Danone Boosts Customer Intelligence and Addressable Reach Using LiveRamp’s Safe Haven

Solution

In order to meet its lofty, long-term objectives, Danone turned to Numberly and LiveRamp to design a test centred on two objectives: 1. Understanding the impact of digital advertising on audience segments and their path to purchase at one large retailer. 2. Optimising omnichannel activations based on transactional data provided by that retailer. Danone started by tapping the data marketing experts at Numberly who deployed LiveRamp’s Safe Haven. Safe Haven is an advanced data foundation that supports omnichannel activation, deep record-level modelling, and full data security to enable collaboration among multiple trusted business partners. Safe Haven would provide the people-based regulatory compliant foundation for refreshing Danone’s consumer intelligence.

Next, Danone worked with Numberly’s analytics experts to assist with test design, historical segment refinement, campaign strategy, and marketing measurement. Numberly also helped identify missing data elements sourced from retailers in order to complete the analytic work with high precision. Finally, retail partners were invited to discuss the needed data collaborations with Danone. These partnerships progressed quickly due to the data security embedded in Safe Haven and the fact that any retail data would be pseudonymised and free of directly identifying information. Data was always managed and under control, as Safe Haven did not allow the copying or exporting of data records. All the pieces were now in place for Danone to run an experiment that would lay the groundwork for its larger goals of moving to post-cookie advertising and owning its data future.

Here’s How They Got There Pre-study:

Segment Review and Experimental Design The Numberly team recommended that Danone’s existing segments and campaign audiences be reviewed with respect to the new retail shopper information. Retail records were linked to revised audience definitions built with LiveRamp’s durable identifier as a join key for the data. Analysing historical revenue performance for these segments was now possible at the record level, and led to one segment being dropped from poor sales performance and two segments merging due to highly correlated audience behaviours. Additionally, two new test segments were created that were considered performance candidates from the linked shopper data. Based on this pre-study, an experimental design could be set up to measure and optimise real incrementality on those six segments.

Cookie-Free Audience Activation

Numberly activated segments directly from the Safe Haven environment, enabling a consistency across channels in defining Danone’s audiences. This consistent definition increased confidence in post-campaign insights and was also key to a successful holdout strategy. Accurately maintaining control groups across channels for each segment was important to not only measure lift, but also to achieve a true understanding of marketing’s incremental impact. The people-based segments had already helped by ensuring that there were no overlaps to complicate the analysis with shared impressions. Numberly was now prepared to execute an accurate 80/20 (exposed/ unexposed) audience split to provide a baseline for optimising the media mix customised for each segment during the campaign.

Execution and Measurement

Numberly activated segments directly from the Safe Haven environment, enabling a consistency across channels in defining Danone’s audiences. This consistent definition increased confidence in post-campaign insights and was also key to a successful holdout strategy. With Numberly’s assistance, Danone selected Facebook and DV 360 as the first channels to pilot. Numberly developed campaign media alternatives to test traditional segment-specific appeals alongside new creative informed by data similarities with neighbouring segments. Numberly could then optimise the creative and the bidding strategy per segment to maximise real incrementality over the course of the campaign.

One surprising result that was discovered early in the analysis was a weakened correlation of engagement metrics (e.g. video completion rates) and sales performance. Without access to people-level revenue analysis, previous campaigns had assumed publisher-provided impression and engagement metrics were indicators of sales intent. Numberly’s daily analysis and weekly strategy reviews revealed that this was not the case, and future campaign decisions had to separate these factors and apply them independently to specific campaign objectives.

When the campaign ended, the overall lift in conversion rate was an impressive 7.9%. During this time, the retailer’s website also saw increased traffic and engagement on Danone’s product pages: e-commerce conversions increased by 22.4% and overall total revenue grew 24.7%. Surprisingly, Numberly’s two newly discovered audience segments were the highest performers overall. Danone’s initial goal of improving its media strategy succeeded, as did its aim of increasing customer intelligence, challenging long-held assumptions, and unlocking insights it would have not obtained otherwise.

Result

Danone successfully optimised audience segments and achieved a strong level of consumer addressability and measurement, enhancing their consumer intelligence in the process. This test was groundbreaking for Danone. The techniques and teams used to achieve these results now form the basis for its new internal marketing operations, allowing the brand to own its data and build direct relationships with consumers—a rare feat for a CPG, but not an impossibility with data collaboration.

Current State

Danone’s journey brought it to a place of increased marketing agility. By collaborating with retail partners and leveraging LiveRamp’s Safe Haven and Numberly’s data scientists, Danone is now able to:

1. Evaluate business objectives alongside marketing budget per channel.
2. Accurately measure KPIs on a weekly basis.
3. Optimise retail channels to align with business priorities.
4. Run campaigns using a durable, non-cookie-based ID.