The insight they brought to the campaign, was that by using GroupM technology to perform a cluster analysis of visitors to the Promote Iceland website, we could segment visitors — based on their behaviour, demographic properties and interests — in a way that would not be possible if the client solely relied on the data and tools available from commonly used providers such as Facebook and Google.

Using this highly detailed segmentation, we were confident it would allow us to create detailed and accurate lookalike audiences. This would allow us to increase click-through and view-through rates by targeting exactly the right consumers for us to be able to drive highly-qualified traffic through to the site. We also expected to be able to work with Promote Iceland, using this insight gathered in the discovery phase to build creative precisely customised to each audience.

Based on the client’s objectives and the results of the cluster analysis, we decided to run display, native, video and social strands to the campaign. For each audience, they developed a bespoke buying strategy. Because audiences were scored against their interest in travel and their likelihood to convert, we could address the most valuable consumers more frequently, optimising as the campaign ran. This increased relevance and receded wastage.

Because the cluster campaigns were tightly segmented both conceptually and within the PBU, they was able to keep the data for each audience clean. Not only did this inform our optimisation strategy, it also allowed us to gather detailed insights for each segment.

Using the insights gathered during the campaign, they gave the client an audience framework. Using the framework, Promote Iceland can take more informed decisions about programmatic buying and creative targeting in future campaigns, helping increase conversions and reduce wastage.

The clusters significantly outperformed the broad-based activity. The click-through rate (CTR) on the broad strategies was 1.2%. In comparison, clustered audiences had a CTR of 6.5%. Taking the broad-based campaign as an index, with a value of 100, the cluster campaigns had a performance value of 540. The picture was similar for the video view-through rates, which had a performance value of 194 compared to 100 for the broad-based campaigns.

As this method was new for lookalike models, we were given 5% of the budget to use with our bespoke audiences. The remaining 95% was allocated across broad-based buying. The results were surprising: 20% of all interactions were generated by the bespoke audiences, despite these only having 5% of the budget.

And Promote Iceland also received the audience framework, containing insights into their core audience.