> For the complete documentation index, see [llms.txt](https://help.smartico.ai/welcome/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://help.smartico.ai/welcome/products/ai-models/bonus-eligibility-model.md). # Bonus Eligibility Model ## Bonus Eligibility - Help Guide ### What it is Bonus Eligibility tells you, for each player and every day, whether it makes sense to give them another bonus. It looks at how a player has been behaving lately (their real-money play, their deposits, and the bonuses they've had) and gives you a simple answer you can act on: yes, no, or not enough information yet. It's there to guide your decisions, not to make them for you. Nothing is granted or blocked automatically. You choose how to use the answer in your journeys, host tools, and reports. ### What you get One row per player, per brand, refreshed every day. Each row gives you two things: * A verdict, the short answer: Eligible, Suppress Bonus, or Unknown (unset). * The reasons behind it, populated for suppressed players in plain terms (for example "Bonus Dependent" or "Wins over time"), explaining why the model recommended suppression. | Verdict | What it means | What you might do | | -------------- | --------------------------------------------- | ------------------------------------------------- | | Eligible | Recent behaviour supports another bonus | Go ahead and grant | | Suppress Bonus | Recent behaviour argues against another bonus | Hold off, or try a non-bonus offer instead | | Unknown | Not enough data to decide | Fall back on your usual rules. This is not a "no" |
Every brand only ever sees its own players. ### How the verdict is decided The verdict comes from four separate signals, each looking at the player from a different angle. Each one has a weight, so some count more than others: * Profitability - 35% * Bonus Effectiveness - 30% * Bonus-to-Deposit Ratio - 25% * Bonus Share - 10% A player is only given a verdict when at least three of the four signals have enough data to run. The scores that did run are then re-balanced so the missing one doesn't skew the result. This keeps things fair, so a player is never judged on a single signal. Here's what each signal asks, and what a good or bad answer looks like:
SignalWeightThe question it asksGood sign (leans Eligible)Warning sign (leans Suppress)
Profitability35%Does the player win or lose over time?Loses over time, so bonusing is sustainableWins over time, so bonusing eats into margin
Bonus Effectiveness30%Do bonuses lead to extra real-money play, or just replace money the player would have spent anyway?Bonus drives extra real-money activityBonus replaces the player's own spend
Bonus-to-Deposit Ratio25%Is the player taking a normal amount of bonus for what they deposit, compared with others?In line with peersBonus-heavy for what they put in
Bonus Share10%Is the player mostly funding their own play, or leaning on bonuses?Self FundedBonus Dependent, likely just chasing bonuses
Profitability carries the most weight because it's the clearest sign of whether it's worth keeping the bonuses going. Bonus Share counts the least, because on its own it only tells you how a player funds their play, not whether bonusing them pays off. ### Verdict Scoring Threshold Each active signal assigns a score of 1.0 for a good sign (leaning Eligible) or 0.0 for a warning sign (leaning Suppress). The combined score is the weighted average of all active signals divided by their total combined weight. A combined score of 0.50 or higher produces a verdict of **Eligible**, while a score strictly below 0.50 produces **Suppress Bonus**. ### How each score is worked out, and what it needs Each signal only runs when the data behind it is coming through your integration. If that data isn't there, the signal sits out. And if fewer than three signals can run, the player comes back as Unknown. The model would rather say "I don't know" than guess. **Profitability (35%)** Evaluates whether the player wins or loses over time across a 90-day window (with a 180-day fallback). For players with 10 or more active betting days, it calculates their proportion of winning days (days where wins exceeded bets); for 5 to 9 active days, it evaluates net gaming revenue. It needs casino and sport betting and win records. It sits out if the player has fewer than 5 active betting days. **Bonus Effectiveness (30%)** Evaluates whether bonus grants drive incremental real-money betting activity over a 180-day history. It compares the player's real-money wagering across periods with bonus grants against periods without bonuses. If real-money play increases around bonuses, the bonus is effective; if real-money play declines or stays flat, the bonus simply substitutes for the player's own funds. It requires at least two bonus grants and sufficient active betting history. **Bonus-to-Deposit Ratio (25%)** Compares the number of bonuses a player received against the number of deposits made over the last 90 days (with a 180-day fallback), benchmarking them against other players on the same brand. Players exceeding the 75th percentile are flagged as over-bonused. It needs bonus redemption and deposit records. For edge cases where deposit or bonus counts are zero, the ratio evaluates deterministically: players with deposits and zero bonuses are marked In line with peers, players with bonuses and zero deposits are flagged Over-Bonused VS. Peers, and players with neither are treated as Unknown. **Bonus Share (10%)** Measures how much of a player's play is funded by bonuses rather than their own money, over the last 30 days. For this to work, the player's bets need to be cleanly split into real money and bonus money. If a brand's bets can't be split cleanly, this signal sits out. That's the single most common reason a whole brand shows up as Unknown. #### What the model needs from your integration | Data coming through the integration | What it powers | | ------------------------------------------------- | ------------------------------------------------------------------------------------------------------ | | Redeemed bonuses | Effectiveness, Bonus-to-Deposit, Bonus Share | | Bets cleanly split into real money vs bonus money | Bonus Share | | The value of each bonus | Used in general financial reporting and bonus ROI analytics, but not required by the eligibility model | | Bets and wins (casino and sport) | Profitability | | Deposits | Bonus-to-Deposit Ratio | | Recent activity | Whether the player is scored at all | ### When the answer is "Unknown" A player shows as Unknown when fewer than three signals could run. This is a real "we can't tell", not a quiet "no". Treat it as a cue to use your normal rules, not a reason to hold back. It usually comes down to one of these: * The player is new or barely active, so there isn't enough recent history to read. * The brand's bets can't be cleanly split into real and bonus money, so Bonus Share and the other real-money reads can't run. This is the most common cause at the brand level. * Bonus Effectiveness isn't available for that player or brand yet. When that happens, the other three signals carry the verdict between them, and the player picks up the fuller picture automatically once Effectiveness is available. There's nothing you need to do. * The player was previously active but has ceased activity. When a player drops out of active scoring, their bonus eligibility user properties are cleared to avoid acting on stale decisions. ### Each of the four signals also works on its own The combined verdict is one way to use this, but each of the four signals is also a player property in its own right. You can use any of them directly without going through the verdict. All four sit alongside the verdict in Lightdash and the user-property feed, ready to segment, target, or report on.
A few ways people use them on their own: * Bonus Share (Self Funded or Bonus Dependent) tells self-funding players apart from bonus-chasers, which is handy for VIP, retention, or abuse checks. * Bonus Effectiveness (helping or replacing spend) gives you a responsiveness group: the players who actually react to bonuses. Useful for targeting and planning campaigns. * Bonus-to-Deposit Ratio (in line or out of line) is a quick read on bonus efficiency and possible abuse. * Profitability (wins or loses over time) is a straight value-and-margin read, useful anywhere you'd otherwise reach for a value score. #### User Properties in the Back-Office All signals are exposed under the Core property group for audience segmentation and journey branching: * **Core: bonus eligibility - verdict**: `Eligible` or `Suppress Bonus` (unset if Unknown). * **Core: bonus eligibility - share result**: `Self Funded`, `Bonus Dependent`, or `Unknown`. * **Core: bonus eligibility - effectiveness result**: `Bonus driving extra Real Money activity`, `Bonus not driving extra Real Money activity`, or `Unknown`. * **Core: bonus eligibility - ratio result**: `In line with peers`, `Over-Bonused VS. Peers`, or `Unknown`. * **Core: bonus eligibility - profitability result**: `Loses over time`, `Wins over time`, or `Unknown`. Each one refreshes daily, is per player, and shows Unknown when the data isn't there, exactly like the combined verdict. ### Using it in your CRM The output is available to your CRM journeys, host tools, Lightdash reports, and the user-property feed. Common ways to use it: * Only run a bonus journey for players marked Eligible. * Send Suppress players down a non-bonus path instead, such as a tournament or a content offer. * Let Unknown players fall through to your existing rules rather than blocking them. #### Viewing Eligibility in the Back-Office * **Player Profile**: In the user profile screen under the Bonus Eligibility card, operators can view the current verdict, suppression reasons, individual pillar outcomes, assigned weights, and the number of contributing signals. * **Reports**: Estate-wide distributions of player eligibility and model coverage are available in the Back-Office under Reports → Insights → Bonus eligibility. The verdict is always a suggestion. Your own campaign rules, budget limits, and responsible-gambling and affordability controls always come first. ### What good looks like You'll know the model is doing its job when, over time, your bonus spend and the number of bonuses go down while play stays flat. In short, you spend less on bonuses without losing activity, because you've stopped bonusing the players who don't need it, don't respond to it, or already cost you money. ### Common questions **How often does it update?** Every day. **Can I override it?** Yes. It's only a suggestion. Your campaign rules and your responsible-gambling and affordability controls always take priority. **Why is one of my best players marked Suppress?** Usually because of profitability. A player who keeps winning gets flagged to protect your margin, even if they play a lot. The reasons on the row will show you which signal drove it. **Why is a player Unknown?** Fewer than three signals had enough data, usually because of thin history or because Effectiveness isn't available for that brand yet. It isn't a suppression. **Why is the verdict property blank in the player's profile?** When a player is Unknown (due to insufficient activity or fewer than three active signals), the `Core: bonus eligibility - verdict` property is left empty rather than storing a text value. In CRM journeys, filter for the presence or absence of the property. **Does Suppress mean the player is a problem?** No. It just means another bonus isn't the right move for that player right now. **Does this model calculate the bonus amount?** No. The Bonus Eligibility Model only determines whether a bonus should be granted or suppressed. To calculate personalized deposit and bet amounts or dynamic bonus values, see the [Recommended Bet and Deposit Amounts](/welcome/products/ai-models/recommended-bet-and-deposit-amounts.md) guide. **Does Bonus Eligibility block manual bonuses in the Back-Office?** No. The model is advisory and never blocks or grants bonuses on its own. Operators can always issue bonuses manually in the Back-Office or configure custom overrides in campaign rules. **Can I segment audiences using individual pillar results?** Yes. All four individual pillar outcomes (`Core: bonus eligibility - share result`, `Core: bonus eligibility - effectiveness result`, `Core: bonus eligibility - ratio result`, and `Core: bonus eligibility - profitability result`) are available as user properties under the Core group for segmentation and journey conditions. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation by asking a question. Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters: ``` GET https://help.smartico.ai/welcome/products/ai-models/bonus-eligibility-model.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.