How to Deliver Real-Time Sports Personalisation

Future Anthem's Content Recommendations module delivers sportsbook personalisation across every surface, from navigation to bet slip, in real time.

Table of Contents

1. What ‘full-surface personalisation’ actually means

2. The core challenge: familiarity vs discovery

3. How it works

4. Solving the cold start problem

5. Accas and bet builders: the high-margin personalisation gap

6. Personalising navigation and discovery

7. Integration: how recommendations reach your platform

8. Operational impact

9. What’s next?
   
   

Sports betting is dynamic, shaped by player preferences and personal passions. But many sportsbooks rely on static logic that treats all sports bettors the same.

Flagship sporting events are surfaced to everyone in the same way. Navigation is fixed. Bet builders and accumulators are left to the player to construct manually.

The problem is exacerbated during peak moments like the FIFA World Cup, Cheltenham or Wimbledon. It doesn’t matter whether the end user is more interested in football, horseracing, tennis or something else entirely; everyone sees the same thing.

This creates three core problems for operators:

• Low relevance — experienced players must search for what they already know they want;
• Poor discovery — casual players are overwhelmed with generic and undifferentiated content; and
• Missed value — high-margin products like accas and bet builders are under-utilised.

The challenge extends beyond recommending a bet; it is to deliver sportsbook personalisation at scale.

That is the issue Future Anthem’s Content Recommendations module is built to solve with its sports recommendations feature.

"The challenge extends beyond recommending a bet; it is to deliver sportsbook personalisation at scale."

What ‘full-surface personalisation’ actually means

Personalising a sportsbook is a system-level challenge where multiple layers of decisioning must be powered by a shared understanding of each player.

A modern recommendations system needs to operate across every touchpoint:

• Navigation menus and sport ordering — surfacing what matters to players before they search for it;
• Competition and event listings — prioritising the tournaments and fixtures most relevant to their history;
• Market and selection recommendations — presenting the bet types they are most likely to engage with; and
• Bet builders and accumulators — personalising multi-leg construction, not just single selections.

Each layer requires a consistent behavioural view on what players bet on, how they bet, and when they bet. But it’s also critical to consider how those behaviours evolve over time, particularly during high-intensity periods like major tournaments.

The core challenge: familiarity vs discovery

At the heart of sports betting personalisation sits a fundamental tension between showing players more of what they already like, or helping them discover something new.

Favour familiarity too heavily and the experience becomes repetitive. Push too hard on discovery and the relevance drops. But a robust system must do both.

In practice, this means:

• Prioritising familiar competitions and markets for most recommendations, reinforcing what the player already engages with;
• Allocating controlled space for exploring new content, surfacing new content within relevant contexts for the player; and
• Continuously adapting based on recent behaviour signals, and not just historical patterns.

This balance is not just a UX principle; it is encoded directly into the ranking and allocation logic of the system.

How it works

Content Recommendations builds a behavioural representation of each player and creates recommendations that will resonate with them.

1. Player representation

A profile is developed for every player, combining:

• What they bet on (e.g. sports, tournaments, markets, outcomes)
• How they bet (e.g. singles vs accumulators, odds preferences, staking patterns)
• When they bet (e.g. time of day, session behaviour, in-play vs pre-match)

Structured behavioural features are combined with learned betting patterns to capture preferences and implicit relationships, such as which markets tend to be selected together or how players structure their bets.

2. Player similarity

Players are matched with others who behave similarly, not based on identical bets, but on shared behavioural patterns over time.

3. Candidate generation

From these similar players, the platform collects:

• Recently placed bets on upcoming events; and
• Popular selections across the market.

This creates a dynamic pool of recommendations, reflecting both peer behaviour and current market activity.

4. Scoring and ranking

Each candidate is scored using a combination of signals:

• Similarity — how close the neighbour is to the player;
• Recency — how recently the player engaged with similar content; and
• Popularity — how widely a selection is being bet.

These signals are normalised and combined to produce a baseline relevance score.

Additional adjustments are then applied:

• Market diversity balancing — to avoid over-concentration;
• Tournament familiarity weighting — to prioritise known contexts; and
• Favourite team alignment — to increase engagement.

The result is a ranked list of recommendations tailored to each user.

Solving the cold start problem

One of the hardest problems in sports betting personalisation is the cold start challenge: delivering relevant recommendations for new or infrequent events where direct player signals are limited or absent.

This is particularly acute for a new tournament on the calendar, seasonal events like Cheltenham, or global competitions like the World Cup that produce sudden spikes in activity.

Future Anthem’s Content Recommendations handles this in two ways:

a) Behavioural generalisation

Player preferences are learned at multiple levels — sport, competition, market —allowing the system to infer interest in new events within familiar structures. A player with a strong history of UEFA Champions League betting, for example, is likely to engage with a new European club competition even without prior direct signals.

b) Fallback ranking

When direct signals are weak, recommendations are generated using recent activity patterns, market-level popularity and content similarity, ensuring the output remains relevant without defaulting to generic or manually curated content.

The result: every player receives relevant recommendations from day one, in real time, with no manual intervention required.

Accas and bet builders: the high-margin personalisation gap

Multi-leg bets are critical for both engagement and margin, but they are among the hardest products to personalise effectively.

Most approaches attempt to construct synthetic combinations based on individual selection scores. Future Anthem’s Content Recommendations module takes a different approach: rather than constructing synthetic combinations, the module uses real tickets placed by similar players.

Each ticket is evaluated based on:

• Relevance (how well each individual leg matches the player’s known preferences)
• Structural similarity (how closely the combination mirrors the player’s historical betting behaviour)
• Diversity (spread across markets and competitions to avoid repetition)
• Favourite team alignment (prioritising combinations that include teams the player already follows)

The result? Accumulator and bet builder recommendations that are more intuitive, credible and likely to drive meaningful engagement by reflecting how players naturally construct bets.

Personalising navigation and discovery

Personalisation can also be applied to the structure of the sportsbook itself.

That means:

• Sport ordering — presenting the sports most relevant to each player at the top of the navigation menu;
• Competition prioritisation — surfacing the tournaments and leagues each player is most likely to engage with; and
• Event sequencing within pages — ordering fixtures within pages based on individual player preference.

The effect is a sportsbook that feels instinctively relevant and reduces friction by ensuring the most relevant content appears earlier in the journey, improving both engagement and conversion.

Integration: how recommendations reach your platform

From an operator’s perspective, the Content Recommendations module is designed to integrate into existing platforms with minimal disruption to existing roadmaps.

Recommendations can be delivered through:

• API endpoints, for real-time surfaces across web and app;
• CMS-driven carousels, for front-end placement without engineering overhead; and
• CRM and marketing integrations, for off-site engagement across email and push channels.

This means personalisation can be deployed consistently across every channel — web, app, and marketing — from a single integration point.

Operational impact

Future Anthem’s Content Recommendations module allows operators to:

• Increase engagement through higher relevance across every surface;
• Improve discovery without sacrificing familiarity or overwhelming the player;
• Drive adoption of high-margin products like accas and bet builders; and
• Maintain performance during peak events.

Crucially, the module is designed to handle large volumes of players and events with no manual intervention required. That is what sportsbook personalisation looks like in practice.

The impact is measurable, and across deployments the pattern is consistent:

• Around 30% engagement at the bet selection/outcome level
• Closer to 40% engagement at the market level
• A 5-7% uplift in daily football stakes
• On average, recommended bets generate roughly 20% more revenue per bet than non-recommended ones
• They also carry over 4x higher median odds and around 2x more legs per bet

What’s next?

The next frontier for Content Recommendations is real-time intent modelling.

By incorporating live signals such as betslip activity, systems can move from: “What does this player usually do?” to “What is this player trying to do right now?”

This shift enables even more precise and timely recommendations, particularly in fast-moving in-play environments, delivering relevance at the exact point it matters most.

Ready to deliver real-time personalisation at scale — through the new season and beyond? Book a demo