How much should a bonus cost as a share of GGR?

There is no safe industry number to copy — the ratio depends on market maturity, product mix, tax treatment and how aggressively the brand acquires. What matters is that you measure bonus-to-GGR on a consistent definition (consumed bonus value, not granted value) and watch its trend against retention and NGR. A ratio that is stable while NGR grows is healthy; a ratio that climbs while deposits flatten means you are buying activity you already had.

What is the difference between granted, consumed and cost of a bonus?

Granted is the face value you credited. Consumed is the part players actually wagered into the game. Cost is what left the business: consumed value multiplied by the expected payout of the games it was played on, plus any withdrawable balance released. Reporting granted value as cost overstates spend, usually by a wide margin; reporting only released withdrawals understates it. Pick one definition, write it into the metric dictionary, and never mix the three in the same chart.

Do higher wagering requirements make a bonus cheaper?

They lower the share of bonus value that converts into withdrawable cash, but not linearly, and not for free. A high wagering requirement also lowers take-up, pushes value toward the players most able to grind it, and raises support and complaint volume. Treat the requirement as a price signal to the player, not just as an accounting brake: the cheapest bonus is the one nobody claims, and it produces nothing.

How do I know whether a bonus produced incremental revenue?

Hold out a comparable group. Without a holdout or a defined baseline cohort you can only observe that bonused players deposited more, which is true almost by construction — they self-selected. Define the eligible population, randomise or match, fix the observation window before launch, and compare net revenue after bonus cost. Anything else measures correlation and reports it as ROI.

Should bonus rules live in the platform or in the CRM?

The rules should be configurable in one engine and consumed by both. Splitting eligibility logic across a CRM campaign tool and a platform bonus engine is the most common source of double-granting, wrong game weighting and abuse windows. One rule set, one audit log, one place where finance can reconstruct why a specific player received a specific offer.

Bonus Economics for iGaming Operators

The number most operators get wrong first

Almost every bonus argument inside an operator is really an argument about definitions. Marketing reports granted value, finance reports released withdrawals, and BI reports something in between — so the same campaign looks cheap, expensive and neutral in three decks on the same day.

Fix the vocabulary before the strategy:

Term	What it counts	Where it misleads
Granted	Face value credited to player accounts	Counts bonuses nobody touched; overstates spend
Consumed	Bonus value actually wagered into games	Ignores what the house expected to return on those wagers
Cost	Consumed value × expected payout of the games played, plus released withdrawable balance	Requires per-game weighting; breaks silently when game mix shifts
Bonus-to-GGR	Cost as a share of gross gaming revenue	Moves for reasons that have nothing to do with promotions — check the denominator first

Write the chosen definitions into the metric dictionary and make every dashboard reference it. This single act removes more recurring conflict than any change to the offers themselves.

The four levers you actually control

Bonus design looks like an infinite space. Operationally there are four levers, and everything else is packaging:

Grant size and shape. Fixed amount, percentage match, free rounds or cashback. Shape decides who finds the offer attractive as much as size does.
Wagering and eligibility. The requirement multiplier, the qualifying deposit, maximum bet while a bonus is active, expiry, and game weighting. This is where cost is genuinely controlled.
Targeting. Who is eligible and how often. An offer sent to everyone is a discount; an offer sent to a defined segment is an experiment.
Frequency and stacking. How offers overlap over a player's lifetime. Most runaway bonus costs are stacking failures, not pricing failures.

Game weighting deserves specific attention. Weighting contribution by game type is how a bonus programme survives contact with a lobby whose game mix changes monthly. If weighting is static while the mix moves, the effective cost of every live bonus drifts without anyone changing a campaign.

Pricing a bonus before launch

A bonus can be priced the same way any promotion is priced: expected cost per accepting player, multiplied by expected acceptance, against expected incremental margin. The inputs are operator-specific, and that is the point — the model is portable, the numbers are not.

Input	Where it comes from	Failure mode if guessed
Take-up rate	Prior campaigns on the same segment and channel	Budget sized for a take-up the offer never reaches — or blows through it in a weekend
Expected payout of eligible games	Certified RTP and observed play distribution	Cost modelled on the lobby average while play concentrates in the highest-payout titles
Completion rate	Historic share of players who clear the wagering requirement	Cost looks low because most value expires — until the requirement is loosened
Incremental margin	Holdout or matched-cohort result, after bonus cost	The campaign is judged on revenue it did not create

If a required input does not exist yet, that is a measurement task, not a reason to launch on instinct. Run the first version deliberately small and instrumented rather than large and unexplained.

Measuring: the part that decides everything

Bonused players deposit more than non-bonused players in essentially every dataset, at every operator, always. That comparison proves nothing: players who claim offers are players who were already engaged. The only honest question is incremental — what happened that would not have happened otherwise.

Define the eligible population before the split, not after seeing results.
Hold out a comparable group, randomised where possible and matched where not.
Fix the observation window in advance, long enough to include the second deposit cycle — see FTD2SD.
Compare net of bonus cost, using the cost definition above, not granted value.
Report the denominator. A conversion rate without its base population is a decoration.

The same discipline applies to cohort retention and LTV claims that follow a promotion. State the horizon, the revenue definition and the censoring rule, or do not state the number.

Where bonus programmes leak

Stacking. Two campaigns, two owners, one player. Overlap rules belong in the engine, not in a spreadsheet of who is running what.
Weighting drift. Game contribution set once and never revisited while the lobby changes underneath it.
Abuse windows. Every new mechanic ships with an exploit path; assume it exists and instrument it. See bonus abuse prevention.
Reconciliation gaps. Finance cannot reconstruct why a specific player received a specific offer. If the audit log cannot answer that, the programme is unauditable regardless of its ROI.
Vanity reporting. Campaigns judged on claims and clicks rather than incremental NGR.

What good looks like

A healthy bonus programme is boring to operate and easy to explain: one rule engine, one cost definition, offers targeted at defined segments, a standing holdout, and a monthly review that compares incremental net revenue against bonus cost by segment. When those five things exist, the interesting conversations become product ones — which is where they belong.

Continue reading: Game mix optimization — what the lobby data says about the shelf your bonuses point at. The Turbo Stars platform — how the bonus engine, CRM and wallet are wired together.

Bonus economics for operators.