What iGaming’s Personalisation Experiment Can Teach Fintech

By Fraser Dunk, CEO and Co-Founder, Jurnii 

Gaming and fintech are converging faster than either acknowledges — and gaming has  spent years, and serious money, learning personalisation lessons fintech is now starting to  repeat. 

Few words are deployed as confidently, or defined as loosely, as “personalisation.” Across consumer  fintech it now sits near the top of most roadmaps and most budgets, sold as the route to  engagement, loyalty and lifetime value. iGaming has been running the same experiment for longer  and at higher intensity, and the results are worth studying — because the two industries share more  DNA than they admit. Both are high-frequency and habit-driven, both run on fragmented data, both  are heavily regulated, and in both, customer trust is fragile and easily spent. Where gaming has  already burned budget learning what personalisation does and does not deliver, fintech has a chance  to skip ahead.  

The first lesson is that the word itself is doing too much work. In gaming, “personalisation” describes  at least five different things: content, UX, journey, offer and product configuration — each a distinct  problem with distinct data, capabilities and success metrics. A banking or payments app faces the  same sprawl: a spending-insights feed, a nudge to move money, a pre-approved credit offer, a  reordered home screen and a configurable dashboard are all filed under “personalisation,” yet have  almost nothing in common operationally. Fund them as one initiative and you spread resource across  all of them without solving any properly — or attributing return to any one of them.  

The attribution illusion

The measurement problem runs deeper than definitions. In gaming, discovery overwhelmingly  happens off-platform — through affiliates, search, social and communities — so players tend to  arrive with intent already formed. An on-site recommendation engine can nonetheless claim credit  for any engagement that follows an impression, regardless of whether that intent existed  beforehand. The result is a systematically inflated return: credit booked to a layer that did not earn it.  

Fintech should recognise this immediately. A customer opens their banking app to pay a bill or check  a balance — intent formed off-app, often before the phone left their pocket. An in-app  recommendation sitting in that flow can claim the conversion that was always going to happen. In  both industries, the personalisation layer is frequently just the last touch before an action with a  cause somewhere else entirely.  

The behavioural reality compounds it. Gameplay is far more habitual than exploratory; a player  loading the same games every Tuesday evening is completing a task, not browsing, and once a  behaviour becomes habitual, deliberation drops out of it. Payments and everyday banking are among  the most habitual digital behaviours that exist — a transfer in progress is fintech’s equivalent of an  in-play bet: high intent, low tolerance for friction. Personalisation that reorders the route, or asks the  user to navigate past a “recommended for you” panel to reach a known destination, makes the  product worse in the name of relevance — imposing a relearning cost on people who simply want to  pay and leave.  

A foundation thinner than the ambition

Then there is the data. In gaming, players routinely hold three to five accounts across competing  brands, so any single operator sees only a fragment of a customer who is signalling preferences  elsewhere — and even within multi-brand groups, those identities are rarely unified. Fintech has its  own version: open banking widens the picture, but customers still spread their financial lives across  multiple apps, cards and providers, and no single firm sees the whole. Building sophisticated  inference on a partial view reliably turns spend into noise, in either industry.  

There is also the question of what these systems optimise for. In gaming, a recommendation engine  might score one game at 0.81 and another at 0.79 — a negligible relevance difference — while the  first returns 12% of gross gaming revenue and the second 8%. Most engines maximise their own  internal metric, not the commercial outcome. The fintech parallel carries an extra edge: a model  optimising an engagement score is not optimising for the customer’s suitable outcome, and the two  can diverge sharply. Recommending the “most relevant” product is not the same as recommending  the right one — and in financial services, that gap has conduct consequences attached.  

That points to the most serious shared risk. Machine-learning systems trained on historical data  replicate and amplify whatever is in that data. If a gaming operator’s history reflects aggressive CRM  or products that disproportionately engaged higher-risk players, an automated system will optimise  to recreate those conditions — more efficiently than any human team could. The UK Gambling  Commission has flagged exactly this in its guidance on AI in marketing. Fintech’s regulators arrive at  the same place from another direction: the FCA’s Consumer Duty and its work on AI make harm amplifying automation a direct fair-treatment exposure, not an abstract ethics question. Both circle  the same concern — that automated systems can identify and act on customer vulnerability more  effectively than people can, which is a risk dressed up as a capability.  

Where the return is real

None of this means personalisation has no value. It means its value is concentrated. The return is  genuine where intent is unformed or has lapsed — at onboarding, before habits are set, and at  reactivation, where a relevant prompt gives a dormant customer a specific reason to return. In  gaming, these are the moments a good VIP account manager has always handled instinctively; that  relationship-led model remains the most effective form of personalisation the industry has produced.  Fintech’s nearest equivalent is the private banker. The lesson in both cases is to direct spend where  the marginal return is highest, not where the technology is most impressive.  

That argues for sequencing. Before funding advanced personalisation, three things earn their return  first. Fix the core journeys — a failed payment, a stalled transfer, unnecessary KYC friction — because  no recommendation engine recovers a customer lost at a broken payment screen, and those fixes  consistently outperform the engine on return. Get honest about data hygiene before buying systems  that depend on data you do not fully hold. And define precisely what you are personalising, so each  initiative is measured against an outcome rather than a model’s own score.  

The convergence makes this urgent

This is also where the two industries genuinely merge. As large language models move into the  interface, the UI increasingly becomes the prompt, and both gaming and payments are heading  toward conversational, intent-mapped, embedded experiences. The line between the product and  the personalised version of the product starts to collapse; the experience responds in real time to  what a customer is trying to do, not to who they have historically been. That rewards firms that have  built — or tightly control — their own intelligence layer, and exposes those leaning on generic tools,  because generic AI delivers generic advantage.  

The question I would put to any product or risk leader in fintech is the same one I put to gaming  operators. If you switched off your personalisation layer tomorrow, could you prove — in customer  and commercial terms — what you had lost? If the honest answer is no, you are not measuring an  asset. You are funding an assumption. Gaming has already paid to learn that. Fintech does not have  to.