E-commerce is more competitive now than it’s been in years. There is a structural shift as marketplaces are becoming the dominant entry point for shoppers, user behaviour is changing rapidly and acquisition costs are being pushed upward by new tax rules, supply chain pressures and rising returns. In this landscape, many advertisers have hit a wall – manual optimisation simply cannot keep pace with the volume, speed and complexity of decisions required to grow.
What we’re seeing across thousands of campaigns is that AI-driven optimisation has become a fundamental requirement for scaling efficiently in 2026. The combination of marketplace-first shopping, fragmented user journeys and fast-moving creative dynamics has elevated AI from optional to essential.
Why marketplace-first shopping matters
One of the biggest shifts we’ve observed is the steady move toward marketplace-first buying with 72% of global e-commerce revenue coming from marketplaces. Whether it’s Shopee and Lazada across Southeast Asia, SHEIN and Temu in cross-border commerce, or travel platforms like Trip.com and Traveloka – marketplaces now convert better at scale. This is because they remove friction. Users trust their payment flows, delivery expectations are clear and most importantly, the buying experience is familiar.
For advertisers, this creates a new challenge. Instead of sending traffic to a single brand site and optimising one funnel, growth now depends on managing dozens of micro-journeys inside platforms they do not fully control. Product pages, category listings, seasonal deals and app-specific features all behave differently. The shopper who clicks an In-Page Push notification may land on a flash-sale item. A popunder user may arrive directly on a product detail page. Meanwhile, Telegram Mini App traffic tends to follow an entirely different interaction pattern.
This level of variation simply can’t be managed manually at scale. The decisions pile up fast, selecting GEOs or matching traffic sources to product types, and quickly exceeds what human teams can track. This is why automation and predictive systems have become core to how top performers operate. They help teams react in real time to signals that would otherwise be lost, such as marketplace ranking shifts or time-of-day behaviour.
Creative-led performance is now an AI challenge
Five years ago, advertisers could compensate for weaker creative with targeted segmentation. That logic no longer applies. We have seen across our network that creative strategy consistently outperforms traditional targeting as the main driver of ROI.
The formats that now dominate performance are Push, In-Page Push, Popunder and increasingly Telegram Mini App ads. These are all extremely sensitive to creative variation, as industry data shows that a single change in a headline, icon or angle can significantly shift click-through and downstream conversion rates.
This is where AI becomes indispensable. Machine-learning powered systems can test creative variants continuously, identify patterns across GEOs and highlight which combinations deserve more spend. They can also detect early signs of fatigue and automatically rotate in new variants before performance collapses.
Performance data shows that engagement typically drops 20-30% week over week as high performing ads near the end of their natural run, making continuous creative rotation essential.
Creative-led performance now means maintaining a constant pipeline of variations and allowing AI to determine which ones resonate with each audience segment. This gives advertisers a major advantage, especially in e-commerce, where users expect immediacy and relevance.
Why rising costs demand predictive optimisation
Competition is not the only force pushing advertisers towards AI-led optimisation. Costs are rising for reasons outside of marketing control.
In the US, the removal of tax exemptions on low-cost imports has increased prices for items from major cross-border retailers. In Europe, customs checks are tightening, which will directly impact e-commerce profitability. Even slight increases in cost can reduce conversion rates, meaning advertisers must compensate through more efficient media allocation.
E-commerce advertisers are already managing average acquisition costs of $45.27 for search and $65.80 for display advertising. As external costs rise from tax changes and increased returns, these figures climb further, leaving little room for inefficiency.
Consumer behaviour is also creating new inefficiencies. Industry data shows around 30% of purchased items are returned, driven partly by the rise of “keep or return?” influencer content and creators who purchase items solely for filming before sending them back. This trend inflates operational costs for retailers, which eventually flow back into acquisition costs for advertisers.
Predictive models help stabilise performance. They analyse how different combinations of GEO, device, product category and format behave under changing market conditions. If a particular offer becomes less viable due to increased tax, volatility in returns, or seasonal fatigue, AI systems can automatically redirect spend to more resilient segments. This prevents wasted budget and keeps campaigns profitable when external factors change suddenly.
E-commerce in 2026 requires a different operating model
Advertisers who rely solely on manual optimisation will struggle to sustain growth. AI does not replace performance teams, it expands their capacity. Teams can focus on strategy, product selection and creative direction, while the system handles the thousands of micro-adjustments required to keep campaigns profitable.
In 2026, the brands growing fastest are the ones using automation to understand where their next buyer is coming from, how marketplaces are evolving day by day and which creative elements truly drive conversion. As the market becomes more competitive, AI is becoming the deciding factor in whether e-commerce advertisers can scale profitably.
For advertisers, the immediate opportunity lies in shifting from manual optimization to automated decision-making, particularly around product selection, creative testing and GEO-level performance. This allows teams to spend less time reacting to volatility and more time shaping sustainable growth strategies.
