AI has dominated marketing technology conversations for the past two years, and personalisation is one of the areas where its promise has been loudest.
Yet across media, entertainment, sports, and gaming, the brands getting measurable results from AI are not the ones with the most sophisticated models. They are the ones with the cleanest data foundations, the clearest use cases, and the strongest measurement discipline. AI amplifies whatever strategy and infrastructure sit underneath it, for better or worse.
How brands are building those foundations, measuring what works, and deploying AI responsibly is one of the central themes of a new report from Braze, the Media & Entertainment Personalisation Report.
Drawing on direct practitioner experience from streaming, sports, gaming, and music brands across Asia, ANZ, and the GCC, the report examines the market pressures driving change in the industry, the capability gaps holding brands back, and the personalisation strategies, data foundations, and AI applications delivering measurable results.
This article, which features excerpts from the main report, looks at the tooling decisions that enable or constrain personalisation at scale, the metrics that prove it is working, and the role AI is actually playing in the brands seeing returns.
Moving from fragmented tools to unified platforms
Most media and entertainment brands have accumulated tools over time, adding point solutions as new channels emerged and new needs arose. The result, for many, is a patchwork of systems that were never designed to work together.
Close to a third of media and entertainment marketers still use channel-specific solutions that fragment the customer experience. The question now is how to move from that patchwork toward something more unified, and what successful implementation actually looks like.
The shift requires both technical integration and organisational alignment. Brands need to simplify their tech stacks, connect their data, and make sure teams are set up and confident using the platform day to day.
Centralised data warehousing is now a necessity. The warehouse must connect directly to activation platforms such as CRM and automation tools, and success looks like immediacy; data is captured, interpreted, and acted on quickly.
Consolidation efforts tend to fail when they are sold into the business as cost reduction. The business case is realised when brands can launch personalisation at speed and scale. Most businesses are still not fully using their martech capabilities because they bought tools without a solid view of their strategy.
Consolidation should be treated as capability building rather than technology replacement. Otherwise the outcome is automated fragmented personalisation.
Moving to a unified platform does not necessarily mean moving to a single provider. Signing up for everything on one tech stack risks ending up with some, or sometimes none, of the elements being best in breed.
A more effective approach is an open architecture where the best solution for each element can be plugged in at any given time. A brand might keep its data warehouse and recommendation engine while routing all customer engagement through a single orchestration layer, combining specialist capability with coordination.
The technology landscape is changing quickly, and brands need the ability to move with it rather than locking their roadmap to a single provider.
"You don't want to be stuck in the mud of 'I've gone all in with this provider, and therefore my roadmap becomes their roadmap for the next ten years.' You want the openness of a buy model for modules that you add on, and the ability to replace them when you need to," Anthony O'Byrne, Managing Director of Growth at Kayo Sports, said.
The metrics that show personalisation is working
Proving that personalisation works requires connecting it to outcomes that matter to the business. Engagement metrics like open rates and click-through rates can indicate direction, but they are not the destination.
The metrics that demonstrate real impact are retention, repeat engagement, lifetime value, frequency, churn reduction, and conversion velocity. The ability to connect personalisation to revenue is what defends the investment.
The most reliable indicators are behavioural deltas rather than vanity metrics. When personalisation works, acquisition becomes cheaper because retention improves. That is the business outcome brands should measure against.
Measuring personalisation effectively requires a testing culture. Control groups on every initiative, whether on-product, off-product, across communications or back-end services, make it possible to isolate the impact of personalisation from other variables.
Revenue is the metric that matters most to the business, but it is not always the metric that aligns teams internally.
Finding a North Star metric that correlates strongly with revenue and that everyone across the organisation can rally around is a practical step. For a sports streaming platform, that metric turned out to be incremental viewing.
Tracking how much additional viewing each initiative drives across product, communications, and back-end services gives teams a shared measure they can act on daily. When viewing increases, customers tend to stay longer and become reliable, long-term subscribers.
"If you can't connect personalisation to revenue, you will be unable to defend the investment. You need to determine what happened because of personalisation that wouldn't have happened otherwise," Tim Armstrong, Director at Mangrove Digital, said.
How brands are using AI to decide what to recommend and when to engage
Marketers across the industry are excited about AI, primarily as a way to automate repetitive tasks and spark creative ideas.
The practical reality of AI in personalisation is more specific than the broad promise suggests. AI is being used as decision support rather than decision authority. It helps with timing, prediction, recommendations, and automation. Humans control voice, values, ethics, risk, and boundaries. The goal is to scale judgement rather than replace it.
Access to AI-driven personalisation has become significantly easier because intelligent features are now built directly into engagement platforms. Marketers can use AI to recommend content, optimise send times, prioritise audiences, and trigger engagement based on real behaviour without needing heavy data science resources.
For fast-paced, event-led businesses, the challenge is choosing the right use cases. Moments move quickly, and AI works best when applied to clear, high-impact scenarios such as surfacing the most relevant content, engaging customers at key moments, or preventing drop-off between events.
AI can assist in solving personalisation at scale, determining what to recommend, when to send, and who is at risk. Recommendation engines process millions of interactions and support capabilities like churn prediction that personalise before customers leave.
The risk is optimising toward the wrong outcome. Activities that do not lead to high-value outcomes can become the focus of optimisation and result in poor return on investment. AI amplifies strategy but does not create it.
Human oversight remains essential at the points where AI cannot substitute for judgement: setting brand boundaries, defining which outcomes the model should optimise for, and intervening when recommendations conflict with editorial or cultural standards. Even mature AI systems need humans to set the guardrails.
AI that learns over time
Self-learning reinforcement AI models offer a practical example of what sustained AI investment can deliver. Running thousands of micro-tests on customers every day, the system identifies what leads to good outcomes and does more of it, continuously refining its decisions over time.
The learning curve is measurable, and the results of sustained AI investment compound over time. AI is most useful when it continuously refines choices inside a well-defined use case. It does not invent the strategy.
"When we launched our reactivation use case, only 50% of people that we were sending a special reactivation offer to were incremental. So 50% of them were going to come back regardless," O'Byrne said.
"After two months that had gotten to 80%, and after six months it got to and has consistently stayed over 95% incrementality. We could see firsthand how the AI decision agents were learning to get better at which customers to target, and what was the best offer for every individual."
Controls remain essential across all of these applications. Editorial oversight, rights validation, cultural sensitivity, and explainable recommendation logic need to be built in at multiple levels, particularly when operating across regulated environments, children's platforms, health use cases, and global markets.
The brands getting durable returns from personalisation are the ones treating tooling, measurement, and AI as parts of the same discipline. Unified platforms make coordinated action possible. Behavioural deltas tied to revenue make the investment defensible.
AI deployed against well-defined use cases compounds value rather than activity. Without all three, personalisation remains a series of campaigns that may move engagement metrics without moving the business.
For the full picture, including the market pressures driving these shifts, the regional dynamics shaping personalisation across Asia, ANZ, and the GCC, and the first-party data strategies that underpin all of this, read the complete Braze Media & Entertainment Personalisation Report.