As organizations continue to invest heavily in data, there is a growing assumption that better data naturally leads to better decisions. Dashboards are more advanced, metrics are tracked in real time, and performance is more visible than ever before.
But while data has expanded at an unprecedented rate, insights have not kept up.
This gap between visibility and understanding creates a dangerous side effect: false confidence in the numbers.
When visibility creates false confidence
An ongoing U.S. class action lawsuit, DZ Reserve et al. v. Meta Platforms, illustrates this issue. The plaintiffs, advertisers on Meta’s Platforms, allege that the “potential reach” metric estimated for Facebook and Instagram advertisements was overstated by as much as 200-400%. This inflation was largely due to duplicate and fake accounts. In one example, Meta’s reported potential reach for 18-34 year olds in Chicago was nearly 4 times higher than the actual number of users in that demographic.
If true, the commercial implications are significant since the metric is used by advertisers for audience sizing, pricing and budget allocation. According to the claims, advertisers have spent more and paid higher prices than they otherwise would have, effectively overinvesting based on a distorted view of the market.
This points to a broader issue that extends beyond a single platform. The risk is not just misinterpretation, but mismeasurement. More fundamentally, it raises a question of trust. If something so fundamental is inaccurate, what else might be?
Basic descriptive analytics is about understanding what is happening through data. But in practice, this only works when the underlying data reflects reality. When validation is overlooked, even well-reasoned analysis can lead to flawed and potentially costly decisions.
Why this matters in Thailand
The implications of this become even clearer in markets like Thailand, where digital adoption is high and customer interactions happen inside a few large platforms. Internet penetration is 91.2%, and 71.1% of the population is active on social media. Nearly 68.2% of Thai consumers indicate a preference for e-commerce over physical stores, making digital channels core to customer acquisition and retention in Thailand.
Thai businesses’ reliance on digital is also reflected in their spending. In 2025, 33 billion THB was directed to digital advertising, much of it concentrated within a small number of dominant platforms, including Meta, TikTok, and LINE.
Concentration creates scale, but it also creates dependency. When platforms dominate demand generation, control over distribution and performance measurement becomes increasingly centralized. As businesses operate across multiple platforms, analytics becomes fragmented, with data split across systems and metrics defined inconsistently.
This is compounded by the nature of platform-generated data. The systems are opaque, offering limited visibility into how performance results are calculated. Platforms also increasingly own the customer relationship, leaving businesses dependent on data they cannot fully validate or control.
In markets like Thailand, a coherent view of performance is difficult to establish, increasing the risk of misallocating marketing spend and misjudging channel effectiveness.
Improving validation with AI
Addressing this requires a shift in how organizations approach data. Not as a fixed source of truth, but as something to be continuously validated and contextualized across systems.
This is where AI becomes operationally critical. By integrating data across platforms through pipelines and unified data layers, machine learning can flag anomalies, detect inconsistencies and reconcile across sources. For example, anomaly detection models can identify unusual spikes in platform reported metrics. More advanced approaches use cross-platform attribution to estimate each channel’s true contribution rather than relying solely on platform metrics.
A major source of distortion is duplicate users across platforms, which inflates reach and skews performance metrics. AI-based probabilistic matching addresses this by identifying when activity across channels likely comes from the same user, reducing double counting.
Together, these methods allow organizations to benchmark platform reported metrics against first-party data, enabling indirect validation even when platform metrics remain opaque.
Beyond validation, AI helps construct a coherent view of performance by identifying patterns across fragmented datasets that are not visible within an individual platform in isolation. This shifts organizations from passively consuming data to actively validating and interpreting it, turning data into a foundation for confident, forward-looking decisions.