iGaming retention KPIs: what to measure and how

Look up "casino retention metrics" and you'll find versions of the same list: retention rate, churn, ARPU, LTV, CAC, sometimes NPS. The formulas are often loose. One widely shared KPI guide defines retention one way in its summary table and another way in its body text. The benchmarks are rarely sourced: a typical glossary says casino 30-day retention is "between 20% and 40%" and cites nothing.

The problem is not that these metrics are wrong. It's that they lack the three things that make a retention number usable: a precise definition, a control group, and a counterweight. This guide sets out a KPI stack built around those three.

First, define the words

Two operators can report "30-day retention" and mean completely different things. Write these definitions down before you build a dashboard, and don't change them without restating history.

The deposit-based definition matters most. A player who logs in to claim free spins and leaves hasn't been retained. They've been paid to visit.

Layer 1: outcome metrics

These tell you whether retention is working at all. Review them monthly.

Cohort retention (deposit-based). Of the players whose first deposit was in month M, the share who deposited in month M+n. Plot it as a cohort triangle, not a single number. A single blended retention rate hides whether newer cohorts are getting better or worse.

Cohort retention(M, n) = players from cohort M who deposited in month M+n ÷ players in cohort M

NGR retention. The same triangle, using net gaming revenue instead of headcount. It shows whether the players you keep are worth keeping. NGR retention below headcount retention means you're keeping low-value players and losing high-value ones.

Player lifetime value. Use a fixed horizon (90-day or 12-month NGR per depositing player), not "lifespan × ARPU", which depends on a lifespan nobody knows until it's over. The player LTV calculator works this out from your own inputs.

Bonus cost ratio. Bonus cost ÷ gross gaming revenue for the cohort. Retention that only holds while this ratio climbs is rented, not earned.

Layer 2: lifecycle metrics

These show where in the lifecycle players are lost, and they are where retention work actually happens.

Registration-to-FTD conversion. Players who registered in a period and made a first deposit within N days. This is often the biggest leak in the whole funnel, and the cheapest to fix. Calling players who registered but didn't deposit is covered in registered no-deposit players.

Second-deposit rate and time to second deposit. A first deposit is a trial. The second is the first real sign of retention. Track the share of FTD players who make a second deposit within 14 or 30 days.

Dormancy inflow. The number of players crossing your dormancy threshold each week. It's an early warning. If inflow rises, retention will fall in next month's cohort chart. Segmenting the dormant base is covered in dormant player segmentation, with a ready-made dormancy segmentation template.

Incremental reactivation rate. The KPI most lists get wrong. Raw reactivation counts every dormant player who came back after contact, including those who would have come back anyway.

Incremental reactivation = reactivation rate in the contacted group − reactivation rate in the holdout group

To measure it, randomly hold back part of every eligible segment, say 10%, and contact nobody in that group. Same criteria, same window, no calls, no emails from that campaign.

Re-lapse rate. Of the players you reactivated, the share who are dormant again within 60 or 90 days. A campaign with a high reactivation rate and a high re-lapse rate has mostly bought one bonus-driven session.

Layer 3: channel metrics for calls

Phone campaigns add a funnel of their own. Each step explains movement in the step after it.

Consent reach deserves its own line on the dashboard. In Great Britain, SR Code 5.1.12 has required opt-in per product and per channel since May 2025, with phone calls as a separate channel. In the Netherlands, the customer-relationship route for telemarketing closed on 1 July 2026. Your callable base is now a KPI you can grow, through better preference prompts, and it caps everything below it in the funnel. The market rules are in UK Gambling Commission rules for marketing calls and calling players in the Netherlands.

The counterweights: safer-gambling KPIs

Every retention metric above can be pushed up in ways that harm players. Bigger bonuses, more frequent calls, or calling players who stopped for good reasons will all lift short-term reactivation. So every retention report needs counterweights on the same page:

• Harm flags per 1,000 conversations. Calls in which the player showed signs of harm or asked to exclude. A rise after a script change is a reason to roll the script back.
• Opt-out and complaint rate. Per conversation, by script and by segment.
• Share of reactivated players who set or lowered a limit within 30 days. Not a target to push, but a sign of whether reactivated players are in control.
• Interaction coverage. Players flagged by calls who were then reviewed by your safer-gambling team within your own service level.

This isn't only ethics. The UK Gambling Commission's customer interaction code requires licensees to take all reasonable steps to evaluate the effectiveness of their overall approach, for example by trialling and measuring impact, and to be able to show the results. A KPI pack with harm metrics next to deposit metrics is also the evidence you'd show.

Worked example: one month of a reactivation campaign

Illustrative figures for a mid-sized operator, to show how the layers fit together. None of these numbers are benchmarks.

Say you have 12,000 players who crossed a 60-day dormancy threshold last quarter.

1. 1

   Eligibility

   4,200 have phone-marketing consent and pass the self-exclusion, harm and age filters. Consent reach = 4,200 ÷ 12,000 = 35%.

2. 2

   Holdout

   10% of the eligible players, 420, are held back at random. 3,780 are called.

3. 3

   Funnel

   1,134 answer as the right person (30%). 1,020 get past the identity check (90% conversation rate). 163 accept the offer (16%). 106 of those deposit within 30 days (65% accept-to-deposit).

4. 4

   All depositors, not only accepters

   Across the whole called group, 249 players deposit within 30 days, a 6.6% raw reactivation rate. Some declined the offer and came back anyway. Some were never reached.

5. 5

   Holdout comparison

   In the holdout, 13 of 420 deposit in the same window: 3.1%. Incremental reactivation = 6.6% − 3.1% = 3.5 points, or about 132 players (3,780 × 3.5%).

Raw reactivation says 249 players came back. Incremental reactivation says the campaign caused about 132 of them. Every cost metric should use the 132.

On the counterweight side, the same month shows 9 harm flags (8.8 per 1,000 conversations) and 41 opt-outs (4.0% of conversations). If next month's script lifts acceptance to 19% but harm flags double, the old script wins.

To turn this funnel into cost per reactivated player and ROI, the reactivation ROI calculator runs it with your own inputs, call prices and bonus costs.

A one-page pack of iGaming retention KPIs

If you can only put a dozen numbers in front of the leadership team each month, use these. Each comes with the question it answers.

Three mistakes show up again and again in packs like this. The first is mixing definitions across teams, so CRM reports login-based retention while finance reports deposit-based retention, and the two never reconcile. The second is reporting campaign results without a holdout, which credits retention with players who were coming back anyway. The third is leaving the safer-gambling row for a separate compliance report, where nobody who sets retention targets ever reads it.

Where these numbers come from in practice

Most of these KPIs need data from three systems: your platform (deposits, NGR, limits), your CRM or journey tool (consent, segments, holdouts), and the call system (outcomes per call).

Dial Milo records an outcome for every call: offer accepted, declined, callback requested, opted out, wrong person or no decision. Each call also gets a transcript, summary and recording, and the AI campaign report labels every player with a next step, with safer-gambling labels taking priority over everything else. If you A/B test scripts, the scripts tab compares called, answered, accepted and, with deposit data connected, deposit rate and cost per depositor, and only declares a winner once it's at least 95% sure the difference is real. With Customer.io connected, outcomes come back to the player profile as attributes and events (milo_offer_accepted, milo_opted_out and so on), so your holdout and deposit analysis can sit in the system that already owns the player.

Two practical rules for the dashboard:

1. Freeze the window. Report 30-day reactivation only once 30 days have passed for the whole cohort. Early reads flatter the campaign, because fast responders always come first.
2. Report by segment, not in total. A 3.5-point incremental lift could be 8 points in 31–60-day dormants and zero in 180-day dormants. The total hides which segment to stop calling.

The broader strategy behind these numbers, including segment choice and offer design, is in the player reactivation guide. For how a CRM structures the data, see iGaming CRM.