Why AI cannot yet solve advertising’s measurement puzzle

While much of advertising these days is being transformed and increasingly automated by AI, including the creation and targeting of ads, one crucial aspect remains stubbornly human: measurement.

 

For the most part, there is still no single AI tool that can navigate the fragmented and walled data ecosystems encompassing earned media, organic views, share of voice, social metrics, and analytics providers. Agencies often juggle multiple platforms: Meltwater for LinkedIn and Instagram, separate tools for organic views, others for Chinese social media, and more.

 

"There is no real solution yet apart from continuing to export raw data into central analytical tools of various kinds," says Martin Bertilsson, founder and CEO of Multipole.AI, a consultancy and ad tech startup focused specifically on LLM-related advertising and marketing. "While these AI-powered tools make them easier to work with and allow for more productivity, the core problem remains."

 

 

The core problem of fragmented measurement is one of the toughest challenges, especially with walled ecosystems and region-specific platforms. Since no single AI tool can move seamlessly between them, most media agencies rely on a careful balance of custom integration, automation, and human oversight.

 

Aditya Shah, an insights and data lead at dentsu Singapore, explains that they use a good mix of tools to automate workflows and pull data from multiple platforms into a central cloud. Once the data sits in the cloud environment, AI models are then deployed for reporting, anomaly detection, and insight generation.

 

"We are homogenising storage, measurement, reporting, and analytics into a singular platform that will help reduce fragmentation," Shah adds. "While AI alone is not a one-stop fix for fragmentation, our cloud-first integration, AI acceleration, and human judgement strategy enable us to deliver both scale and speed while staying ahead of client needs."

 

For now, pulling data into one central reporting platform seems to be the preferred work around.

 

However, Geoff Clarke, COO at IPG Mediabrands says they have solved the fragmentation issue by building a custom, automated data pipeline that ingests information from all platforms including Meltwater, organic tools, and Chinese social networks via API (application programming interface) where possible. For any platform without an API, a robust semi-automated process is employed to collect and integrate the data.

 

"This unified approach consolidates all your disparate metrics into a single Tableau or PowerBI dashboard," Clarke says. "The result is a single source of truth that breaks down these walled ecosystems, providing a holistic, transparent, and actionable view of your cross-channel performance, all updated on a fixed, reliable cadence."

 

 

Despite advances in AI, measurement remains complex. Entry-level staff still spend considerable time on manual, repetitive tasks such as pulling platform data, reconciling discrepancies, and formatting client-specific reports. Full automation is not straightforward because every client has different KPIs, sales cycles, and data ecosystems. Additionally, custom storytelling and business context require human interpretation.

 

So, contrary to some reports about AI wiping out graduate hiring and entry-level roles in advertising, measurement tasks remain very much human. Junior roles may not be disappearing but evolving.

 

"AI is definitely changing the nature of junior roles in advertising but is not eliminating them," says Shah. "As AI advances, junior roles will shift from 'data wranglers' to data interpreters and enablers. Instead of spending hours piecing together reports, they will focus more on quality control, insight generation, and business alignment. They will learn how to train, prompt, and govern AI tools to produce smarter, faster outputs, building skills in storytelling, visualisation, and domain expertise—skills AI alone cannot replicate."

 

So, far from replacing junior roles, it seems AI is actually shifting them from repetitive execution to higher-value analytical, strategic, and client-facing responsibilities.

 

Bob Du, Jellyfish managing director, Singapore, says that while AI can automate routine reporting when data is clean and systems connect smoothly, those conditions do not always exist today.

 

"Junior staff are increasingly valuable for spotting data gaps, strengthening publisher relationships, and understanding how APIs really work in practice. As AI agents take on standardised reporting, these early-career professionals can potentially shift towards quality assurance and contextual interpretation, skills that make them indispensable to both clients and technology."

 

 

Beyond fragmented measurement and the grunt work of data integration, client demands add another layer of complexity. Reporting formats and requirements vary widely by industry, with each client having their own taxonomy, dashboard structure, and reporting cadence.

 

"For example, FMCG brands with high purchase frequency focus on reach and uplift, while financial or luxury clients with longer cycles and high-ticket SKUs prioritise lead quality or lifetime value," Shah explains. "Many clients request niche KPIs sourced from exclusive platforms. Building APIs, aligning taxonomies, and standardising such data can be time-consuming and hectic."

 

However, AI advancements promise to reduce much of this friction.

 

"AI can automate data harmonisation, dynamically generate customised reports in the exact formats clients need, and shift measurement from reporting to predictive and prescriptive insights," Shah says. "This allows agencies to spend less time stitching data and more time delivering business impact."

 

 

Even as AI matures and becomes more agentic and self-learning, able to anticipate client preferences, flag anomalies, and recommend new measurement frameworks, human oversight remains essential.

 

"Our philosophy has always been that we are the minds behind the machines," says Dane Buchanan, chief data and analytics officer at M+C Saatchi Performance. "AI accelerates the process, but human oversight keeps it accountable and decision-ready. Clients do not just need accurate numbers; they need confidence that those numbers can drive real decisions. That is where human judgement makes the difference."

 

Veron Dai, VP of data and research consultancy APAC at Assembly, views AI and human expertise as complementary. "We let AI handle heavy lifting, from automating large-scale data harmonisation to report customisation and real-time analysis, while our team focus on problem recognition, strategy, and communication, tasks that require human judgement and contextual interpretation."

 

To successfully balance AI automation with crucial human oversight, companies must employ a hybrid approach that integrates AI-driven processes with human expertise at critical workflow points.

 

"Human-based feedback loops, alongside rigorous quality assurance protocols, will increasingly be called for in master service agreements," says Clarke. "Transparency is crucial for sustainable industry practices, ensuring open client communication that manages expectations about AI’s role, guarantees output quality, differentiation, and overall satisfaction."

 

 

There is no doubt that AI offers powerful opportunities to enhance creativity, efficiency, and effectiveness in advertising and marketing. Yet, its use must be guided by responsibility, transparency, and respect for ethical and societal values.

 

"It is also important to note that in five years’ time, AI advantage will be largely ubiquitous," adds Clarke, "meaning those who invest behind craft now will possess tomorrow’s competitive advantage. All companies will need to refine their ways of working and policies to ensure human intelligence and craft remain central to output value."