LC Waikiki is a fashion retailer offering men’s, women’s, and children’s clothing in more than 50 different markets.

Definitions: 

BigQuery is a robust cloud-based database and analytics tool that enables the quick and efficient querying, analysis, and visualization of large datasets. With BigQuery, we can perform scalable, high-performance data analyses and make decisions based on information. The rapid query processing allows instant exploration of extensive datasets, providing significant advantages in developing data-driven solutions and obtaining business intelligence. 

Meta ads are effective and targeted advertising solutions offered through Meta’s advertising platform. These ads enable reaching a broad user base on platforms such as Facebook, Instagram, WhatsApp, and other Meta services. Meta ads offer comprehensive targeting options, interactive ad formats, and detailed analytics capabilities, allowing brands to reach their audiences in a more personalized, effective, and measurable way. 

Approach: 

This project, focusing on transferring data from the Google ecosystem to Meta through BigQuery, represents a significant milestone for LCW TR. The process, completed and operationalized last month (automatically transferring audiences from BigQuery to Meta using customer match), has been shared with the brand after thorough evaluations, receiving positive feedback. 

Google’s default predictive audiences can be shared anonymously with ad accounts connected to GA4. However, to further optimize these audiences, we conducted a specific study during the peak communication period in November. We personalized this estimated user audience even more by using our existing measurements, aiming to separate users who would generate higher revenue than those targeted with coupon code communications in push notifications, with the goal of improving customer attraction quality and optimizing ROAS. 

We transformed this customized audience into a daily updated audience table on BigQuery. Using first-party data, we segmented the audiences based on specific rule sets using mobile device IDs (madid). These tables were fed into ML-based predictions, and through an API developed in-house, new users were added to the desired audience ID in Meta every day at a specific time, keeping the information up to date. 

Execution: 

Although this audience comprised only application campaigns, we closely monitored the development and changes of these four campaigns by adding a new audience in addition to our existing campaigns. Due to the fact that the campaigns were not exclusively application campaigns, we also focused on the web audience within. 

On the web side, a few days before our audience started being added to the campaigns, we transitioned to consent mode. This allowed us to continue measurements by collecting data only from users who consented to cookies. By utilizing server-side GTM, we enhanced signal maximization on the CAPI side and elevated measurement quality and signaling by feeding advanced matching (1PD) data. In the November period, we took a significant step to ensure that this newly added audience did not affect web conversions and revenue. 

Results: 

To achieve success beyond the increase in the shopping season, we evaluated the November period, taking into account the effects of 11.11, Black Friday and Cyber Monday. During this process, we also specifically focused on calculating the incremental impact of seasonality on direct/organic orders and overall order/revenue data. 

• The different ML model we developed allowed us to go beyond default methods, enabling us to create a personalized audience with more accuracy and higher returns.

• Feeding our personalized audience through 1PD via mobile device IDs, we created a more active audience compared to users with ad interactions.

• Utilizing our technological infrastructure, we built an API that automatically transfers data from the Google ecosystem to Meta.

• Thanks to the reliability of integrations, we continued to enhance measurement signal quality on the web without experiencing any potential loss due to consent. 

The results demonstrate the achievements we obtained with a reduced budget between November 10th and 30th: 

• Compared to the average upward trend in overall, our order trend increased by 2.7 times, and our revenue soared to 5.4 times higher. This indicates that the average increase achieved in the second period is 2.7 and 5.4 times higher than the first period.

• Following the addition of the audience to these campaigns, we achieved 65% more orders, 101% more revenue, 38% lower CPA, and 99% higher ROAS compared to previous periods.

• Despite the cost spending among all campaigns, these campaigns contributed to 21% of all orders and 20% of revenue. On average, these campaigns have 3 times higher ROAS and 35% lower CPA compared to the general performance.

• With our uploaded Customer Match audience reaching 1.7 million users, we achieved a high match rate with the targeted 1.5 million users. 

 

Leveraging BigQuery, lcwaikiki.com achieved data interoperability between Google and Meta, optimizing campaign parameters for heightened user engagement and conversion efficacy. The integration of personalized ML algorithms and advanced matching mechanisms underscores Meta’s proficiency in refining targeting precision, ultimately enhancing campaign efficiency and yielding more impactful outcomes. – Berkan Şişman

Through BigQuery, lcwaikiki.com strategically harnessed data, optimizing Meta campaigns for substantial improvements in user engagement and conversion rates. The personalized ML models and advanced matching techniques enhanced targeting precision, leading to increased campaign efficiency and more impactful results. – Elif Buse Betoner