How Charlotte Tilbury Used AI To Deepen Its Data Pool

Despite what you may have heard, artificial intelligence (AI) is not a magic solution for every problem facing the ad industry.

But AI can help advertisers optimize the performance of their campaigns.

Luxury beauty brand Charlotte Tilbury, for example, recently tested a custom integration with Scibids, a company that specializes in AI-based marketing solutions for the demand side.

Over the course of two months, the brand was able to reduce its average cost per acquisition (CPA) by 29%.

More data, more predictive

Scibids created custom bidding algorithms for Charlotte Tilbury to deploy via Google’s DV 360 DSP.

The algorithms were designed to take upper-funnel data into account, including site visits, clicks and ad viewability. Giving Charlotte Tilbury more data to optimize off of – luxury brands with higher price points often have fewer online sales data to look at – helped expand the reach of its campaigns while still prioritizing performance.

Scibids was also able to reduce the frequency of retargeted impressions so people who had already been exposed to the brand’s messaging wouldn’t be bombarded. This led to reduced media waste and gave Charlotte Tilbury an opportunity to spend more on acquiring new customers efficiently.

Since it started using custom bidding algorithms, Charlotte Tilbury has seen a 60% increase in the conversion rate of its acquisition campaigns.

The performance is due to multiple factors working in tandem, said Andrew Leung, Charlotte Tilbury’s global head of performance marketing, including a “full-funnel learning strategy, accurate conversion weighting and the AI being able to refresh models every few hours based on its constant monitoring of the campaign signals.”

Fewer conversions means less data

Layering in more upper-funnel signals proved to be a viable work-around to an issue facing many luxury brands: Fewer people tend to purchase expensive products, especially online, which means ad campaigns promoting these products drive fewer conversions. The result is less data to pull from to inform future campaigns.

Acquisition campaigns are especially vulnerable to this dynamic, because the number of people willing to buy an expensive product from a brand they’ve never purchased from before is often low. That can lead to a higher average CPA compared to acquisition campaigns for lower-cost products.

Pulling in data that is not typically used for targeting to predict which impressions are more likely to yield conversions can change that dynamic, said Scibids CMO Nadia Gonzalez. That data can include customer relationship management data as well as measurement and attribution data.

“Scibids made it possible for us to optimize toward multiple KPI scenarios efficiently and effectively within our DSP at scale, which is difficult to do manually,” Leung said. “Multiple KPI scenarios means the ability to buy media against custom quality metrics and business goals – essentially not your typical KPIs.”

To measure the effectiveness of its custom bidding algorithms, Charlotte Tilbury used the experiment and brand lift features within DV 360 to A/B test insertion orders (IOs) using the Scibids AI against a control group.

Within three weeks, the AI-based custom bidding IOs began to exceed the control group’s CPA performance, Leung said.

Achieving conversions at a more effective price allows the brand to use its campaign budget to buy more impressions, which increases the likelihood of additional conversions.

The custom data sets created by the AI are exclusive to Charlotte Tilbury, and the brand plans to continue to use AI to inform future campaigns.

“The world is only becoming more digital, which means more data to make sense of,” Leung said, “and AI is part of our strategy as [we] continue to grow.”

Enjoying this content?

Sign up to be an AdExchanger Member today and get unlimited access to articles like this, plus proprietary data and research, conference discounts, on-demand access to event content, and more!

Join Today!