In 2020, I gave an interview for a joint project by vc.ru and Rocket10 about mobile marketing. I was Head of Growth at AliExpress Russia, managing user acquisition and retention for one of the country's largest e-commerce apps every day. Today, in 2026, I'm CMO at Lesta Games — a mobile-first gaming studio. Different industries, same mobile. And mobile itself has changed more in these five years than in the decade before them.
I want to put down what exactly changed — without nostalgia, but grounded in direct experience.
What I Believed About Mobile in 2020 — and What No Longer Holds
In that 2020 interview, I talked about LTV as the north-star metric and about how we at AliExpress looked beyond the first conversion toward the long-term value of a user. That part hasn't aged poorly — it became mainstream. But I was wrong about something else.
Back then, I believed in the relative predictability of mobile performance marketing. The rules of the game seemed clear: there were MMP platforms (AppsFlyer, Adjust), there were channels, there was tracking — and if you set up the funnel well, everything would be measurable. The main challenge, I thought, was scale and bid optimization, not the nature of the data itself.
It turned out I was underestimating how fragile the entire infrastructure was. The predictability I believed in was an illusion — built on a silent consensus among market participants that the rules of measurement would stay fixed. iOS 14.5 dismantled that consensus within a few quarters.
In that same period, I was talking about social commerce and live streams as the next big wave. That came true — but not the way I expected. I was anticipating TikTok shopping taking hold in Russia; instead I got a heavily regulated market with traffic redistributed toward domestic platforms. The directional forecast was right; the forecast about the environment was not.
Attribution: From the MMP World to Probabilistic and Privacy-First
The biggest shift in mobile marketing didn't happen in 2020 or in 2026 — it happened in April 2021, when Apple released iOS 14.5 with mandatory ATT consent prompts. That's when the model underpinning all of mobile performance collapsed: deterministic attribution based on IDFA.
Before that, we lived in a world where you could identify with high precision which specific channel and creative had brought in the user who made a purchase on day thirty. After — that world was over.
CAC and LTV in mobile stopped being precise numbers and became estimated ranges. The teams that understood this early won: they moved to probabilistic attribution, aggregated models, SKAdNetwork, and first-party data. Teams that kept optimizing against old MMP reports were burning budget without understanding why.
In gaming — where I work now — this pain is especially acute. Game apps historically lived on precise cohort analytics. When attribution degraded, teams had to rebuild their entire decision-making approach: leaning on media mix modeling (MMM), building proprietary LTV prediction models, shifting from session-based to behavioral signals.
Privacy-first isn't a trend that will pass. It's a new market state. Android is moving in the same direction. A marketer who can't work with incomplete data is simply not competitive in 2026. And to be honest, most teams I've seen needed at least a year to genuinely shift their mindset — not just adopt new tools, but stop expecting precision where it no longer exists.
My observation: the best mobile growth teams today look less like performance specialists and more like data analysts with a marketing context. They build models, validate hypotheses through incrementality testing, and don't pretend that aggregated statistics equal individual tracking.
The Creative Economy: AI Changes the Rules (UGC, CGI, Generation)
In 2020, producing ad creatives for mobile cost real money and real time. You needed a designer, a copywriter, a producer. The cycle from concept to publish took days. This was a genuine bottleneck: good performance marketing requires constant creative refresh, and production couldn't keep up.
Today that constraint is essentially gone. Generative tools removed the production barrier entirely. A video that used to take a small team a week to create now gets produced in hours — in the right format, aspect ratio, and localization.
You'd expect this to mean the victory of quantity over quality. In practice, the opposite happened.
When everyone learned to produce "decent" content quickly, the value threshold moved higher. Users now instantly recognize synthetic content — and trust it less. UGC from real players or real customers converts better than any CGI perfection. Not because CGI is bad — but because it looks too correct.
This is especially visible in gaming. A video from an actual player recorded on their phone, with a genuine reaction to gameplay, can outperform a professionally shot cinematic trailer in CTR. We see this in tests constantly.
This changes the role of the marketer in the creative economy: the primary job is no longer managing production, but managing selection and distribution. Finding signals in what works. Testing more hypotheses in less time. Building a system that continuously refreshes creatives rather than locking in a "winner" for three months.
AI in marketing is not about replacing people. It's about shifting attention — away from execution, toward judgment. The teams that understood this early restructured their creative workflows accordingly. Those who treated generative AI as a production shortcut, without rethinking selection and testing discipline, just ended up with a higher volume of mediocre assets.
From Performance to Product Growth — Why the Boundary Disappeared
In 2020, I had a clear mental model: marketing handles acquisition and communication; product handles functionality and retention. They interact, but they're separate domains with separate metrics and separate teams.
Today I understand that division was a convenient abstraction, not reality.
Retention in mobile is not a marketing problem. It's a product problem with a marketing measurement layer. Push notifications, onboarding, the first gameplay experience, engagement loops — all of this determines whether a user comes back on day two, and no UA budget compensates for getting it wrong.
The most effective mobile growth teams I've seen across these years — in e-commerce, gaming, fintech — don't draw an internal line between "marketing" and "product" in their day-to-day work. There are people responsible for cohort P&L: they manage CAC, LTV, and product mechanics simultaneously. The marketer knows product metrics. The PM can read an attribution report. The blurring of roles isn't organizational chaos — it's a sign of a mature mobile growth team.
This shift has also changed how I think about the CMO role in a mobile-first company. The old model: CMO is about communication, brand, channels. The current model: CMO is about understanding the unit economics of each channel, managing creative velocity, reading cohort-level data. It's closer to Chief Growth Officer than to the classical CMO.
What's Coming in Five Years
I'm deliberately not going to make precise predictions — five years ago I got the direction right and missed most of the details. But some vectors feel durable.
Attribution will continue to degrade — not in the sense of "getting worse," but in the sense of "becoming less granular." Individual tracking as a decision-making foundation will disappear entirely. Marketers will work with aggregated models and causal experiments — and this will be treated as the norm, not a workaround.
Voice and native interfaces will change distribution. If users receive recommendations from an AI assistant rather than from search results or app stores, the entire logic of acquisition changes. Optimization for recommendation algorithms will become as foundational a skill as ASO is today.
UGC will become infrastructure, not a tactic. Brands that are building communities and user-content generation systems right now will, in five years, have a competitive advantage that money cannot buy outright. This isn't about virality — it's an asset.
UGC as a strategic asset requires real infrastructure: creator onboarding systems, reward mechanics, editorial curation. This is not about launching a hashtag — it's a sustained operational capability.
The line between gaming and commerce will keep blurring. Gamification in e-commerce and monetization in gaming are converging at the same point: users don't just want to buy or play — they want to interact. This insight will keep redefining product and marketing strategies for years to come.
In 2031 I'll probably re-read this and smile at the parts where I was wrong. That's fine. Reflection isn't weakness. It's the only way to avoid repeating the same mistakes across different markets.