRT purposes while keeping ART and FCR tied to the original ticket ID, or exclude reopens from FCR entirely and track them as a separate reopen rate.
Channel matters enormously here. Freshworks’ 2025 benchmark report shows that conversation channels like live chat can achieve sub-minute first responses among top performers, while ticketing-based channels such as email carry median FRTs measured in minutes to hours rather than seconds. A single company-wide SLA ignores this gap and sets the wrong expectation on at least one channel.
- Set FRT targets under 1 minute for live chat and messaging apps, and under 1 hour for email.
- Calculate SLA compliance as (tickets meeting target ÷ total tickets) × 100, tracked per channel, not blended.
- Reserve the tightest FRT tier for VIP and risk-flagged queues, since these carry both commercial and compliance weight.
- Review FCR monthly against reopen rate, since a high FCR paired with a rising reopen rate signals false closures.
Pro Tip: Set separate SLA tiers for instant channels (chat, SMS, Telegram) and asynchronous channels (email), rather than one blended target that under-serves one side.
Productivity and efficiency: AHT, occupancy and backlog
Productivity metrics measure how much capacity your team has and whether it is being used well, but they are also the easiest to over-optimise in ways that damage satisfaction and compliance.
Average Handle Time (AHT) is the average time an agent spends on a ticket, including active handling and after-call or after-chat work, calculated as (total handle time + total after-contact work time) ÷ number of tickets. Occupancy is the percentage of logged-in time an agent spends actively handling contacts, calculated as (handle time ÷ logged-in time) × 100. Agent utilisation extends this to include all productive work, not just live contact handling. Backlog age tracks the average age of unresolved tickets in the queue, flagging where volume is outpacing capacity.
The pitfall is treating AHT as a target to minimise in isolation. A support agent who rushes a risk-flagged conversation to hit an AHT target may miss the indicators that LCCP Condition 3.4.3 requires licensees to act on, and a rushed VIP conversation can cost more in retention than it saves in handle time. AHT works as a capacity-planning input, not a quality target.
- Calculate AHT per channel and per query type, since payment disputes and account queries carry naturally different handle times than promotional questions.
- Track occupancy alongside CSAT: rising occupancy with falling CSAT signals agents are overloaded, not more efficient.
- Weight VIP and risk-flagged cases separately in workforce planning, since these need more time per contact, not less.
- Monitor backlog age daily on live queues and weekly on asynchronous ones.
Cost and economics: cost per ticket and AI deflection
Unit economics in casino support now hinge on how much of the ticket volume AI can safely absorb, and how that shifts cost per ticket without compromising the human review that compliance in regulated markets requires.
Cost per ticket is calculated as total support costs (staffing, tooling, overheads) ÷ total tickets handled in the period. A useful refinement splits this into cost per AI-resolved ticket and cost per human-handled ticket, since the gap between the two is now the main lever on overall spend.
AI deflection rate is the share of contacts resolved without human intervention, calculated as (tickets resolved by AI ÷ total tickets) × 100. Bot resolution rate is a narrower measure of how many AI-handled conversations close successfully without escalation. Both numbers should be read alongside reopen rate and escalation rate, since a high deflection rate achieved by closing tickets prematurely will show up later as rework and lower CSAT.
Industry projections put AI deflection rising from 18% in 2024 to around 50% in 2026, according to knowledgelib’s 2026 customer support benchmark summary, with a corresponding divergence in cost per ticket between AI-resolved and human-assisted cases. For gaming operators, savings from deflection are best partly reinvested into compliance and responsible gambling capacity rather than taken entirely as margin, since regulatory expectations on timely intervention are rising at the same time that ticket volumes grow.
Compliance and risk KPIs mapped to LCCP 3.4.3
Gambling Commission guidance requires remote licensees to identify, act on and evaluate indicators of harm, which means safer-gambling metrics need to sit inside the same dashboard as satisfaction and speed KPIs, not in a separate compliance silo.
Measurable risk indicators include spend patterns, time spent on site, account behaviour changes, customer-led contact about gambling concerns, and use of safer-gambling tools such as deposit limits or time-outs. LCCP Condition 3.4.3 expects tailored, timely action once these indicators appear, with stronger indicators triggering faster and firmer responses, up to suspending service.
Two KPIs turn this into something measurable: flag-to-intervention time, which tracks how long elapses between an indicator firing and an action being taken, and automated-action review rate, which tracks the share of automated interventions that receive manual review before or shortly after action. The Commission’s guidance also expects licensees to evaluate whether interventions actually reduce harm and to keep records that demonstrate this to regulators on request.
| Indicator | KPI | Target |
|---|---|---|
| Spend pattern change | Flag-to-review time | Under 24 hours |
| Strong risk indicator | Flag-to-intervention time | Under 1 hour |
| Automated suppression action | Manual review completion rate | 100% reviewed |
| Customer-led contact about harm | Response time to self-identified risk | Under 1 hour |
- Log every automated safer-gambling action with a timestamp and the reviewing staff member’s identity.
- Track intervention effectiveness by monitoring post-intervention behaviour change, not just whether an action was logged.
- Keep audit trails detailed enough to reconstruct the full decision chain for any flagged account.
Using benchmarks and percentiles to set realistic targets
External benchmarks are a starting point for calibration, not a target to copy directly, since your channel mix and player base will differ from the aggregate sample behind any published figure.
Freshworks’ 2025 benchmark report shows wide spread between median and top-quartile performers on FRT and resolution time, particularly between conversation channels and ticketing channels. Rather than adopting a single external number, translate percentiles into your own tiers: treat the top-quartile benchmark as your stretch target for VIP and live-chat queues, the median as your baseline target for standard queues, and the lower quartile as the floor below which a channel needs investigation.
To run a baseline benchmarking exercise:
- Pull 90 days of your own FRT, ART, FCR and CSAT data, segmented by channel and cohort.
- Compare your median against the external median for each equivalent channel.
- Set stretch targets at the external top quartile only for channels where you are already near the median.
- Re-baseline every two quarters as channel mix and AI deflection shift.
- Never apply one blended SLA across chat, email and messaging; benchmark each channel separately.
- Treat VIP and risk-flagged cohorts as their own benchmark category, since they carry different acceptable thresholds.
- Revisit targets after any major process or tooling change rather than waiting for the next scheduled review.
Building the KPI dashboard: data, formulas and governance
A working KPI dashboard pulls from several systems and applies the same counting rules consistently, or the numbers drift apart within a quarter.
- Consolidate data from the ticketing system, chat transcripts, CRM records, payment platform events and safer-gambling signal feeds into one reporting layer.
- Fix canonical formulas for every metric in this article and document edge rules, including how reopened tickets count towards FCR and how multi-channel conversations are stitched into a single ticket ID.
- Build three dashboard layers: a live operations view showing current queue depth and FRT against SLA, a weekly health view covering CSAT, FCR, AHT and deflection trends, and a monthly executive view combining cost per ticket, compliance KPIs and cohort-level satisfaction.
- Set alert thresholds that trigger a review, such as FRT breaching SLA for 15 minutes on live chat, or flag-to-intervention time exceeding 24 hours on any account.
- Apply change control to formula updates: log any change to a KPI definition, notify stakeholders, and keep both old and new calculations running in parallel for one reporting cycle to validate consistency.
Pro Tip: After any process or tooling change, run the old and new KPI calculations side by side for at least two weeks before retiring the old method, so you can confirm the shift in the numbers reflects reality rather than a counting change.
How Pitbot supports these KPIs in practice
We built Pitbot around the idea that casino support KPIs and compliance obligations should be measured from the same data, not reconciled after the fact between separate systems.
The platform combines multiple AI providers with the option to hold AI-drafted replies for staff approval before they reach a player. With auto send off, AI-drafted replies on email, live chat, SMS and Telegram wait for staff approval, so faster drafting can bring first response time down without removing the human check. Every reply a staff member approves is recorded with who approved it. Coverage across common communication channels means response and resolution metrics can be tracked consistently across every channel a player might use, rather than only on the channel with the best tooling.
Responsible gambling and regulator flags on email, live chat, SMS and Telegram are a hard stop: the thread is escalated to a person instead of answered, and a sensitive case opens for someone to pick up. The analytics dashboard counts those escalations alongside time to first response, AI handle rate, and reopen, repeat-contact and first-contact resolution rates. Automated sentiment detection also feeds the dashboard, where KPIs can be sliced by channel, sentiment, VIP level and language. Satisfaction scores and flag-to-intervention time are not Pitbot metrics, so measure those in your own survey tool and case records.
- Map FRT and AI handle rate improvements against your pre-existing baseline before and during any pilot.
- Track reopen and escalation rates alongside AI handle rate to confirm quality holds as AI handles more volume.
- Check that every responsible-gambling escalation was picked up and handled by a named member of staff.
Why single-metric thinking fails casino support teams
The biggest mistake support leaders make going into 2026 is chasing one headline number, usually AI deflection rate or average handle time, while treating compliance metrics as a separate reporting exercise for the legal team. That split is exactly backwards.
My three priorities for the next 12 months: build one dashboard that puts satisfaction, speed, cost and safer-gambling KPIs on the same screen; resist any target that rewards closing tickets fast over closing them correctly; and put a human reviewer in the loop on every automated action that touches a player’s account status, not just the ones that feel sensitive in hindsight.
Channel proliferation, particularly messaging apps, will keep stretching response-time expectations upward. The operators who hold their ground on compliance quality while still improving speed will be the ones still trusted by their regulator in a year’s time.
— pitbot
Measure these KPIs with AI and human review built in
Tracking the metrics above by hand across five channels and a compliance spreadsheet is the kind of work that eats a support leader’s week before the real decisions even start. We built Pitbot to put time to first response, AI handle rate and responsible-gambling escalations on one analytics dashboard, fed by the same conversations your team is already having across email, live chat, SMS and Telegram.
You can hold AI-drafted replies for staff approval before they reach a player, and every approved reply records who approved it, so the speed gains from automation show up in your FRT and cost-per-ticket numbers without creating a gap in your audit trail. Our features page covers the channels, analytics and compliance tooling in detail, and our responsible gambling automation escalates conversations flagged for responsible gambling or regulatory concerns to your team instead of answering them, and opens a sensitive case for someone to pick up.
If your team is ready to see what balanced KPI reporting looks like with AI and human oversight working together, book a look at Pitbot for your support operation and bring your current benchmarks to compare against.
FAQ
What are key performance measures for call centre agents?
The core measures are first response time, average handle time, first contact resolution and CSAT, each calculated per agent and rolled up to team level. For casino support specifically, agents handling VIP or risk-flagged accounts should also be measured on flag-to-intervention time and manual review completion rather than speed alone.
What are good examples of KPIs for casino support?
Strong examples include CSAT, first response time, first contact resolution, cost per ticket, AI deflection rate and flag-to-intervention time for safer-gambling indicators. The strongest KPI sets combine satisfaction, speed, cost and compliance measures rather than relying on any single number.
What is KPI gamification?
KPI gamification applies game-like elements, such as leaderboards, badges or point systems, to performance metrics to motivate agents around targets like resolution speed or CSAT. In casino support, it needs care so that gamified targets never reward speed at the expense of the manual review steps that compliance requires.
What are the key performance indicators for customer service?
The main KPI categories are customer satisfaction (CSAT, NPS, CES), response and resolution speed (FRT, ART, FCR), productivity (AHT, occupancy, backlog) and cost per ticket. Casino and sportsbook operators add a further category: safer-gambling and compliance indicators tied to regulatory requirements such as LCCP Condition 3.4.3.
How do I measure casino support performance against industry benchmarks?
Pull your own FRT, resolution time, CSAT and FCR data segmented by channel, then compare medians against published benchmarks such as the Freshworks 2025 report. Set stretch targets at the top-quartile benchmark only for channels already performing near the median, and re-baseline every couple of quarters.
Sources
- Customer interaction guidance - for remote gambling licensees (Gambling Commission)
- Customer Service Benchmark Report 2025 (Freshworks)
