How casinos rate players' LTV and ROI

Introduction: Why know LTV and ROI

In iGaming LTV (Lifetime Value), the player answers the question "how much net value does the user bring over the entire life cycle," and the ROI shows "how much the marketing investment in attracting it has paid off." The correct LTV/ROI score manages performance budgets, bonus policies, payout limits, and VIP prioritization and retention programs - subject to Responsible Gaming.

Basic definitions and formulas

Revenue bases

GGR (Gross Gaming Revenue) = Bets − Wins.

NGR (Net Gaming Revenue) = GGR − bonuses − jackpot contributions − provider commissions.

Net Revenue (operator) = NGR − payment commissions − affiliate/media commissions − fraud/chargebacks.

LTV (historical and predictive)

Historical LTV (on date t): sum of the actual Net Revenue by player from registration to t.

Predictive LTV (T-day horizon): Net Revenue expected for T period discounted:

LTV_T = Σ_{d=1..T} E [NetRev _ d ]/( 1 + r) ^ {d/30}, where r is the monthly risk discount rate.

ROI и Payback

CAC (Customer Acquisition Cost): all costs for attracting a player (media, creatives, agency, bonuses-for-regu, tracking).

ROI_T = (LTV_T − CAC) / CAC.

Payback Period: The minimum t at which cumulative Net Revenue ≥ CAC.

Data and accounting model: what to add to LTV

Income: deposits → rates → GGR → NGR (including provider/jackpot deductions).

Reductions:

bonuses and freespins (at fair value, breakage is recorded later);
payment commissions (MDR, fix per transaction, FX margin);
affiliate/streaming (CPA, RevShare, Hybrid by attributed players);
fraud/chargeback/chargeback-fee;
operational discounts (payments on personal offers).

Not to be confused: deposit turnover ≠ income. LTV includes Net Revenue, not cash flow.

Cochort approach

We consider LTV by cohort (month of registration × channel × GEO × brand × vertical).

Key curves:

Retention D1/D7/D30/D90/D180/D365 (share of active players);
ARPPU/ARPDAU by week of life;
Bets/sessions/hours as behavioral proxies.
Cohorts are needed to see the impact of seasonality, regulation and bonus policy changes.

LTV Prediction Models

1. Retention curves + average check

E[NetRev_d] = P(active_d) × E[NetRev	active, d].
P (active_d) is taken from the survival model (Weibull/Log-logistic), a check from historians/GLM.

2. Markov/Survival

States: "new," "active," "dormant," "reactivated," "gone."

The transition matrix gives the probability of activity on day d and the probability of reactivation.

3. ML approach (GBM/LightGBM/NN)

Features: first 72 hours (deposits, sessions, games, device, time zone, payment path, reaction to bonus), as well as risk signals (RG).

Target: Net Revenue over the horizon of 90/180/365 days with a log transformation.

Regular retraining and shift tests are mandatory.

4. VIP-models

We predict the probability of entering the VIP segment and the size of future net income; add human-in-the-loop (VIP managers) and RG restrictions.

What is critical to correct in LTV

Bonus economy: count at fair value (EV) and take into account deferred breakage.

Payments: blended MDR and cashout-fees depend on AWP/GEO (crypto/instant banking/card).

Provider commissions: different in verticals (live is more expensive than RNG).

Taxes/Levi on NGR: in "white" GEOs reduce margins; consider by GEO cohort.

Risk and discount: Use r> 0 for horizons> 90 days, especially for new channels.

RG/AML restrictions: deposit and self-exclusion limits reduce the expected LTV - this is a blessing, not a "loss": this is how you reduce penalties and reputational risks.

Attribution and incrementality

Last click/first touch - convenient, but distort ROI; for toll channels use MMM/geo-elevator, A/B geo-holdouts, PSA-control.

Incremental ROI: LTV difference between test and control group/incremental expenditure.

Reactivation vs is new: don't mix - they have different payback curves.

Calculation example (simplified)

July-registration cohort, GEO A, channel X (180-day horizon).

P(active_d) по survival → 0. 55 (D7), 0. 38 (D30), 0. 22 (D90), 0. 15 (D180).

Average Net Revenue per active: $3. 2/day (D1-D7), further decreasing to $0. 7 to D180.

Integrating by day and discounting r = 2 %/month, we get LTV_180 ≈ $126.

CAC (Channel X): $70 (media + creative + trackers + part of welcome bonus).

ROI_180 = (126 − 70)/70 = 0. 8 (80%); Payback ≈ D112 (the day when cumulative Net Revenue covers $70).

For VIP subsegment (top 5% VIP probability) LTV_180 ≈ $520 at CAC $140 → ROI = 271%, but with upper RG limits.

Dashboards for operational management

1. LTV-Cohort View: by month of registration × GEO × channel × vertical; LTV_30/90/180, Payback, CAC, ROI.

2. Unit-Economics Live: approval rate, MDR, cashout T-time, share of bonuses in NGR, affiliate share.

3. Retention/RG: retention curves, percentage of self-exclusions, triggering of RG triggers, average time to limit.

4. Attribution/Incrementality: geo-holdouts results, uplift by channel/creative.

5. VIP-panel: VIP probability forecast, marginality after taking into account personal offers and service-bones.

Practice rules and thresholds

Payback target: 90-120 days for mass channels; 180 + is acceptable for high quality (brand/organic).

CAC-guardrails: CAC ≤ 0. 6 × LTV_180 for paid channels (with risk margin).

Bonus-CAP: share of bonuses ≤ 25-30% of the NGR cohort (depending on market and vertical).

Payments: approval> 88%, blended MDR <2. 5%, cashout T-time < 12–24 ч.

Models: retrain every 2-4 weeks, monitor drift and calibration.

Common mistakes

1. Count LTV on deposits, not Net Revenue.

2. Ignore taxes/levi and payment fees.

3. Mix reactivation with primary recruitment.

4. Rely only on last-click without incrementality.

5. Do not discount long horizons.

6. Overestimate VIP LTV due to outlier effect without averaging over VIP probability.

7. Ignore RG limits and self-exclusions in the forecast.

Loop LTV/ROI implementation checklist

Unified Data Schema: Bets/Winnings → GGR → NGR → Net Revenue.
Cohorts on registration and traffic source; separate accounting of reactivations.
Survival/Markov + ML scoring of early LTV (72 hours).
Full accounting of bonuses, MDR, affiliate payments, fraud and taxes.
Dashboards by LTV_30/90/180, Payback, ROI, with alerts for deviations.
Incrementality experiments (geo-holdouts, PSA control).
RG restrictions are built into targeting and offers; report on the impact of RG on LTV.
Model retraining procedures and data shift control.

Casino LTV and ROI scores are not a "single digit" but a live circuit: the correct Net Revenue base, cohort analytics, predictive models, and incrementality checking. When LTV is considered fair (taking into account bonuses, payments, taxes and RG), it becomes a reliable driver for marketing, VIP programs and product strategy - and directly converts into sustainable margins and predictable growth.