Advanced Automated A/B Testing (DCM/DBM)


In 2018 our advertiser, active in the restaurant industry, was ready to approach the market with remodeled restaurants, their own CRM systems and profound market research. Their biggest objective was to obtain a strong uplift in restaurant visits & customer retention while also increasing revenue generated.


The advertiser is a new player in the digital landscape with limited experience and knowledge but they wanted to launch a complete digital strategy in 2018. However, they were uncertain about the message and themes they wanted to use during their campaigns.

Therefore the advertiser wanted to run an ‘AB’-test with 18 themes (per language), that each had their own landing page adapted to that message. In combination with this AB-testing, the advertiser wanted to test a downloadable coupon per theme which also changed weekly. Additionally, for the coupons they wanted to know which one was downloaded the most and which ones were redeemed the most offline.

Technical Tool

In the initial planning stage, we advised the client that it would be absolutely paramount to collect enough data in order to generate statistically relevant results and ultimately, conclusions. Therefore, we used statistical tools to determine the amount of impressions the advertiser would need in order to successfully test 18 themes against each other. Afterwards, the next logical step was advising the client to use DCM as it allows to add multiple creatives to one ad and apply an even creative rotation between them.

Secondly, in order to address their challenge regarding the coupon tracking we pointed them towards applying custom floodlight variables to their tags in DCM. Since adding a custom variable of couponID permitted the client to specifically track which coupon has been downloaded after a users saw or interacted with the campaign. Adding a URL variable also enabled them to check on which page the coupon was downloaded. This allowed the client to check the post-view behaviour on the website

After assisting the client in the planning phase, they implemented 18 creatives per Ad for each creative dimension. They then made sure that all creatives followed the ‘even’ creative rotation. When all creatives were in place, they used the campaign spreadsheet and applied a formula in order to set the specific landing page per theme for each creative.

Results and Business Insights

The campaign did not have the “classic” campaign objective of branding or sales. The objective was to analyze the behavior of customers regarding the different messages and different coupons and to transform this data into actionable insights. However we also assisted the client in looking for the best performing creatives of the campaign.

Furthermore, based on the advertiser’s data of the coupons used in physical stores they saw 4x more offline ‘conversion’ compared to the same period last year, when they did not run programmatic campaigns.