Cookie Life – Part 4: Data Clean Rooms

I wrote this article as part of a series published on the Mercury Media Technology GmbH blog.

As part of our series discussing the evolution of digital advertising triggered by the concerns about privacy on the web, we have published articles on how to target consumers without third-party cookiesand how to start collecting first-party data. This piece aims to clarify how you can benefit from data clean rooms when you use your first-party data for advertising purposes. 


  • What are “data clean rooms” in the context of advertising? 
  • How do data clean rooms work?
  • Why set up a data clean room for your company?
  • How to get started with a data clean room?
  • What are distributed clean rooms?
  • What are deficiencies of data clean rooms?

What are “data clean rooms” in the context of advertising?

The name “clean room” originates from manufacturing sites that require extreme protective measures to prevent contamination. 

Definition Data Clean Rooms

data clean room in advertising is a safe space that offers access to sets of aggregated & anonymized PII data that can be cross-matched for measurement, analytics, and targeting purposes.

Foggy? Let start with some explanations: 

  • Safe space: privacy-first closed-environment with pre-defined requirements to keep the data anonymized.
  • Aggregated & Anonymized: individual first-party user information is #ed (read hashed for unrecognizable characters) and grouped by clusters. No cluster can be limited to an individual or a few profiles. 
  • Matching: level of relevance of the two distinct audiences, one versus the other. 
  • PII data: (Personally Identifiable Information) any information that can identify a user: email, phone number, name, etc.

How do data clean rooms work?

Let’s take an example! The most famous use case is the walled gardens: they have individual data from billions of users, but they only provide access to anonymized data in cohorts to the marketers willing to leverage it. Advertisers of all sizes can define target groups by selecting a set of variables, from location to interests. The profiles of users are anonymized, and the advertisers can see the delivery data but no individual identities (PII). A clean room is a “neutral“ storage in the sense that no personal data can be extracted by any stakeholders: in the case of walled gardens, the advertisers can leverage it without seeing the names and profiles of users, which solves the issue of user privacy. 

If the advertiser has its own first-party data collection solution, it has the capacity to create its own cohorts. By building a proprietary solution or using partners available on the market, it can aggregate and anonymize its own data to match with other partners. No individual profile will ever be shared.

Data Clean Rooms_mmt_Cookie Life 4

Why set up a data clean room for your company?

  • Accessing data from partners without giving away your first-party data: Brands can protect the PII information they have been entrusted with by their customers and fans. Still, they can use it to find their alter-egos on the web (open or walled gardens) without any risks of breaching privacy laws. 
  • Scaling your first-party audience from the start: When you decide to ramp up your first-party data collection effort, optimising the strategy and acquiring a loyal customer base takes time. Clean rooms allow brands to multiply their audience very quickly. 
  • Assessing external partners based on the audience matching rate: Building the right partner synergies requires thorough scrutiny of both sides – the most efficient partners and audiences. 

How to get started with a data clean room?

Data clean rooms provide the right technical environment to apply your first-party data before interacting with other data sets. There are precise requirements to fulfill in order to connect CRM solutions to it. They need good quality data and good data governance to be used to the fullest, which may entail hiring data scientists and an analytics team to manage clean rooms. 

To get you started in your setup, please consider the following: 

  • If you have already set your own tech stack, which clean room is compatible with it? If you have not started your first-party data collection, you may need to consider choosing your tech stack and clean room accordingly.
  • If you use the service of a network agency, which solution are they powered by? Most agencies have chosen a partner to power their media buys, it is worth checking with them. 
  • How competent is your current data science team in terms of leveraging the system? Clean rooms require specific efforts and their power needs technical skills to be properly unleashed.
  • Do not underestimate the legal aspects of data privacy and data sharing. It takes time and should be anticipated as part of all partners’ negotiations. 

There are multiple providers in the market who help to set up a data clean room and analyze the various requirements and options available. We are there for you if you need first-hand help.

What are distributed clean rooms?

“Distributed”, or “decentralized” clean rooms means that different partners keep their databases on different platforms without the need to migrate, share or centralize the data – it can still be analyzed in a seamless way. 

This concept is relatively new. For example, InfoSum lets advertisers load their CRM data into a personal “Bunker” whereas media owners can upload their addressable audience data into another “Bunker”. After that, multiple Bunkers can be virtually connected through anonymous mathematical representations, which enables audience matching and subsequent comparison of conversions to advertising exposure to attribute new conversions to specific channels and measure the efficiency of the campaign. 

Data Clean Rooms MMT
©Kim Heisler

Image based on Merkel’s graphic

This distributed setup, however, implies difficulties with governance and security and should be handled professionally.

What are the deficiencies of data clean rooms?

Most data clean rooms nowadays are still limited to one specific platform, which means marketers cannot gain a full picture of a customer’s journey or trace cross-platform attribution.

For example, the walled gardens only allow the use of their data on their own platform, thus forcing advertisers to focus their budgets on respective channels. But that may change in the future – and we will make sure we keep you posted. Stay tuned! 


Special thanks to Lily T. who provided me with insider tips on clean rooms considerations.

Some data clean rooms are available on the market:

Cookie Life – Part 2: Targeting without third-party cookie

I wrote this article as part of a series published on the Mercury Media Technology GmbH blog.

To follow up on the status quo on cookieless future, we would do our first dive on cookieless targeting as it offers an opportunity to launch very quickly.

We will give here a quick look into what are these targeting options. How to leverage it properly and why contextual is the trendy solution that all media agencies are talking about.

Which targeting relies on what? What is targeting? Which targeting is affected by third-party cookies?

Definition Targeting

Targeting means the precise addressing of target groups in online marketing. The most important prerequisite for this is determining the target group in advance of any advertising campaign. Advertisers have numerous options to target an online campaign precisely.

To go back to the status quo article, I mentioned different targeting options currently in use. Let’s clarify which ones will remain alternatives in the future framework.

MMT - Cookieless advertising - targeting option without cookies
  • SocioDemo: using sociological and/or demographic characteristics to target eg. Age group would be possible via Cohorts / FloC or opt-in data. We will cover it in part 3
  • Geolocation: relating to the use of lat-long positioning of the device use, this relies on realtime data that can be deactivated but do not rely on third-party cookie. 
  • Behavioural: privacy laws also aim at the capacity to follow users from one site to another. So this will be affected by the loss of third-party cookies. 
  • Time/season based: based on external circumstances- e.g connected to the weather forecast, time of the day, or temperature, it is not affected by cookies. 
  • Contextual: based on the content of the hosting page, this is not influenced by cookies. 

Why contextual targeting is (once again) cool?

The name is explicit – advertising in context – car insurance presented on a blog presenting car model reviews and comparisons. It is not new, we even used it before the internet existed! Yes, the print channel had Interest titles which was a great opportunity to talk to niche audiences. It is still used for specialized products like hobby groups or sport material aficionados. 

It emerged in digital advertising in the same way as an offering to help focus advertising based on interest, digitally, it requires to employ a technology which recognizes index or “read” the content of the page to analyze its content and allocated it to a “context” group.

Google Adsense (created in 2003 thanks to the acquisition of Applied Semantics) is a good example of the usage of contextual advertising solutions based on keyword-indexing. It has its detractors as it also created an enormous volume of sites uniquely design to attract Adsense ads by aggregating some other sites’ content on a topic, not bringing particular quality to the market. 

The main benefit of contextual advertising is relevance: there is no need of knowing who the person is if you just need to know what she wishes. It is based on real-time interest instead of past behavioral data analysis. It increases the chance of conversion when approached properly (let’s talk about next) as it focuses on the theme of the content and not the person reading it, It is offering an alternative to all companies which have not started collecting their own customer’s data (let’s talk about that too). 

How to leverage the targeting options that are not relying on cookies?

As said, Google pushed back the end of third-party and many marketers sighed with relief. It is just an opportunity to ensure we do not fall behind. There is no time to waste in starting to develop internal knowledge of the most efficient channels for your brand outside of this cookie world. 

One essential solution entails allocating a budget for a clear test and learn approach potentially using the SMART framework: Your general objective remains to understand which targeting is the most relevant to create awareness or convert to sales (as both are important to run on a long term focus) 

  1. Establish hypothesis
  2. Determine how you will measure success
  3. Define a time and budget limit
  4. Assess results 
  5. Share learning

To make your “test and learn” budget in contextual advertising a success.

  • Adapt the creative to a contextual environment, relating to the right topic assignment rather than the consumer-centric approach.
  • Monitor closely the performance of each context and associated creative to learn which are the best matches.
  • Internal factors can play a role in each activity, do not hesitate to use granular monitoring solutions to deep dive into your results when needed.

For context

The scalability of the campaign also depends on the quality of the contextual sorting. IAB has released a list of general categories on which website pages can be classified. It offers an opportunity for any publisher to list itself in a category of content, but this is still generic. In contrast, the niche of a context can end up impossible to scale as not enough content will be available. 

Good AI leverage Natural Language Processing (NLP) can now determine the sentiment of the pages which reduces the risk of brands being presented against non-relevant or offensive content. It avoids excluding entire categories out of fear of falling against the wrong news (e.g. breaking news section, user-generated content such as reviews and forums). There are good pre-bid solutions now starting to be integrated as part of the brand safety offer of adservers. Check your options. 

For other external factors

Build your company knowledge of the influence of external factors by approaching such categories with solid test & learn hypotheses and structure including the creative drawing a clear picture of this option for you. Depending on your service or product, the advertising can be tested on:

  • Location 
  • Weather 
  • Time of the day 
  • Period of the year

It is worth paying attention to these factors when running campaigns, it can save you cash by adding an extra relevance layer to your campaigns. 

To close

Remaining independent from large providers is a strategic decision to make, whereas it is always possible to choose to rely on a walled garden solution to provide the required data to leverage. No company should feel doomed to adhere to the technological limitations dictated by the market. 

Alternative targeting does not rely on third-party cookies can yield great performances once the understanding of the best approach is integrated. As it relies not on consumer-centric data, it requires assimilating a new mechanic and that implies testing. As long as Google offers time for that, there is not a second to waste in starting to build knowledge.

The next stop will be, first-party data collection and leverage, stay tuned.


Cookie Life – Part 1: Status Quo

I wrote this article as part of a series published on the Mercury Media Technology GmbH blog.

Google announced at the beginning of July that it will postpone the launch of the Privacy Sandbox solution in Chrome to the “end of 2023” hence maintaining the support to third-party cookies. It gives the industry some air to implement their new strategies and arbitrage for future open-web campaigns. Let’s use this time to understand what is at stake. 

Why is scraping third-party cookies from Chrome bringing so much trouble to the advertising industry?

  1. Because Chrome represents 65% of all web navigation (cf. Gartner) and 75% of mobile ones (cf Visual capitalist). 
  2. Google alone attracted almost 31% of the total US ad revenue in 2019 (eMarketer)  (2020 being disturbed by Corona) across all formats (display, search, video- desktop, and mobile) and the revenue is growing.

Data collection and capacity to leverage it is becoming concentrated in the hands of a very few actors(such as Google, Facebook Group, and Amazon) due to the limitations that various data protection laws have triggered in the western world. 

To set the scene, let’s double-click on the context:

  • What is Data Privacy Protection?
  • What is classified as Personal Data?
  • What are advertising Cookies?
  • What is changing after the Elimination of Cookies?
  • What are the Alternatives to Cookies?

What is Data privacy protection?

Multiple examples of privacy laws around the world (non-exhaustive list):

  • The European Union General Data Protection Regulation (GDPR) in 2016
  • The Personal Information Protection and Electronic Documents Act (PIPEDA) in Canada in 2018
  • The General Data Protection Act (LGPD) in Brazil in 2018
  • The Data Protection Act in the United Kingdom in 2018
  • The Personal Data Protection Bill in India in 2019
  • The California Consumer Privacy Act (CCPA) in California (US) in 2019
  • The Protection of Personal Information Act (POPI) in South Africa in 2020
  • The Data Security Law in China in 2021
  • Various privacy laws in effect in Australia 

Something all these laws have in common is that:

  1. There are sensitive personal data that should be protected 
  2. Businesses can be held financially liable for the mishandling of data at different pipeline stages including secure processing, storage, and transfer (free or at a cost) 

Commonalities of data protection laws

Sensitive personal data should be protected. Businesses can be held financially liable for the mishandling of data at different pipeline stages including secure processing, storage, and transfer (free or at a cost).

What is classified as Personal Data?

Definitions vary from country to country, for example in Europe it says that personal data is any information that relates to an identified or identifiable living individual: 

  • Name and surname
  • Home address
  • Email address 
  • Identification card number
  • IP address
  • Advertising identifier of your phone
  • Health data which could be a symbol that uniquely identifies a person
  • Cookie-ID or location identifier

What are Cookies?

The identification cookies are files created by the websites you visit. They make your online experience easier by saving browsing information (the name or address of the site you visited, your ID, and timestamp). 

There are two types of cookies:

  1. First-party cookies are created when one visits a website. They are set by the site shown in the address bar.
  2. Third-party cookies are created by other sites that own ads or images on the visited website.

With technical cookies, sites can keep you signed in, remember your site preferences, and give you locally relevant content; such cookies do not require the user’s consent. 

Third-party cookies provide support to advertising targeting: they enable the selection of ads according to behavior as well as historical partners identifying user’s interests and potential needs.

First things first, let’s clarify what has happened to data collection via cookies:

Advertisers rely on cookies to track behaviors across the open web. Cookies help follow users from one site to another to serve them with ads that are coherent with their actions. At the same time, cookies record what the users have already seen for future reference (for example, for retargeting purposes).

Here is a timeline of the milestones on the path towards cookie eradication:

MMT - overview of advertising cookies evolution that affects targeting

What is changing after the Elimination of Cookies?

The elimination of cookies compromises the measurement of campaigns on the web and cross-device attribution. We are turning to a consented information gathering: cookies will be saved in the browser and will get no access to the user’s history of actions prior to sending an ad. Campaigns will be running “blind” without any historical information on users.

This entails the following:

  • It will incapacitate many actors involved in digital retargeting 
  • It will limit the ability to create lookalike and similar audiences based on collected first-party data 
  • Performance marketing activities, in general, are going to be handicapped
  • Lookalike and enhanced targeting will be limited 
  • Digital attribution models outside of walled gardens and cross-device targeting are going to be extremely limited.
For advertisersLosing the base for attribution hence no capacity to build conversion funnelsDecreasing cost efficiency of digital campaigns
For publishersLower ad revenue => need to find new sources of fundingConsent world for user identification
For Walled GardensMonopolization of the identification capacityNew sources of revenue
For Open Web data providersLosing the key source of revenue
For end-consumersLower ad relevanceLess intense ad stalking

What are the Alternatives to Cookies?

The whole market is trying to find the most efficient solution to overcome this challenge, with or without leveraging Google’s new offer:

  • First-party data, e.g. email addresses are back as the golden ticket to recognize users across devices and platforms and collect user data
  • Walled gardens will keep offering the same level of granularity in targeting, thanks to the user consent upon login and personal data processing agreements. They have developed network outreach => advertising offers outside of their walls.
  • Cookie alternatives like various technological solutions aiming to replace cookies are already either available or in development. We are planning to give you an overview of these in the next articles. 
  • More traditional targeting options including contextual advertising, geotargeting, seasonality, and timing can be leveraged to guarantee some sense of going back to classic tactics – brands can rely on the specialized websites and relevance of influencers to promote their brands to the users, pending on season, hours of the days and locations, thus limiting wastage. 

To close up

The topic is ongoing – it is important to grasp the context to understand what is coming and build a customized solution for each individual business. There is no one-size-fits-all approach, as business needs and target group behaviors are genuinely different from one sector to another. This series of articles aims to help you navigate through the changes – MMT is here to help you build your #futureproof data solution.


Link to legislative documents:

Keeping Advertising yearly commitment in check

I wrote this post for the company I work for – MMT – it is also posted on their blog

Every year in June we hear about the Upfront market in the US, and this COVID year, particularly as it, could not happen in its normal sense. But the other markets or media are not exempt from this seasonal exercise: the negotiation of yearly advertising is a global practice of the industry and exists for all media. Let’s clarify what these are, who they benefit, and why it can make sense. 

What are media yearly commitments? 

The idea behind yearly media commitment is for advertisers and agencies to commit to a level of spend for the year with a specific partner. Long term plan of media investment is signed as a contract with publishing houses, tv channels, web giants, and so forth. The contract is binding and disengaging has consequences – usually financial – to guarantee respect for the conditions: spend per quarters, level of visibility, etc. 

What are the benefits of yearly commitments? 

An essential reason for commitment is visibility: it keeps publishers able to provide the quality content that brings the quality audience to tune in. 

In return for their engagement, advertisers benefit from:

  • Discounts: promising funds offer great discounts rates on top of agency ones. 
  • Beat yoy inflation: As part of the discount, there is also the engagement of not applying year on year inflations. Long term partnerships have led to some advertisers having discount rates off the charts. 
  • Visibility: it guarantees the impressions and can also include the placement quality 
  • Mention: it is the unofficial part of the deal where editors accept to mention or not mention brands as part of their “support” to the publishing house. It is very important in the fashion industry in the female magazine industry
  • Access to data: Data from WG are so scarce that advertisers are trading data as part of the commitment: “tell me who are interested in my product and I engage more”. 
  • Free services: As part of the package, Publishing houses can offer a strategic recommendation, community management, content production, insight, and so on. 

The goal is to secure as much budget as possible from the advertisers and new revenue streams for the teams

What are the cons of committing?

Advertisers are legally bound: they actually have to spend! You are committed to an entire year; so change of market – like 2020 –  requires a lot of negotiations to maintain flexibility, goodwill, and rates. That requires a lot of creativity in the way of building the right balance between benefits and engagements

There is no reason for any publishers to not try to help you remain competitive and accept to modify when commitments cannot be met. But this requires an investment in resource and time, strong negotiation skills, and grit!

Which type of clients are recommended to engage in yearly commitment? 

Any advertisers investing over 300,000€ on a single partner may consider negotiating extras by committing their budget. Obviously, this commitment makes sense when there is a clear correlation between advertising and sales volumes / ROAS and financial flexibility is not in discussion.

Also, when premium visibility/positioning is a priority and guaranteeing presence around key events, long lead time helps to secure these spaces. 

Finally, when budget performance is your priority, it can push the needle an extra bit for ROI results. 

How does it work in practice? 

As a newcomer, you benefit from new business negotiation rates/agency rates if you use one. Generally, major advertising investors (two digits million euros) are organizing these sessions to gain the benefits mentioned above. 

German market negotiations kick in in October for all media publishing houses, OOH, audio, walled gardens and so… The objective is for everyone to maintain a form of visibility on a base level of investments. Obviously, advertisers do not have to engage all budgets in commitments, they can decide to keep some “buffer” for last-minute deals.

Right message at the right time – why is timing important ?

« there is a time and a place » for everything …

Over the last two months we mentioned the evolution of media and the importance of the message to be heard in a fragmented space. We now need to approach another important point of communications. Timing !

The Status quo is simple : No one can live without habits or our brain would be overloaded with permanent choices to make, a study says that we make 35,000 decisions per day that we need to prioritise. Everyone builds habits, patterns they follow in their daily routine, and that includes the media they trust to provide information and news.

When talking business and time together, we all have in mind « time is money » about the financial trade in general. It is not completely excluded of Communications as well, we can define for reasons for which time as to be taken into account for any marketing plans:

The objective of your communication influences its timing:

When one knows that the actions led by a communication will take a long time and will require planning to execute (for instance offline sales vs. online phone based registration), it is essential to understand the best timing for your clients to act – for exemple catch them on the way to the shop on the radio, with a SMS in front of the shop, or online when they search for ideas.

Modern platforms offer opportunity to adapt the message to the time of the day and the context

TV has always offered day-part copy rotation, digital communications increased a great level of flexibility including contextual adaptation. Morning advertising for a restaurant can push breakfast menu where evening could present today’s dinner special’s adapted to the type of audience.

Communicating all the time is expensive, observing right timing create efficiencies.

If your business have opening hours and your communications objectives if for people to call you, you may consider advertising during your opening hours only, or change your company client support system to adapt to the marketing campaign.

Catch your audience at the right moment for them to be open to receive the message.

Attention span is shorter, people are over exposed with thousands messages a day. It is necessary to cut through the clutter, but with what? If you have bigs bucks, the sprinkler approach works : you communicated all the time, which reach in mind to guarantee high repetition. If you do not have big bucks : choose your moment.
There is a space in which your place can be found in the routine, which would instil the message your company wish to pass.

Need any help to create efficiencies in your marketing plan?
Get in touch.

Stay tuned and do not hesitate to request more details.

What is Media vs Message ?

After a conversation on communications strategy with one of my clients, I realised it is not generally clear to businesses what the pillars of communications – media and advertising – are.

These are two distinct but intertwined elements that we are going to discuss over the next few posts covering:

  1. What is media vs. message ?
  2. The right message at the right time – what is the right message?
  3. The right message at the right time – when is the right time?

Just to clarify – let’s start with definitions :

‘Medium’ (singular of media) is the vehicle through which information is transmitted.

Any type of vehicles that carry information can be considered media that can influence our view of the world: publisher platforms, social media, word-of-mouth, goodies, branded content, sponsoring, endorsement, influencer program, video games… The number of media vehicles available not only have increased but their « quality » -or trustworthiness- vary greatly.

Graph of the evolution of media offer – based on a Perspective graph edited by Carat in 2008

Where ‘Message’ is the content that is transmitted.

Messages can take the form of news, educational content, stories, anecdotes, satires, propaganda, advertising, and so on…

What are the characteristics of Media?

  • Shelf-life: Each medium has a length of time impact: where an outdoor poster can stay up for a month to be seen 60 times by commuters. A social media post would have a few hours – if not minutes – of impact to be seen once by an audience.
  • Memorizing stickiness: Each vehicule needs a number of repetition before the message sticks, where 75% of cinema goers remember an ad in one shot, TV needs of 3 to 4 to reach the same.
  • Engagement: Each vehicle can instil specific action or response; like create curiosity, provoque a reaction, searching about a product, participating into a poll, keeping an information top of mind for a few hours, or just build trust.
  • Image: Each vehicle has a perceived image by the public: trustworthy, statutory, specialised, entertaining, the choice of media and so on..
  • Reach: Each medium has a particular capacity to reach an audience over a limited period on time.
  • Message complexity: Each vehicle carries a specific depth of details: where print permits long explanations, display ads allow only short headlines.

The combination of these characteristics plus the study of the audience to address is the basis for the media strategist to advise on the best vehicle to use for a specific communication objective.

Nowadays – with the fragmentation of media channels – it is sensible to involve the media agency into the creative process to ensure that the message is related in the relevant way for the environment.

This will be the topic in our next post, stay tuned.