2025: When scale stopped signalling value in adtech

As AI-driven automation floods digital media with synthetic activity, 2025 marks the moment attribution, verification and signal integrity replace raw scale as the true engines of growth.

For most of the last decade, the ad tech industry has operated on a simple assumption. More impressions, more clicks and more installs eventually translated into more value. In 2025, that assumption finally broke.

Data from the Singular Quarterly Report 2025 Q4 shows a market where activity continues to rise, but confidence in what that activity represents is eroding fast. Global ad spends jumped 24.4 percent quarter-on-quarter as marketers ramped up for the holiday season, yet this surge masked a deeper imbalance. On iOS, impressions actually fell 3.4 percent while spend rose 24.3 percent. On Android, impressions climbed 30.9 percent alongside a similar increase in spend. The result was predictable. More money chased fewer impressions on Apple’s ecosystem, driving a sharp escalation in acquisition costs and reinforcing the idea that scale alone no longer explains performance.

This divergence is central to the industry’s current reckoning. As AI, automation and programmatic systems make it easier to generate impressions, clicks and even conversions, volume has become abundant and cheap. Trust has not.

The end of volume as a growth proxy

The most striking signal in the report is cost inflation on iOS. Average CPI on Apple devices surged 44.4 percent quarter-on-quarter to $9.11, the steepest rise in more than a year. Android CPI, by contrast, rose a modest 7.9 percent to $0.68. The reason goes beyond seasonality. Advertisers are increasingly confident that iOS users deliver higher lifetime value, even though they represent a shrinking share of total installs.

Retail offers a clear example of this shift. Retail app installs are now the most expensive across both platforms, with CPI rising roughly 40 percent on Android and 61 percent on iOS in a single quarter. Despite this, advertisers continue to bid aggressively for premium users ahead of the festive season, indicating that return, not reach, is driving decisions.

Nikhil Kumar, Chief Growth Officer at mediasmart powered by affle, believes this marks a structural change. “Real growth in 2026 will no longer be defined by volume metrics alone. With synthetic impressions, clicks and installs flooding the ecosystem, growth will be about quality, signal integrity and actionable outcomes. Attribution and verification will be the new currency of growth,” he says. According to Kumar, the industry is already seeing stronger impact when interactions translate into meaningful business outcomes rather than inflated dashboards.

Attribution moves to the core

As optimisation cycles compress, attribution is no longer a reporting afterthought. It is becoming core infrastructure. The Singular data highlights why. Across most verticals, Android continues to deliver 70 to 95 percent of installs, but iOS drives 70 to 95 percent of revenue. The skew is extreme in categories like Retail, where Android accounts for 71 percent of installs while iOS captures 99 percent of revenue. Health and Fitness shows a similar imbalance, with installs split evenly but iOS generating 94 percent of revenue.

This gap makes accurate attribution existential rather than optional. When optimisation engines rely on flawed or inflated signals, they do not simply misallocate budgets. They compound errors at machine speed.

Siddharth Jhawar, Country Manager at Moloco India, notes that this problem predates digital advertising but has intensified with automation. “The lack of clarity on the true impact of advertising is not a new problem. Unfortunately, progress in digital advertising has also created more sophisticated methods of beating measurement and attribution systems,” he says. Jhawar argues that the next phase of ad tech will be defined by incrementality. By separating users into test, ghost and placebo groups, advertisers can isolate whether advertising actually drove incremental outcomes. “The next phase is not about blind trust. It is about scientifically driving and measuring incrementality,” he adds.

Also read: Are lines between agencies and tech platforms blurring in India’s digital marketing?

Big Tech Q3: Search holds ground, as retail media networks challenge Google–Meta duopoly

Fraud becomes structural, not seasonal

Another uncomfortable truth running through the data is that fraud is no longer an edge case. The report treats it as persistent and adaptive, embedded within automated media flows. Rising CPIs, especially on iOS, make the cost of polluted signals far higher than in the past. Bad data no longer just wastes spend. It distorts optimisation models and strategic decisions.

Utilities highlight the paradox. On Android, Utilities installs surged 110 percent quarter-on-quarter, making it the fastest-growing category. CPI remains among the lowest across regions, and both CTR and conversion rates are strong. Yet this efficiency also makes Utilities a magnet for low-quality traffic if verification is weak. The same automation that enables scale can just as easily accelerate leakage.

Vishal Singh, Vice President Brand and Agency Partnership at Globale Media, frames this as a shift from growth curves to credibility curves. “Growth curves can be engineered. Credibility curves must be earned over time. Signal integrity, fraud resistance and consistency of outcomes will emerge as long-term success metrics,” he says. In an AI-driven ecosystem, Singh believes credibility becomes a defensible moat that cannot be bought or automated overnight.

Android scales, iOS pays

The platform split remains stark. Android dominates volume, particularly in Tier 2 East which includes India and Tier 2 West markets, where CPIs are low and install volumes massive. These regions are ideal for top-funnel growth and experimentation, though monetisation per user often lags. iOS, by contrast, continues to concentrate value. In Financial apps, Android drives 82 percent of installs but contributes only 22 percent of revenue. In Retail, the revenue skew is even more dramatic.

This reality explains why advertisers are willing to tolerate soaring iOS CPIs. They are not buying installs. They are buying belief in downstream value.

Kumar describes this shift as ad tech selling confidence rather than media. “Ad tech’s value is increasingly about enabling confident decisions at scale. With AI optimising media in real time, the key differentiator is verifiable performance and trustworthy attribution,” he says. Platforms that ensure every impression, click or conversion is credible are empowering advertisers to act with certainty rather than assumption.

Automation accelerates the pressure

The report also shows how quickly AI-led automation is being adopted. TikTok Smart Plus campaigns grew from just 9 percent of performance ad spend in Q1 2025 to 42 percent by Q3. Instagram Reels now account for 26 percent of ad impressions, pushing down CPMs while expanding inventory. YouTube Shorts, once considered mobile-native, now sees nearly half its ad spend coming from TV screens.

These shifts are expanding reach but also intensifying noise. When optimisation is automated and creativity is generated at scale, the cost of trusting the wrong signal rises sharply.

Singh puts it bluntly. “In a world where machines generate, buy and optimise media, ad tech is increasingly selling belief in outcomes, not just advertising. The platforms that succeed will be those that safeguard reality, not just performance.”

The business implication for 2026

2025 is the year where scale stopped being sufficient. Rising CPIs, shrinking premium inventory and AI-driven signal inflation have forced the industry to confront uncomfortable questions about truth and trust.

For brands, this means demanding proof rather than promises. For agencies, it means defending restraint and verification even when automation rewards speed. For platforms, it means choosing whether to amplify noise or enforce credibility.

As Jhawar advises, when a channel offers the sun and the moon, the response should be simple. Ask for log-level data. Ask for incrementality. The answers will reveal whether growth deserves belief.

In 2026, the winners in ad tech will not be those who generate the most activity. They will be the ones who can prove which activity is real.