Relevance
Content matches the customer's interests, context and current intent.
Trend #7 · Product · 2027
Personalisation: one sportsbook, many personalised experiences
Personalisation in 2027 is no longer just about recommending similar events. It is a system that decides what content to show, through which channel, at what moment — and when it is better not to contact the customer at all.
Executive summary
Good personalisation answers not only the question “what should we offer?”, but also “when”, “where”, “why” and “should we make an offer at all?”.
Real time
The next decision reflects what the customer did only seconds earlier.
Discovery
The system does more than repeat known preferences; it also helps customers discover something new.
Responsibility
Risk signals can reduce marketing, bonuses and contact intensity.
Why now
Why personalisation is becoming a core product layer
Modern platforms already use betting history, customer preferences, inventory and real-time events to personalise recommendations, the homepage, bet slip and CRM.
60%
prioritise relevanceIn Optimove's survey of US bettors ahead of the 2026 World Cup, relevance to a favourite team or player was the main factor in opening a message.
Optimove · World Cup 202664%
already see personalisationThat share of respondents in the same study felt their preferred sportsbook already personalised communications well.
Optimove · 202672%
prefer the right timingIn the same study, most respondents preferred to receive communication the day before a match — timing matters as much as content.
Optimove · 20262027
personalisation = orchestrationBetting Trends editorial view: competition is shifting from isolated recommendations to a unified decisioning system.
Betting Trends frameworkPersonalisation loop
Observe → Understand → Decide → Deliver → Learn
Effective personalisation is a closed loop: every customer action becomes a new signal for the next decision.
01OBSERVE
Observe
Capture first-party customer data and behavioural signals in real time.
- Viewed events
- Placed bets
- Search queries
- Channel activity
02UNDERSTAND
Understand
Build a dynamic profile of interests and context.
- Sports affinity
- Market affinity
- Timing
- Risk context
03DECIDE
Choose
Determine the next-best content, channel or whether contact should be suppressed.
- Recommendation
- Offer
- Channel
- No action
04DELIVER
Deliver
Deliver the relevant experience at the right touchpoint.
- Homepage
- Bet slip
- Push notification
- email / SMS
05LEARN
Learn
Measure the outcome and update the customer profile.
- Open / click
- Conversion
- Discovery
- Suppression outcome
Customer model
Not a segment, but a live customer context
Traditional segmentation places a person in a static group. Modern personalisation adds changing context: favourite teams, recent actions, active events, channel, timing and risk status.
A personal profile should not justify unlimited data collection: purpose, customer consent and retention period should be clear from the outset.
Customer context · Demoupdated 8 seconds ago
Sports affinityFootball · Tennisbased on recent first-party customer activity
Market affinityMatch Result · Totalsdynamic preference score
Current contextLive football sessionmobile app · evening
Preferred channelPush → emailconsented channels only
Discovery scoreMediumopen to adjacent content
Risk statusNormalmarketing contact permitted by policy
Use cases
Where personalisation changes the sportsbook in practice
One decisioning layer can work across different touchpoints — from the homepage to CRM — but eligibility and player-protection rules should be applied before ranking.
The homepage becomes different for each customer
Ranking can account for favourite sports, teams, leagues, live status, recent activity and current context.
- Favourite leagues first
- Relevant live events
- Personalised market ordering
- Discovery slots
- Eligibility filtering
Inventory1,240 events
Eligibility816
Affinity ranking120
Context ranking28
Homepage slots8
From popular markets to an individual relevance score
The recommendation engine ranks events and markets by an individual's likely interest, not only by overall popularity.
- Sport / league affinity
- Team preference
- Market preference
- Pre-match vs in-play
- Recent context
Recommendation ≠ unrestricted promotion
Jurisdiction, consent, player-protection and marketing-eligibility rules should be applied before ranking.
Personalisation can extend beyond the event to the combination of selections
Modern recommendation APIs can already suggest personalised bet-builder selections within a specific event.
- Selection relevance
- Event affinity
- Valid combinations
- Confidence ranking
- Dynamic recommendation count
Where does convenience end?
The faster and easier it becomes to build a complex bet, the more important it is to assess product intensity and player-risk context.
Search can account for the customer too
The same query can produce different ordering and suggestions depending on known interests and the current session.
- Personalised autocomplete
- Context-aware ranking
- Related-event discovery
- Popular for you
- Diversity control
Do not turn the profile into a filter bubble
Part of the ranking should remain exploratory; otherwise the system simply repeats the past and stops discovering new interests.
Content, channel and timing are personalised
A next-best-action system chooses not only the message, but also the timing and permitted communication channel.
- Push notification
- SMS
- In-app message
- On-site placement
InterestFootball
Next eventTomorrow
Preferred channelPush notification
FrequencyLow
DecisionSend tomorrow morning
Sometimes the best personalised decision is to show nothing
Responsible decisioning should be able to block marketing, bonuses and high-intensity content when risk status changes.
- Marketing suppression
- Bonus suppression
- Channel reduction
- Content de-intensification
- Player-protection intervention
Protection > relevance
If customer-risk policy prohibits marketing contact, a high relevance score should not override that restriction.
Recommendation engine
From inventory to the next-best experience
The recommendation engine should first establish what may be shown to the customer, then what is most relevant, and only then optimise ranking.
What is available?
Events, markets, content, promotions and permitted communication surfaces.
What is permitted?
Jurisdiction, age, consent, product rules and player-protection policy.
What fits?
Affinity, history, context, similarity and predicted interest.
What should be added?
Discovery slots stop the algorithm from endlessly repeating past preferences.
What should happen?
Show, recommend, message, delay, suppress or trigger a protective action.
Real-time context
The customer profile changes during the session
Modern systems use real-time events — such as accepted bets — to update customer interests and recommendations almost immediately.
Context from the last 90 seconds
Illustrative example of how an activity stream can change ranking within a single session.
20:41:03Opened live football+ affinity
20:41:19Viewed Event 1context
20:41:42Bet acceptedstrong signal
20:42:10Returned to lobbyre-rank
20:42:18New recommendation setdeliver
Accepted ≠ attempted
Event semantics matter. For example, a recommendation system may treat an accepted bet as a preference signal, while a rejected attempt should not necessarily be interpreted in the same way.
Discovery vs repetition
Personalisation should not simply repeat the past
If the recommendation engine only shows what the customer chose before, personalisation becomes an expensive history filter.
Repetition loop
The system reinforces known interests and gradually narrows the catalogue.
The same teams again
The same markets again
The same offers again
Less discovery
Possible fatigue
VS
Relevant discovery
The system combines known interests with controlled exploration of new events and markets.
Known interests
Adjacent content
Contextual discovery
Diversity constraints
Feedback learning
CRM orchestration
Next-best channel, not more messages
Communication personalisation works when the system decides not only the content, but also the channel, timing, frequency and when contact should stop.
On-site
Lobby, banners, recommendations and search.
Push notification
A fast channel for relevant contextual messages.
A longer format for digests and event-led communication.
SMS
High visibility, but requires strict consent and frequency control.
Silence
Sometimes the best decision is to send nothing.
Offers & incentives
A personalised offer also needs limits
The more precisely the system understands the customer, the more important it is to limit how that knowledge is used for promotional pressure.
Relevance without pressure
Useful personalisation helps customers find relevant content and reduces unnecessary noise.
- Relevant sport / event
- Preferred channel
- Reasonable frequency
- Transparent offer terms
- Easy preference controls
What should not be optimised at any cost
Commercial objectives should not override regulatory constraints, customer consent or player-protection rules.
- No marketing where policy requires suppression
- No unconsented channels
- No hidden cross-product bonus logic
- No opaque use of urgency
- No reward optimisation without policy constraints
Privacy & consent
Personalisation begins with permission
The more a recommendation system predicts preferences, behaviour and interests, the more important transparency becomes: which data is used, for what purpose, and how the customer can change their choice.
Purpose
Define in advance why specific data is needed by the personalisation system.
Consent
Account for consent requirements for advertising, tracking and communication channels.
Preference control
Allow the customer to change marketing choices and communication channels.
Withdrawal
Withdrawal of consent should propagate correctly across systems and external partners.
Responsible Personalisation
Risk status should take priority over the marketing score
In regulated betting, the hardest personalisation problem is not maximising engagement, but making the system behave correctly when risk indicators change.
Relevance
Personalised content and permitted communication within consent, frequency and product rules.
De-intensify
Reduce frequency, remove aggressive mechanics and strengthen protective messaging.
Suppress
Restrict targeted marketing and new bonus offers in line with policy and regulation.
Protect
Player-protection action takes priority over CRM, recommendations and revenue objectives.
Measurement
CTR is not enough
If personalisation is measured only by clicks and short-term conversion, the system will inevitably optimise too narrowly.
Relevance
Click-through, recommendation acceptance, search success and engagement with relevant content.
Discovery
How well the system helps customers find new relevant content rather than only repeating history.
Retention
Long-term engagement and loyalty, not just immediate response.
Diversity
Whether the recommendation set remains diverse and avoids creating a filter bubble.
Consent quality
Preference accuracy, opt-out propagation and channel compliance.
Safety outcomes
Suppression accuracy, marketing blocks and absence of conflicts with player-protection rules.
Personalisation stack
From first-party data to decisioning
Personalisation is an architecture, not a single recommendation model. Value appears when profile, inventory, decisioning and channels work together.
Layer 01 · Data
First-party customer events
views · bets · search · sessions · CRM
Layer 02 · Profile
Customer context and preferences
affinity · recency · channel · consent
Layer 03 · Inventory
Events, markets and content
sports · leagues · markets · offers
Layer 04 · Decisioning
Ranking, next-best action and suppression
relevance · diversity · policy · timing
Layer 05 · Delivery
Cross-channel experience
app · web · push · email · SMS
Layer 06 · Governance
Consent, risk and measurement
responsible gaming · privacy · experiments · audit
KPI Matrix
How to measure personalisation maturity
Not all use cases are equally mature: real-time recommendations are already available, while fully autonomous commercial decisioning requires much stronger governance.
| Capability | Business value | Maturity | Governance risk | Key metric |
|---|---|---|---|---|
| Personalised lobby | High | High | Medium | Recommendation engagement |
| Bet recommendations | High | High | Medium–High | Acceptance / diversity |
| Real-time recommendations | High | Medium–High | High | Relevance / timing |
| Personalised search | Medium–High | Medium | Medium | Search success rate |
| Next-best channel | High | Medium–High | High | Response / opt-out |
| Marketing suppression | Very High | High | Very High | Suppression accuracy |
| Autonomous offer optimisation | Potentially High | Early–Medium | Very High | Long-term value + safety |
The matrix is a Betting Trends editorial assessment, not an industry standard.
Build or buy
Where proprietary capability creates a real advantage
Recommendation infrastructure can be bought, but unique competitive advantage often sits in customer data, eligibility rules and orchestration.
Recommendation capability
Ready-made components can shorten time to market and provide specialised models for betting and iGaming.
- recommendation APIs
- CRM orchestration
- real-time event infrastructure
- experimentation tooling
- content-ranking components
Decision policy layer
Capabilities that reflect the operator's strategy, consent model, product rules and responsible-gaming policy.
- Customer data model
- Eligibility logic
- Risk-based suppression
- Cross-product orchestration
- Long-term measurement framework
2027 scenarios
Three personalisation scenarios
These are Betting Trends editorial scenarios, not guaranteed market forecasts.
Personalisation OS
A unified decisioning layer manages the lobby, search, CRM, offers and suppression in real time.
Guardrail-first AI
Stricter privacy, marketing and player-protection rules push eligibility checks ahead of recommendation.
Discovery wins
The best systems begin to optimise not only known affinity, but also discovery, diversity and long-term experience quality.
90-day plan
How to move from segmentation to real-time personalisation
Start not with “let's implement AI”, but with customer signals, eligibility rules and one measurable use case.
Map
Map the customer journey, data events and eligibility rules.
- Map first-party signals
- Audit consent
- Define inventory
- Baseline current relevance
Pilot
Launch one high-value recommendation use case.
- Choose lobby or CRM
- Add an eligibility layer
- A/B test ranking
- Track discovery
Orchestrate
Connect product, CRM and responsible-gaming decisions.
- Connect real-time events
- Introduce suppression
- Measure long-term outcomes
- Create a governance review process
Board questions
7 questions before scaling
If the personalisation team cannot answer them, the problem usually sits in data, governance or measurement rather than the model itself.
01
What exactly is being personalised?
Lobby, ranking, offers, timing, channel, search — or all of them?
02
Which data is actually needed?
Is there a clear purpose for every signal, rather than a habit of collecting data?
03
Where does consent apply?
Can the business demonstrate that channel and tracking choices match the customer's preferences?
04
What happens when a risk signal appears?
Can player protection automatically override a marketing recommendation?
05
Is there discovery?
Or does the model simply show more of what the customer did before?
06
What counts as success?
Clicks, conversion, customer lifetime value, retention, diversity or customer trust?
07
Can the recommendation be explained?
Which signals, rules and eligibility constraints influenced the decision?
Related trends
What to read next
Sources & Methodology
Sources and methodology
This page combines product documentation, industry research, privacy guidance and gambling regulation. The KPI framework, maturity assessment and 2027 scenarios are a Betting Trends editorial model.
Personalised bet recommendations, favourite sports, leagues and markets, pre-match and in-play recommendations, push notifications and pre-filled bet slips.
Open source →Description of a model in which inventory and real-time customer events are used as recommendation signals.
Open source →Data on relevance, timing and bettors' expectations of personalised communication.
Open source →Requirements covering tailored action, marketing suppression when strong indicators of harm are present, and review of automated decisions.
Open source →Updated requirements apply from 19 January 2026, including a maximum 10x wagering requirement and a ban on mixing different gambling products within a single incentive.
Open source →Guidance on consent for technologies used to analyse and predict personal preferences, behaviour and attitudes.
Open source →