Cost per acquisition has long been banded around as a key metric to determine the performance of an affiliate. However, doing so is like ‘judging a first date by how cheap the parking was’, according to Casumo’s Chief Marketing Officer, Nikola Jellacic.
Speaking to iGaming Expert ahead of taking part in a panel titled ‘Designing Campaigns the Data Can Read’ at SBC Summit in Lisbon, Jellacic revealed the metric that affiliates should be focused on instead, as well as examined the role of AI in marketing and shared his advice for calming the ongoing battle between creative and performance divisions.
Have marketers become too focused on short-term analytics rather than what actually drives long-term growth?
Guilty, your honour. And iGaming is the worst offender – someone clicks, deposits, and the dopamine hits your dashboard before the kettle’s boiled.
We end up running an entire business at the resolution of ‘what happened in the last four hours’, which is like steering a ship by staring at the foam right next to the hull.
The stuff that actually compounds – trust, being the brand people type in directly instead of stumbling onto via some affiliate’s 47th link – refuses to show up in tomorrow’s numbers.
Though short-term data isn’t the villain, believing it’s the whole story is. You need two clocks running and the discipline not to let the twitchy one make decisions only the slow one can see coming.
What metric or KPI do you think the industry pays too much attention to and what should it be focusing on instead?
We treat cost-per-acquisition like it’s the meaning of life when it’s basically the marketing equivalent of judging a first date by how cheap the parking was. It tells you what you paid to get someone through the door and absolutely nothing about whether they’ll stick around.
The number that actually matters is contribution margin per cohort over six to 12 months. Switch to that, and half your channels stop looking like growth and start looking like a very efficient way to set money on fire.
How should Chief Marketing Officers balance short-term metrics with the need to build distinctive brands that may not show immediate returns?
I stopped treating it as a philosophy seminar and started treating it as a budget with a bouncer.
A fixed chunk of spend gets ring-fenced for brand and explicitly told it does not have to hit a 30-day return on ad spend target — because brand loses that fight every single time. You measure it on the slow instruments instead: branded search creeping up, direct traffic, people actually remembering you exist.
The classic CMO own-goal is defending brand spend with performance metrics, then acting surprised when it loses. Don’t. Argue brand on its own evidence, get the board to agree the split while everyone’s still calm, and guard that ring-fence – because the quarter always gets tight, and the brand budget is the first thing someone tries to ‘borrow’.
How has AI changed the ability of marketing teams to measure the impact of their promotions?
The big leap is going from ‘what happened’ to ‘okay but what would’ve happened anyway, you absolute optimist’.
For years, our attribution was essentially fan fiction — last-click, some platform grading its own homework, a multi-touch model nobody trusted but everyone nodded along to. AI made proper causal measurement actually doable at our speed: geo-experiments, synthetic control groups, uplift models that tell apart the player who deposited because of your promo from the one who was going to deposit regardless.
In an offer-led industry, that is significant, because the most expensive spend is the spend that changed nothing. AI finds that leak.
However, it also makes it dead easy to produce a gorgeous, confident, fully-animated answer that is completely and serenely wrong – which is exactly the trap to keep both hands on the wheel for.
With AI generating more insights than ever, how do you distinguish between genuinely actionable intelligence and simply more noise?
I make every ‘insight’ pass through three bouncers before it gets to call itself that.
One: would knowing this change a single thing we do? If not, it’s trivia – fun at parties, useless in a strategy.
Two: causal or just correlated? Patterns in player data are like pigeons in a square – there are infinite of them, and most mean nothing. I want a mechanism I can actually yank a lever on.
Three: does it survive a holdout, or does it dissolve the moment we test it? The modern problem isn’t generating insight – a model will cheerfully hand you 200 ‘opportunities’ before lunch. The skill is brutal triage. I’ll take three things we’ve genuinely stress-tested over 200 we’ve merely admired. Noise wearing a signal costume is far more dangerous than honest noise, because the costume gets it past security and into your roadmap.
How can creative teams and performance teams work more collaboratively to improve performance outcomes?
Most organisations run this relay race where creative makes the thing, lobs it over a wall, performance runs it, and reports back numbers the creative side never sees or feels. That’s not collaboration, that’s two departments writing each other passive-aggressive postcards.
Put them on one squad, one P&L line, one shared outcome.
Creatives should see which hooks drive retained value, not just clicks, and see it weekly — not in a quarterly autopsy. Performance should brief the actual problem (‘we’ve saturated this audience, we need three genuinely new angles’) instead of just chanting ‘more variants’ into the void.
AI’s quietly forcing the reunion anyway: when you can generate and test 50 variants in a week, production stops being the bottleneck and taste becomes the bottleneck – which is exactly the spot where these two were always meant to be standing together, slightly awkwardly, holding the same coffee.
Is marketing becoming too reliant on AI, or is there scope to use it even more to support campaigns?
Both! We’re over-trusting it for the thinking and under-using it for the heavy lifting.
The over-reliance looks like outsourced judgement – nodding along because the model said it confidently and in a nice font – plus a creeping sameness where everyone runs identical tools on identical data and arrives at the same ‘optimal’ answer.
In a crowded market, sounding like everyone else isn’t a strategy, it’s camouflage for the predators. Meanwhile, there’s a goldmine in the boring layer: automated incrementality testing, real-time budget shuffling, squishing the test-and-learn cycle from weeks down to days. Let the robot do everything that scales. Keep the humans on everything that makes what you do you.
Beyond your panel participation, what are you most excited about for the upcoming SBC Summit in Lisbon, and why is it important for the industry to come together at these sorts of events?
Truthfully? The corridors more than the stages. The panels set the agenda, but the real gold is the unguarded ten minutes over an ice cream with someone wrestling the exact same gremlin you are – what’s actually working on retention, where they got beautifully burned, how they’re handling another market deciding to ‘have a word’.
Our industry talks innovation at full volume, but an embarrassing amount of genuine learning still travels human-to-human, slightly off the record, ideally near the pastries.
Lisbon matters because we’ve got AI, regulation, rising costs and rising player expectations all having a midlife crisis simultaneously, and no single operator can see the whole board.
Events like SBC are where the industry quietly calibrates — where you discover whether the thing keeping you up at night is your problem or everyone’s problem. And honestly, that one distinction is worth more than any keynote, and definitely worth the airfare.
Held in Lisbon from 29 September to 1 October 2026, SBC Summit is one of the world’s largest gatherings of betting and gaming professionals.
The event will bring together 40,000 attendees from across the industry for three days of learning, networking, and discussion, alongside a major exhibition featuring leading brands from around the globe.
For more information and tickets, visit sbcevents.com/sbc-summit.
