The Hybrid method is your go-to solution for automating personalisation efforts while consistently delivering exceptional results. Think of it as your universal model, capable of adapting to diverse scenarios and providing top-tier recommendations with minimal intervention required.

The Hybrid method is a set of sub-models, each crafted and personalised based on the response rates observed from individual models. Over time, this method learns from customer interactions and continuously refines itself by optimising the weights assigned to each sub-model.

It encompasses a comprehensive range of models that utilise all accessible customer and game information, including game launches, financial data, and game metadata sourced from operators and Opti-X.

Excellent for “just for you” or “recommended for you” placements, especially when you want a multi-model approach that continuously refines itself.

The Layout model is an alternative hybrid approach tailored explicitly for use within Placements that you plan to use in Intelligent Layouts.

An alternative hybrid specifically tailored for Intelligent Layouts—ideal for experimenting with multiple recommended modules in dynamic on-site layouts.

The Similar Items method employs a straightforward approach, comparing different items.

Here’s how it works: Opti-X gathers a set of interactions (such as actions or behaviours) from users who have played Game A, labelling it Set A. Interactions from users who have played Game B are collected as Set B. The model then calculates a score by examining the intersection and union of these two sets.

A high score indicates a significant overlap in user interactions between the two games, suggesting a strong similarity in how users engage with them.

Perfect for “If you like this, you’ll also like…” style recommendations (e.g. additional slot or casino games).

The Similar Users method leverages user interactions with games or items by analysing engagement patterns among different users.

Example: Jane, John, and Paul each interact differently with various games. Jane and John engage with Games A, B, and C, while Paul interacts with Games A and B but not C. Since Jane and John’s patterns match more closely, the model might recommend Game C to Paul.

Great for "similar users to you are playing" and introducing games that are popular with players whose betting or playing patterns match the current user’s behaviour.

The Dummy method assigns scores randomly. It’s ideal for test scenarios or showcasing special events before other models are ready.

Great for test scenarios or highlighting certain events via random scoring before other models are live.

The Financial method operates much like Similar Users but also weighs financial data. By considering user engagement patterns alongside spending or win/loss information, it fine-tunes personalised results.

For example, if Jane, John, and Paul have varied financial behaviours as well as gaming interests, the method identifies those similarities to tailor more accurate recommendations.

Incorporates spending or win/loss info alongside user engagement—can help target higher-value players or big winners.

The Popular method is designed to capture recent user interactions within a defined timeframe, often the last 15 minutes.

Here’s how it works: The algorithm analyses user actions such as game plays within that short window, prioritising items receiving the most engagement.

Best for “hot right now” widgets showcasing games with the most engagement in the past few minutes.

The Trending method operates similarly to the Popular method but over a slightly broader timeframe (for instance, comparing the past 15 minutes to activity over the previous 2 hours).

This approach is recommended for brands experiencing high activity, as it balances immediate spikes with sustained interest.

Suited for highlighting games with sustained popularity over a broader time window (e.g. 15 minutes vs. the last 2 hours). For example "games trending right now"

The Popular Near You method also builds on popularity but takes the user’s country into account. It looks at recent interactions in a particular region over a chosen timeframe.

Surfaces region-specific popular content, handy if you want to tailor “top games” by country.

The New Games method is a content-based model that spotlights newly launched items. Instead of relying on user-item interactions, it focuses on the attributes of the games themselves.

It highlights content released within a set period (for example, a week or a month after launch), adjustable according to brand preferences.

Perfect for a “Newly Launched” or “Fresh Releases” widget, focusing on recently added or launched games

The New Games Boost method tackles the “cold start” problem by giving newly launched items a temporary boost until they accumulate sufficient user interaction data.

Here’s how it works: New items receive pseudo-recommendation scores so they don’t languish unseen until enough user interactions naturally build up.

Addresses the “cold start” problem—use when you want new games to get a fair chance in recommendations before enough interaction data accumulates.

The Release Date method ranks games according to their launch dates, assigning higher scores to the newest releases—an effective way to spotlight fresh content.

Lists games in descending order of their launch date—simple way to spotlight the newest titles on the platform.

The Themed Content method recommends items aligned with a customer’s specific thematic preferences.

How it works: By analysing plays or views and matching them to content with similar themes, the system personalises recommendations around those interests.

Ideal for holiday or seasonal promotions—surfaces thematically related games (e.g. spooky slots at Halloween).

The Behavioural method uses content analysis from text fields in the inventory (such as the name, theme, or description), ranking items higher based on how closely they match a user’s known preferences.

Here’s how it works: For example, if Megan frequently plays Irish-themed games, the system identifies other similar Irish-themed items and ranks them higher in her recommendations.

Great for matching a user’s taste in themes (e.g. “Irish-themed games” or “mythology-themed slots”) based on inventory text fields.

The Recently Played method focuses on how recently a user interacted with each item, prioritising those engaged with most recently.

Example: If Jane played Game A today and Game B yesterday, B outranks older interactions, reflecting her more current interests.

Show games the user engaged with most recently, handy for “pick up where you left off” experiences.

The Previously Played method prioritises items a user has engaged with before, sorted by overall frequency of engagement.

Here’s how it works: If a user played Game A seven times and Game B three times, A is ranked higher than B—regardless of recency— because of the overall higher play frequency.

Ranks games by overall frequency rather than recency—ideal for reminding users of long-time favourites they may have dropped.

The Previous Win method uses a user’s financial history to identify which games they’ve won the most from, ranking those higher in recommendations.

Here’s how it operates: It checks the total winnings for each game and prioritises those with higher payouts.

Excellent for reinforcing success patterns—“You won on these games before; maybe try again?”

The Favorite Game method identifies a user’s most preferred game based on their previous interactions. It’s similar to Recently Played, but zeroes in on the game showing the strongest consistent usage.

Pinpoints the single most-played or most-frequently launched game, perfect for a “Your Favorite” quick-launch tile.

The ‘Remember This?’ method highlights items a user hasn’t engaged with for a set period (e.g. 14 days).

If John has played Game X 50 times and Game Y 70 times, but hasn’t touched them in the last 30 days, the method will reintroduce Y first (due to higher total plays) and then X.

Useful for re-introducing games a user hasn’t played for a while—nudging them to revisit past favourites.

The Hybrid method is your go-to solution for automating personalisation efforts while consistently delivering exceptional results. Think of it as your universal model, capable of adapting to diverse scenarios and providing top-tier recommendations with minimal intervention required.

The Hybrid method is a set of sub-models, each crafted and personalised based on the response rates observed from individual models. Over time, this method learns from customer interactions and continuously refines itself by optimising the weights assigned to each sub-model.

It encompasses a comprehensive range of models utilising all accessible customer and event information, including bets, financial data, and game metadata sourced from both operators and Opti-X.

Suited for broad “Recommended Bets” modules, combining multiple signals like previous bets, popular events, and more.

This method compares in-play betting categories among different users to recommend relevant categories.

For instance, if Jane frequently places in-play bets on football and tennis, while Alex only bets on football, the model may recommend tennis to Alex due to the overlap in preferences.

The method returns categories regardless of the target item in the user interface.

Recommends in-play betting categories to users who match the profile of other in-play bettors—useful for real-time bet prompts.

This method compares pre-event betting categories among different users to recommend relevant categories.

For instance, if Jane places pre-event bets on football and tennis, while Alex only places pre-event bets on football, the model may suggest tennis to Alex due to their overlapping interests.

The method still returns categories regardless of the user’s chosen target item.

Great for users who regularly place bets on specific sports before events start, prompting them to explore similar categories.

This method works much like adding items to a shopping basket. Once users add an item to their bet slip, the method identifies and recommends similar selections based on that initial choice.

Acts like a shopping basket approach—once a user adds one bet, the system instantly recommends related ones.

This approach returns categories, classes, or types in the recommendation, functioning like Similar Users. It leverages betting patterns to propose additional categories.

Example: If John bets on football and horse racing, while Jane bets on tennis, football, and hockey, the model may recommend tennis to John and horse racing to Jane.

Returns higher-level groupings (e.g. “Tennis,” “Football,” “Hockey”) if similar user patterns indicate a preference for certain sports categories.

The In-Play method recommends selections for ongoing events, originally designed for football. It compares the user’s historical betting activity with live events in real time.

For example: If a target user has bet on Arsenal in the past, and other similarly behaved users also placed in-play bets on Arsenal winning the first half, the method might recommend that selection to the target user.

Perfect for dynamic, live bet prompts, especially during big matches or tournaments.

The Dummy method assigns scores randomly. It’s ideal for showcasing particular events (by adding sports rules) or testing before your other methods are fully trained.

Randomly assigns scores—useful for event showcases or test scenarios pre-launch of real methods.

Similar to Dummy, this version assigns scores randomly but is especially handy for showcasing specific categories (again by adding sports rules) or for testing.

Similar to Dummy but focusing on specific categories via sports rules—again, good for highlight reels or testing.

The Extra Leg method enhances user experience by suggesting additional elements for multiple selections. If a user has chosen 2 or 3 items for a bet, this method recommends another relevant element based on similar preferences or patterns.

Perfect for multi-selection bets (2 or 3 picks). Suggests an additional relevant “leg” to help build an accumulator or parlay.

This method is used when users place multiple bets—2, 3, or more selections—and is especially valuable for building Same Game Parlays (SGP). Users can craft wagers involving multiple aspects of the same game.

For constructing multi-selection or same-game parlays (SGP)—especially popular where combined bets within a single event are common.

The Popular method captures recent user interactions within a short timeframe (often the last 15 minutes).

Here’s how it works: The algorithm analyses bets (or other relevant actions) in that short window, ranking items that garner significant engagement.

Highlights “hot” or most-bet-on events in a short timeframe—great for quick-lobby features or “Trending Bets” widgets.

The Trending method is similar to Popular but compares user engagement over a slightly longer period (e.g. 15 minutes vs. the last 2 hours).

It’s recommended for brands experiencing higher activity within that timeframe.

Similar to Popular but covers a slightly broader period—handy for capturing both short bursts and steady interest.

Personalised Popular recommends bets based on a user’s betting activity over roughly the past hour, factoring in trends from other users.

Note: If no bets have been placed for a certain sport (e.g. ice hockey), the method may produce random results or none at all, so applying strict sports rules might yield empty returns.

Balances a user’s recent betting activity with overall popular trends, though no bets in a given category can produce random/empty results.

Explicit Popular also recommends bets based on activity within the last hour but applies sports rules before ranking the final results.

If no bets have been placed in your chosen category (e.g. ice hockey) but you filter for that category, the method returns random or empty results after applying the filter.

Applies sports rules first (e.g. sport = “ice hockey”), then ranks the final results by popularity. Good when you need filters plus popularity.

Behavioural 2.0 is an automated hybrid of content and popularity models. It analyses text fields (e.g. team names, leagues) and users’ bet history.

Here’s how it works: If Jane often bets on Liverpool in the Premier League, the model identifies that preference and might suggest Liverpool in the FA Cup or other relevant matches.

An automated hybrid that uses textual data (team names, leagues) plus popularity—ideal for advanced, personalised bet suggestions.

This method identifies a user's top 5 most frequently bet-on teams (currently supports football only) and recommends upcoming matches or betting options involving those teams.

If a user has fewer than 5 unique teams in their betting history, the system supplements with globally popular default teams to ensure a consistent recommendation experience. Betting selections for each team are ranked by market popularity, ensuring recommendations are relevant and widely followed.

Ideal for sportsbook users with strong team preferences. Helps re-engage fans by recommending matches involving their favourite teams, for example, in a "Your Teams" or "For The Fans" section.

The Prior Win method recommends bets based on a user’s historical winning experiences. If you’ve won betting on certain teams or players in the past, it may suggest similar selections in future events.

Surfaces bets reflecting the user’s historical successes, reinforcing winning patterns.

The Hybrid method is your go-to solution for automating personalisation efforts while consistently delivering exceptional results. Think of it as your universal model, able to adapt to diverse scenarios and provide top-tier recommendations with minimal intervention.

This method is a set of sub-models, each crafted and personalised based on the response rates observed from individual models. Over time, it learns from customer interactions and continuously refines itself by optimising the weights assigned to each sub-model.

It encompasses a broad range of models using all accessible customer and event information: views, add-to-cart actions, financial data, and product metadata.

Perfect for site-wide “Recommended For You” sections—an all-rounder that refines itself with each interaction.

The Similar Items method compares two distinct items based on user interactions.

Here’s how it works: Opti-X gathers interactions (e.g. actions or behaviours) from users who engaged with Item A, labelling them Set X, and from users who engaged with Item B, creating Set Y. The method calculates a score based on the intersection and union of these sets.

A higher score indicates a stronger overlap, implying greater user similarity between the two items.

Use for “Customers who viewed this item also viewed…” or “You may also like…” product widgets.

This method focuses on user events, analysing how different users engage with items. You can also choose specific events that you want to be included or remain with all the available events. Example: Jane and John abandon chairs, storage solutions, and cushions, while Paul abandons chairs and storage solutions only. The model might suggest cushions to Paul, noticing that Jane and John share a similar pattern.

Choose User Activity type: add/remove to cart for cart recovery campaigns—promote items frequently abandoned by lookalike shoppers.

Choose User Activity type: product views for retargeting or “since you viewed these products, you might like…” suggestions.

The Add Ons method addresses the “frequently bought together” scenario. It identifies additional items users tend to purchase alongside or after certain items.

Addresses “frequently bought together” scenarios—ideal for upselling complementary items in the cart (increasing AOV).

This method surfaces products that users often view in the context of a specific action — such as purchasing, abandoning, or wishlisting a product. It uses collective behaviour patterns to show what users are likely to explore after key shopping events.

Example: If a user abandoned "white popcorn kernels" in their cart, the system might recommend a "popcorn storage jar" because other users who abandoned the same product tended to view that item.

Viewed After Purchase: Great for thank-you pages or post-purchase emails. E.g., “Complete the look” or “Customers also explored…”

Viewed After Cart Abandon: Use in cart recovery emails or exit-intent popups. E.g., “Others who left this behind looked at…”

Popular is a personalised recommendation method that tailors results to each user based on their recent activity up to a chosen event depth parameter in Method Settings. It identifies a user’s top categories or brands and ranks products within those categories by popularity. You can also choose to include only specific interaction types, based on the user events you track.

This method differs from Explicit Popular, which applies a global ranking across all users using predefined filters. Popular dynamically adjusts to each user's preferences, making it ideal for delivering relevant and timely product suggestions.

Showcases top-trending products over the past 30 days—e.g. “Best Sellers” or “Trending Now.” Choose User Activity type: product views to focus on most-viewed items—great for “most popular” features or email campaigns highlighting “top-browsed products.”

Choose User Activity type: add/remove to cart for cart to surfaces products often abandoned in carts—useful for follow-up offers or “people left these behind” reminders.

Explicit Popular ranks products based on overall popularity - what’s popular among all users. It uses your top 1 or 5 categories or brands with all candidate products, ranks all of them across all users.

Best for global or sitewide recommendations like homepage carousels, newsletters, or "Trending Now" sections — especially when there’s no user context or minimal personalisation is needed.

Great fallback when you want to show what’s generally popular across your catalog - available as fallback option.

The New Items method is a content-based approach spotlighting items recently added to the catalogue. It doesn’t rely on prior user-item interactions, but rather on item attributes.

By default, it highlights items added within a given timeframe (e.g. a week or a month). You can customise this period according to brand needs.

Ideal for “New Arrivals” or “Fresh In” sections, attracting attention to brand-new catalogue items.

The Release Date method ranks products based on when they were added to the catalogue, giving priority to the newest items—perfect for showcasing fresh arrivals. Displays products in descending order of release date—an easy way to highlight the latest additions to your store.

Ideal for “New Arrivals” or “Fresh In” sections, attracting attention to brand-new catalogue items.

This method focuses on items a user has already purchased, sorted by how frequently or consistently they buy these items.

Perfect for reminding users of items they reorder frequently (e.g. groceries, cosmetics, subscriptions).

The Recently Interacted with method ranks items based on how recently the user looked at them. Items seen more recently appear higher in the recommendations.

Great for a “Continue Browsing” or “You just looked at these” section, especially if they left without adding to cart. Great for "Remember This?" blocks in your templates - recommends products that a user interacted with in the past but hasn't viewed or engaged with recently — helping to bring back items they may have forgotten about or were previously interested in.

This level of customesation could be achived with tailoring additional method parameters to consider only specific user events within last range of events.

Abandoned Cart Items is a dedicated recommendation method designed to help recover dropped purchases. It is based on the Recently Interacted model, but offered as a separate method for convenience. It specifically surfaces products that the user added to cart but did not purchase, using cart-related events from your tracking setup.

Best used for: cart abandonment emails, push notifications, or in-session reminders. You can also combine this method with the Fulfilment Parameter to tailor recommendations based on how the user intended to receive their order — for example, showing different reminders for delivery vs. click & collect shoppers.

This makes your messaging more relevant and increases the chances of conversion.

The Dummy method assigns scores randomly, useful for scenarios where you want to showcase certain items (via tag rules) or run tests before other methods have fully trained.

Assigns random scores. Useful for test scenarios, event-based promotions, or flash sales where you don’t need a real preference algorithm.