Table of Contents
If you work in digital marketing, you have certainly heard of “Data Clean Rooms”.
Generally, the concept is brought up when the conversation turns to proprietary data (first party), GDPR standards, the imminent disappearance of cookies, etc.
- But what is a Data Clean Room?
- How is this technology respectful of data privacy?
- What does this actually do?
- Why did Google decide in 2017 to launch the first data clean room offer with Ads Data Hub?
Follow the guide !
What is a data clean room?
A pragmatic definition
A data clean room is an environment that allows two parties to share data about their users, securely & anonymously. The aim is to carry out analyzes intended to improve an advertiser’s marketing and advertising strategy.
In other words, a data clean room allows an advertiser to “mix” his first-hand data (first party) with the ultra-sensitive data of another partner (such as Google, Facebook or Amazon) to bring a new light on the effectiveness of its campaigns and the profile of its target audiences.
By combining anonymized data sets with those of their partners, marketers can break down silos and obtain a more holistic view of the customer journey, a better understanding of their media attribution, and achieve more relevant segmentation of their audiences…
A data clean room allows 100% respect for data confidentiality
A data clean room acts as a “neutral territory” between two parties who want to combine their data, in a secure and confidential way. To use an analogy, a data clean room is a bit like the “Switzerland” of data.
At no time does either party need to manipulate the other’s data. In addition, such technological infrastructures make it possible to prevent any leakage of confidential data (data leakage).
And to avoid any false notes, the data room will filter all sensitive data upstream and apply a certain number of rules to ensure that user confidentiality is 100% respected.
If we take the example of what Google does with its Ads Data Hub data clean room, we see that 3 principles are in place to guarantee data confidentiality:
- SQL queries made by data scientists that are likely to extract confidential data are identified and blocked upstream.
- Before returning any data, Google ensures that the population studied is large enough to guarantee that the data will remain anonymized.
- Finally, if by combining certain datasets with others it is possible to go down to the user level (i.e. to be able to analyze the figures of a single and same user), then Google will automatically remove the data from the reports.
Data clean rooms are mainly intended for large advertisers, given the complexity of implementation
From the start, data clean rooms (first offered by GAFAM) were mainly intended for large advertisers. Indeed, the implementation is still somewhat “painful”.
Just look at the ecosystem needed to use Google’s ADH solution to understand that it is not within the reach of ordinary mortals.
It is therefore unthinkable to be able to use these data clean rooms without a solid team of analysts and data scientists, experienced in business intelligence and SQL queries. The view of a media / marketing specialist is also recommended, in order to be able to give the commercial sense necessary for the use of the data.
What are the use cases of a data clean room for an advertiser?
Here are several examples of what a data clean room can do for an advertiser.
Analysis of the reach and frequency of advertising exposure
- An advertiser advertises on YouTube, and on premium inventory (e.g. the advertising space available on the home page of the Le Monde site) and wishes to know how many times the targeted Internet users have been exposed to his advertising, total.
- An advertiser who uses all of Facebook’s advertising networks (Facebook, Instagram, Audience Network and Messenger) can find out how their deduplicated audience is distributed within the 4 environments.
- An advertiser who does display advertising on several inventories can use a data clean room to find out if he can generate as many conversions, while lowering his overall repetition (the number of times a user is exposed to the same ad) and thus spend less media budget.
Measurement & Attribution
- An e-merchant who sells their products on Amazon and promotes them through Amazon’s programmatic solutions can analyze the impact of their campaigns with their own attribution window (understand how many sales are generated by the campaign at different intervals of time).
- Thanks to the features of Amazon’s data clean room (Amazon Marketing Cloud), an advertiser can analyze cause and effect relationships and correlations between different marketing factors in order to generate learnings that can improve their return on advertising investment.
Consumer Insights
- Two brands combine their 1P (first party) customer data to measure the overlap between the two audiences (X% buyers of brand 1 are also buyers of brand 2)
- A brick & mortar advertiser can build a profile of their best omnichannel customers with the interests of Internet users that the GAFAMs know (what a person is interested in, the type of sites consulted, what they are about to buy, etc. )
retail media
- A manufacturer can combine its data with that of a distributor to find out if some of its ad campaigns have led to in-store sales.
- A brand can use a data room to combine its data linked to its loyalty cards with several publishers, and thus know which sites / content are consulted by its customers.
Media Activation
- An e-merchant could ask Amazon, via a clean data room, to provide a segment of users who are buyers, both, in category X and Y, to reach them via programmatic advertising, using the Amazon DSP.
- An agency can use a GAFA’s clean data room to analyze the types of audiences that have the best predisposition to buy their client’s product.
History of the data clean room: understanding the problem with the launch of Ads Data Hub
Before, advertisers collected data via third-party tags (called “Pixels”)
It was Google that initiated the “data clean room” movement in 2017, with the launch of ADH (Ads Data Hub).
Before ADH, advertisers and their agencies required pixels to be implemented in their campaigns. These code snippets directly collected data for measurement purposes (how many times was my ad seen? How many times was it viewed at 100%? How many clicks were made? etc.)
This practice, which ultimately allows third parties to collect data directly from the platform, can represent a security risk :
- Who is really behind the collection of this data?
- How will this data really be used afterwards?
- Are the resulting insights respectful of privacy?
Especially since the “piggybacking” was legion; that is to say that a pixel could “embed” others, thus making the collection of information very opaque, even for YouTube itself.
Limit third-party data collection to the user level, for greater security and privacy
Therefore, Google has decided to simply ban the use of third-party tracking pixels on its YouTube platform.
In the same vein, Google has also removed User IDs from its ad server, Campaign Manager, which prevents advertisers from analyzing each user individually, if they wish.
With all these changes, digital advertising has found itself “stripped” of a great power: that of delivering a granular measure of advertising effectiveness and the profile of the audience reached.
It was therefore necessary to provide advertisers with another means of measuring their campaigns, with more granularity, without taking the risk of exposing confidential data.
This is how Ads Data Hub, the first Data Clean Room dedicated to digital advertising, was born.
The use of data clean room is developing today throughout the industry
Today, clean data rooms are no longer the prerogative of GAFAM.
The major retail chains, such as Carrefour, Target in the United States, and even players such as Disney also offer clean data rooms to brands that pay them to use their data for advertising (retail media).
Some companies like Habu, InfoSum and Snowflake offer clean data room platforms, used by these players outside GAFAM, which then allow two entities to share their data, while respecting data confidentiality.
Conclusion
For advertisers, the real challenge behind the concept of data clean room: that of combining their data with that of other partners to collect more insights, with a view to optimizing their media investments.
Advertisers can thus better prepare, activate and measure their advertising campaigns using rich signals.
To see a use case from A to Z of a data clean room, I invite you to watch the following video from MightyHive, which relates the case study of an advertiser who used Ads Data Hub to better measure and optimize its advertising budget.
Data Clean Room FAQs
A data clean room is an environment that allows two parties to share data about their users, securely & anonymously.
The purpose of a data clean room is to carry out analyzes intended to improve an advertiser’s marketing and advertising strategy.
Among the existing data clean romm, we can mention:
- Ads Data Hub (Google)
- Amazon Marketing Cloud
- Snowflake
- Habu
- InfoSum
