Game Recommendations - FAQ
Everything you need to know about integration, implementation and performance insights for our Game Recommendations product
General Overview
Are recommendations based only on the previous day’s transactions, or on a player’s full transaction history?
Recommendations are generated using the player’s entire transaction history, not just from the previous day. While recent transactions have a higher influence, the system takes into account broader patterns from all past activity to ensure recommendations are both relevant and personalised. For inactive players, the recommendation engine will reference their last meaningful activity, even if it was several days or weeks ago.
What happens to Game Recommendations if a player has minimal or sporadic activity?
They will receive recommendations based on their prior activity that exists. If a player has no transaction history, they receive the new player recommendations. These are games most likely to convert new players (i.e. all new players get the same recommendations) designed to activate brand new players who have no prior activity.
If a player becomes inactive, will they see the same recommendations upon return?
For short inactivity (2–3 days), players generally see the same recommendations when they log back in. Recommendations are re-calculated for every player every 7 days maximum, so even during inactivity a player’s recommendations can change.
Use Cases
What channels can I run these campaigns on?
Recommendations are designed to be displayed directly on the front-end (homepage, lobby, category pages) and extended into CRM campaigns (push, email, in-app messaging).
Can recommendations be personalised by attributes like volatility or theme?
Yes. With Metadata Driven Personalisation, operators can create carousels tailored to attributes such as volatility, mechanics, or theme.
Integration & Implementation
How long does it take to typically integrate?
Integration usually takes 6–8 weeks, depending on data readiness and system complexity.
What data is required for Game Recommendations to work?
Player-level session and transactional data., Game metadata (theme, RTP, volatility, mechanics) is helpful, however Anthem has an in-house meta-data store which can fill in this information in most cases
What is the update frequency
As frequently as data is received
Are recommendations unique to each operator, or shared?
All data and outputs are operator-specific. No data is co-mingled across clients.
What is the most common way to integrate recommendations?
Via Future Anthem’s rest API, which provides a single flexible endpoint which provides all recommendation categories configured by your request.
Do you offer sandbox access?
Yes. A sandbox API environment is provided during onboarding for testing and validation.
What support is available during integration?
Whilst we have easy to follow integration guides, your Future Anthem account manager will engage product and engineering specialists as required.
Target / Control Group Logic
How are Game Recommendations handled for Control group players?
Operators can decide how recommendations appear for control group players ("control_group": True). Amplifier AI supports the following options:
- EMPTY: No recommendations displayed.
- RANDOM: Randomly selected games from other players’ recommendations.
- POPULAR: Games from other players’ recommendations, weighted by popularity.
- NONE: Players in the control group receive the same recommendations as if they were in the target group.
Are all players assigned to either the target or control group once they’ve had a transaction?
Yes. Every player who completes a transaction is assigned to either the target or control group based on the date of their first transaction. After 60 days in a group, players are reassigned without prior bias.
If a player has transactions on a given day but isn’t in either target or control group, what does that mean?
They are a new player without prior transactions. Existing players with activity are always assigned to a group.
Does the target/control logic run before recommendations each day?
Yes. Group assignment runs daily. Players are only reassigned after 60 days, ensuring consistency. This process is separate from recommendation generation.
If a player joins and has transactions on September 1st, when will they be assigned to a group?
Typically within 1–2 days, depending on when the transaction data is delivered and processed.
How does Amplifier AI ensure a fair distribution between target and control groups?
Stratification is used so players of varying value are evenly distributed. New players without history are assigned randomly.
What is the ratio of players in the target and control groups?
The ratio is configurable. Amplifier AI recommends a 90:10 split, but operators can adjust this as required.
Performance & ROI
What uplift can I expect from Game Recommendations?
Results vary by market and operator, but customers typically see measurable improvements within the first weeks of go-live:
- 3–4x ROI
- 5–8% increase in NGR
- 2–2.5x increase in games played
These results highlight the impact of personalised recommendations on both engagement and revenue performance.
How is performance measured?
Amplifier AI uses a target vs. control framework, complemented by pre vs. post analysis. Dashboards track key KPIs such as active players, stakes per player, spins per player, and hit rates.
How do I track performance?
Dashboards are released 1 week post go-live, providing visibility of engagement trends, hit rates, and impact on player behaviour.
Do you provide case studies?
Yes. Case studies are available on request, demonstrating how operators have driven measurable ROI with Game Recommendations.
How do I get support or make a feature suggestion?
Visit our 'here to help' form and we will respond accordingly.