Machine learning helps Heureka win customers cost effectively

Heureka aims to make shopping online simple, safe, and fast. Every month over 20 million people use the price comparison site, which is the largest e-commerce platform in Central and Eastern Europe. Visitors can search for products, read and contribute reviews, and buy from a wide range of categories. To attract new customers and bring previous visitors back to the site, Heureka developed a new campaign based around video advertising.

Goals
Attract new customers and increase returning site visitors
Achieve cost per acquisition equal to or better than generic search
Approach
Converted existing video shopping guides into TrueView for Action ads
Tailored messaging to previous site visitors and new users
Ran always-on campaign using Smart Bidding strategy tCPA
Results
13% of users finished entire six-minute video
Cost per acquisition fell by 78% over campaign duration
Achieved same cost per acquisition for new and previous site visitors
33% lower cost per acquisition than generic search

How the e-commerce platform used machine learning to attract site users at an ambitious cost per acquisition target.

Spotting a chance to use existing collateral, Heureka converted video shopping guides that focused on specific product categories (speakers, coffee machines, and headphones) into three separate ad creatives of four to six minutes in length. By developing these for YouTube’s TrueView for Action format, the team wanted to stimulate viewers to click directly to the Heureka website.

The team ran the ads in an always-on campaign and used the Smart Bidding strategy tCPA (target cost per acquisition). By using machine learning in this way, their aim was to do as well or better than the average cost per acquisition of their existing generic search campaigns.

Although the creatives were relatively long, audience retention was high. For instance, of those who started watching the six-minute video, 13% completed it. The ads managed to achieve the same cost per acquisition for new customers as for those who had visited in the previous 30 days. Overall, the YouTube campaign delivered a 33% lower cost per acquisition than Heureka’s generic search campaigns. And thanks to Smart Bidding, over the course of the campaign so far the cost per acquisition has fallen by 78%.

“Our main goal with these videos was to become more relevant in the decision-making phase of customers' shopping journeys. Smart Bidding brought us closer to relevant customers and enabled us to be much more efficient. We plan to test more video creatives and cuts in the future to reach even better results.”

– Marek Dobrý, Brand Manager, Heureka

An integrated online strategy helps Lindex increase summer online sales by 37% YOY