iGaming metrics: formulas & how to read them

Build the funnel top to bottom so every rate has a denominator above it: visits → Visit2Reg → registrations → Reg2Dep → FTD, then AVG FD and Reg2TryDep to separate an offer problem from a payment problem. Put a revenue row next — turnover, GGR, NGR and one ARPU/ARPPU pair so you see value per user, not just totals. Set cost against it on the same screen: CPR, CPD and ROI by NGR; a funnel without acquisition cost is a vanity dashboard. Finish with a retention row — D1, Stickiness, churn, reactivation — read on a cohort basis rather than as one blended figure. Keep withdrawals visible (Out/In, Withdrawal AR, average payout time) because they move churn more than most operators expect. Show every rate next to its two raw counts, and plot a 7- and 28-day trend instead of a single value: a KPI with no trend tells you almost nothing.

Four recur. First, no segmentation: a blended average deposit or ARPU hides that one whale segment funds the product while the mass segment barely converts — always split by traffic source, geo, device and player tier. Second, inconsistent definitions: if "active user" means "logged in" one month and "placed a real-money bet" the next, every downstream metric — DAU/MAU, stickiness, churn — shifts for no real reason. Third, outliers left in: a single jackpot win can swing RTP, ARPU by GGR and InOut for a whole cohort, so report both the mean and a trimmed or median figure. Fourth, reading a rate without its volume: a 40% Reg2Dep on 50 registrations is noise, on 50,000 it is a signal. Two more: comparing cohorts of different ages instead of using a fixed day (D7, D30), and treating turnover or visits as success metrics when they only matter relative to revenue and cost.

Walk the funnel from the money back to the traffic. Start at NGR: did GGR itself fall, or did bonus cost and provider fees rise? If GGR fell, check turnover and RTP — lower turnover means fewer or smaller bets (look at active users, average bet, DAU/MAU), while a spike in RTP means players ran hot for the period and it will likely revert. If turnover is down, move up: FTD count, Reg2Dep, then registrations and Visit2Reg, then visits by source. A break high in the funnel with steady conversion below points to a traffic or campaign problem; steady visits with a falling Reg2Dep points to the offer, payment availability or a broken step. In parallel, check Out/In and Withdrawal AR — a payout problem or a bonus-abuse wave can drain NGR without touching the acquisition funnel at all. Always compare against the same weekday and a cohort of the same age, not against yesterday.

Most of these metrics have two versions and they answer different questions. A period (snapshot) view — "Reg2Dep this month" — mixes players who registered years ago with players who signed up yesterday, so it drifts as your traffic mix changes even when nothing about the product changed. A cohort view groups players by when they joined and measures each group at the same age: Reg2Dep for the March cohort by day 30, D30 ROI for the April cohort, LTV for a cohort once it has matured. Cohorts make campaigns and product changes comparable because acquisition date is held constant. The rule "define active the same way every time" matters most here: churn and retention are only meaningful if the inactivity window (say, 30 days with no real bet) and the cohort boundary are identical across every measurement. Use snapshots for daily operational health, cohorts for anything you plan to decide on.

Because each source delivers a different kind of player and a different level of intent. SEO and brand traffic arrives warm — high Visit2Reg, high Reg2Dep, low CPR, but limited volume. Paid social and pop/push traffic is cold and broad — low conversion, high bounce, cheap clicks, and often a long average visit-to-FTD time as people research before funding. Affiliate traffic sits in between and varies by partner and deal type: a CPA partner optimises to the FTD event and can inflate Reg2Dep while sending players with poor Nth-deposit conversion and low LTV; a RevShare partner is aligned with long-term value. Incentivised or bonus-hunter traffic shows a high first-deposit bonus activation rate, high bonus/GGR, high cancellation and near-zero retention. The practical consequence: never judge a source on a single stage. Line up Visit2Dep, D30 ROI by NGR, FTD-to-STD and LTV per source, and weight budget toward sources that clear ROI at a fixed cohort age, not the ones with the prettiest top-of-funnel rate.

No single bonus number is enough; read them as a set against the promo's goal. Bonus issued is the cost. Bonus issued / deposit sum tells you how much bonus it took to pull a unit of deposits — the acquisition efficiency of the offer. Bonus issued / GGR tells you how much of your revenue the promo consumed — the profitability check; a ratio that keeps climbing at product level means promos are subsidising play that would have happened anyway, or attracting the wrong players. Bonus GGR (bonus bets − bonus wins) shows whether the bonus money actually played through or leaked straight to withdrawable balance. Bonus cancellation rate flags terms that are too harsh or an offer that does not fit the segment. Bonus issued to real is the amount that cleared wagering and became cash — the true payout. A healthy promo has a controlled bonus/GGR, a positive or small-negative bonus GGR, low cancellation, and a lift in FTD-to-STD or reactivation that outlasts the campaign window. If retention does not move, the bonus bought turnover, not customers.

Memorise the chain, not isolated formulas. Revenue: GGR = bets − wins; NGR = GGR − bonus cost (and, depending on the shop, − fees, taxes, provider share); margin / hold = 100% − RTP, where RTP = wins ÷ turnover. Value: ARPU = revenue ÷ active users, ARPPU = revenue ÷ paying users, LTV = cohort NGR ÷ cohort players, and LTV must exceed CAC (= spend ÷ acquired depositors) by a comfortable multiple. Funnel: Visit2Reg = regs ÷ visits, Reg2Dep = FTD ÷ regs, Visit2Dep = FTD ÷ visits, and the three multiply through. Engagement: stickiness = DAU ÷ MAU; a rising ratio means a stronger habit. Retention: D1/D7/D30 are cohort return rates at a fixed age; churn is their complement. Payments: Processing AR is per attempt, User AR is per player. Be ready to explain why a metric moved — mix shift, seasonality, an outlier — and which lever you would pull, because reciting the formula is the easy half of the question.

There is no universal benchmark, and treating a number from a blog post as a target is how teams end up optimising the wrong thing. The right values depend on vertical (casino vs sportsbook vs lottery), geo, traffic mix, product maturity and licensing model. A casino RTP of 96% is normal; a sportsbook "RTP" is really 100% − margin and sits far lower per bet but with much higher turnover. Reg2Dep on brand traffic can be several times higher than on cold paid traffic and both can be fine. Instead of chasing external numbers, set your own baseline: pick a stable recent period, compute each metric by segment, and track deviation from that baseline. Define a green / amber / red band per metric based on what actually threatens the P&L — Withdrawal AR or average payout time slipping is a retention emergency, while a 2-point ARPU wobble on a small cohort is noise. Benchmark against your own past and against cohorts, and use industry figures only as a loose sanity check.