Key outcomes:

• 40 hours per week of manual CSV exports and App Script maintenance eliminated—equivalent to a full-time analyst.
• Sub-2-hour data latency from 11 media platforms into BigQuery enables near-real-time pacing dashboards.
• Single governance layer for 1.9 K ad accounts; schema changes are auto-detected and version-controlled.
• 100% adoption of a standard naming taxonomy across all media teams within eight weeks.

About the client

Omnicom Media Group Hong Kong (OMG HK) is the Asia-Pacific division of Omnicom Group, a global media and communications company. With 1,500+ employees, OMG HK provides media planning, buying, and analytics services across 1,900 ad accounts on Meta, Google, Microsoft, Taboola, and other niche platforms.

Challenges

As Omnicom Media Group Hong Kong’s client base expanded rapidly, so did the complexity of managing their data. With ≈1,900 ad accounts spanning 11 platforms, the volume and variety of data became overwhelming.

The Omnicom’s analytics team faced several key challenges:

• Manual and time-consuming data handling: Analysts had to download multiple reports manually, wrangle data across numerous spreadsheets, and write fragile Python scripts to upload and transform data into BigQuery.
• Data inconsistency and reconciliation issues: Weekly “data-day” meetings required the entire team’s attention just to reconcile mismatched date ranges, currency conversions, and varying data schemas across platforms.
• Scalability bottlenecks: The growing number of accounts and platforms made the existing manual processes unsustainable, causing delays in data availability and limiting the team’s ability to deliver timely insights.
• Fragile infrastructure: The custom scripts and workflows were prone to breaking whenever platforms updated their APIs or data structures, leading to frequent dashboard outages and firefighting.

These challenges not only drained valuable analyst hours but also hindered the team’s ability to focus on higher-value tasks such as strategic analysis and campaign optimization.

Omnicom’s primary target was to automate omnichannel reporting to eliminate manual work and reduce errors.

Solutions

To tackle the complexity and scale of omnichannel data reporting, OMG HK looked for an automated ETL/ELT solution that could seamlessly collect data from multiple platforms and deliver large datasets directly into BigQuery for advanced analysis. Additionally, the tool needed to support Google Sheets and BI tools for automated reporting and data visualization.

After comparing Supermetrics, Fivetran, and in-house builds, the team selected Windsor.ai for three reasons:

Modernised architecture

Here’s what OMG HK’s updated data architecture with Windsor.ai looks like, automating and simplifying their data flows:

• Data sources: OMG HK collects data from 11 different media and analytics platforms.
• Windsor.ai connectors: Our connectors automatically pull data from all these platforms without manual work.
• Raw tables → BigQuery (hourly refresh): The raw data from all sources is integrated into Google BigQuery every hour, ensuring up-to-date, centralized storage of all media data.
• Curated views → Google Sheets: Cleaned, structured summaries (curated views) of this data are automatically streamed to Google Sheets when needed for quick checks or ad-hoc analysis.
• Downstream tools: From BigQuery and Sheets, data feeds into business intelligence and analytics tools like Looker Studio dashboards, Tableau workbooks, and Python notebooks for deeper analysis and custom reporting.

Results

Thanks to the fully automated Google’s native pipelines through Windsor.ai, Omnicom’s analysts become insight partners, not data janitors. Here are their key outcomes:

• 40 hours per week saved by eliminating manual CSV exports and App Script maintenance—equivalent to the workload of a full-time analyst.
• Near real-time data availability: Sub-2-hour latency from 11 media platforms into BigQuery enables campaign pacing dashboards to refresh automatically every morning before trading desks arrive.
• Unified data governance: A single governance layer manages around 1,900 ad accounts with automatic schema change detection and version control.
• Rapid standardization: Omnicom achieved 100% adoption of a standard naming taxonomy across all media teams within eight weeks.
• Increased analyst productivity: With 40 hours freed weekly, the analytics team shifted focus from manual data extraction to creative testing and cohort analysis.
• Enhanced reliability: Windsor.ai’s schema tracking ensures new GA4 fields appear automatically in BigQuery without manual intervention or tickets.
• Improved scalability: Onboarding new clients is now as simple as ticking a checkbox in Windsor.ai, eliminating the need for provisioning new ETL jobs.

Final takeaway

Omnicom Media Group Hong Kong replaced manual ETL workflows with Windsor.ai’s automated pipelines, integrating data from 11 sources into BigQuery with sub-2-hour latency. The new architecture processes up to 1,900 ad accounts and supports downstream tools like Looker Studio, Tableau, and Python notebooks.

Analysts no longer manage scripts or handle reconciliation issues caused by API changes; schema-aware syncs and automated alerts ensure consistency. Standardized naming conventions were adopted across teams within eight weeks. New client onboarding now takes minutes via a UI-driven workflow.

Curious what Windsor.ai can do for your team?

Start your free 30-day trial: https://onboard.windsor.ai/, or book a demo: https://calendly.com/windsorai/guided-onboarding.