2026 Mobile Gaming predictions, Gossip Harbor playable strategy & UA Skills for the ai era
December 2025
Good UA, good performance, is all about battle scars, failure, wins, and glory.
It's about watching 97 creatives die so three can live.
It's about testing fifty hooks before one lands, iterating until the data tells you to stop, finding the creative that scales when everyone said it won’t work.
December is prediction season. Everyone publishes forecasts. Most age like milk. But it’s always great to hear what everyone thinks.
We read through many (many!) dozens of industry leaders’ take across four publications.
The same themes surfaced: AI stops being a talking point. Personalization becomes real. App stores lose their grip.
And everywhere, the same panic. The teams that built foundations during the hype cycle are about to cash in. The ones still waiting will discover they waited too long.
Three predictions that stood out:
"More than 50% of creatives will be AI-assisted or fully AI-generated by end of 2026." (Jakub Remiar) - Half of everything. The creative team that spent six weeks on a hero video will watch an AI spit out forty variations before lunch.
"Rewarded UA finally cracks iOS economics." (John Wright) Improved ML models and smarter web-to-app funnels make it work. The iOS wall that seemed permanent develops cracks.
"AI skill separates architects from spreadsheet survivors." (Joseph Kim) - The teams building AI workflows will orchestrate entire campaigns end-to-end. Everyone else will wonder why they're suddenly uncompetitive.
And everywhere, the same panic. The teams that built foundations during the hype cycle are about to cash in. The ones still waiting will discover they waited too long.
→ Read the full 2026 predictions recap
Unpredictable Hits Podcast E04: Jon Bellamy, Jagex CEO
The great Jon Bellamy runs a studio built around a game people have played for 25 years.
RuneScape has its own rules, votes, and culture. For anyone in games, specifically IP, and games with large fan bases this is a must.
The 70% rule: No feature ships unless players vote for it. The threshold isn't 50%. It's 70%. Players control their world. The tradeoff: if they see everything coming, surprise becomes impossible.
AI stance: AI will not write RuneScape content. Players would feel the difference immediately. But RuneScript, their 20-year-old proprietary language, can now be abstracted from Python. Eight people in the world used to write it. AI expands that circle.
On loyalty: "I can't name a tool from the last five years that I still use and feel loyal to. But I can name brands and games that have built loyalty over a long time. You can't force it or fake it."
Play it while you drive to work, ride the tube, eat a sandwich or as I do - while knitting sweaters.
Sett Labs: How Gossip Harbor Runs Their Playable Machine
Volume matters in UA creatives. But volume alone doesn’t promise you anything - performance is not just a game of volume, it’s also a game of intelligent exploration and exploitation.
The core of Gossip Harbor's UA success: explore and exploit.
The numbers: 407 new creative sets with playables in 30 days. 70-75% iterate on their winning concept (save the family). The rest: wildly different explorations.
How they iterate: They test hypotheses, not cosmetic swaps. Three items vs. one. Story hooks in the opening. Fire theme vs. cold theme.
How they explore: Pure merge (no story). Hidden object + merge. Restaurant simulation. Match-3 tap instead of swipe. Each targets a different audience that might convert.
The machine: Data-driven ideas in. Volume, diversity, intelligent iterations out. Systematic exploration until you hit gold.
Gossip Harbor hit gold. They're outpacing the puzzle category by margins no one predicted.
Read the full breakdown here including 9 examples of different playable ad concepts spotted in the wild.
On-Demand: UA Skills in the AI Era (with Supersonic)
Keren Levy Atias (Creative Growth Director, Supersonic) on what skills matter now.
The shift: Five years ago you controlled bids, audiences, campaign structures. AppLovin killed that world. SDK networks are black boxes. You pick a country, set an economic goal, upload creatives. The algorithm does the rest. Creatives are the last lever you actually control.
What moved: Audience targeting shifted upstream. You lost data access. The targeting that works now happens through creative strategy. Emotions. Motivations. The playable that makes someone feel something, have fun, and engage deeply gets the most share of spend.
AI exposed the truth: Great creative teams didn't suddenly appear. AI revealed who was already thinking deeply. Meanwhile, networks flood with "AI slop." Thousands of videos that don't tell stories. Frankenstein concepts with no hypothesis. Templates copied from competitors with a color swap.
The trash tycoon example: Keren's team wanted to tell a story about pollution and environmental harm. Two years ago? That pitch dies in production. Too expensive. Too risky. With AI? Half a day. The hook boosted performance 30% on a notoriously hard genre. AI is an accelerator. Not a replacement for taste.
Skills that matter:
Emotional intelligence: Algorithms optimize delivery. They don't understand what makes someone feel something. You do.
System thinking: Which audiences have affinity for your game? Which geos are underserved? Build a system that learns from performance data and informs the next round of ideation. Explore the entire space of possibilities systematically.
Game design literacy: Playables are everywhere. The concept might be right, but if the playable isn't fun, players drop. Tutorials break. Loops don't engage.
AI literacy: Know what to delegate and what to keep. AI doesn't have taste. Keep the judgment. Delegate the production.
The bottom line: The teams winning in 2026 treat AI as a way to lower the cost of testing ideas, not a magic button. They think in systems. They invest in creative strategy. They keep taste for themselves.
Industry Buzz
Two Ways AI Shows Up in Game Production
Companies treat AI as an isolated add-on. This creates three problems: artist uncertainty, vague cost savings that never materialize, ownership gaps where no one takes responsibility.
The fix: pipeline integration, team education before expecting delivery, systems thinking.
"That's when AI becomes boring and useful. In production, boring is usually a good sign."
Competitive Intelligence with AI
A workflow for continuous market monitoring. AI tracks competitor activities across app stores, sentiment, social, ad spending.
When competitors face crises, you activate. Marketing launches counter-campaigns. Product expedites features. UA scales budgets to capture displaced players.
Game Design Formulas
Ten years of product management distilled into nine formulas. Four that matter most for UA:
Retention = Fun + Clarity - Friction. Players stay when the game is enjoyable and they understand what to do. Friction kills retention.
Engagement = Progress + Rewards + Clarity. Players engage when they see themselves advancing and know what they're working toward.
Stickiness = Friction + Skill + Rewards. Counterintuitive: friction creates stickiness when paired with skill mastery. The struggle makes the reward sweeter.
Rage-Quit Prevention = Friction - Clarity - Skill. When friction exceeds the player's skill and understanding, they quit. Balance difficulty with guidance.
AI Churn Prediction
By the time manual analysis identifies churning players, they've already uninstalled.
The fix: real-time AI monitoring behavioral signals. Declining sessions. Stalled progression. Skipped daily rewards. Each player gets a churn probability score.
Then segment by reason: difficulty frustration, content exhaustion, social isolation, monetization pressure. Different problems need different interventions.
High-value players get personal outreach. Mid-tier players get difficulty adjustments. Social players get guild introductions. The system learns what works for each segment.
The Greatest Fake Onboarding
Kefir's Sunday City looks like GTA. Open world. Third-person exploration. Missions upgrading cafes.
Then you walk into the casino. That's the real game. Social slots hidden behind an elaborate wrapper.
The numbers: $70k/day within 30 days of launch. Three rewarded ads per day generating $250-500. Almost all revenue from IAP.
Braberg projects Sunday City could hit $7-17M monthly by next year. The playbook: dramatic CPI reduction through disguise.
Wrap Up
That’s it for this week - I wish you happy holidays wherever you are in the world, and next time we speak it’ll be 2026, let’s kill it.
Until next time, Jonathan
About the Author
Fishi is the Head of Marketing at Sett. His brain is a chaotic jukebox of ideas with more cultural references than any feed can handle. He collects sneakers and plays chess while you’re still counting sheep.
