Consumer privacy has become a business resiliency issue for Chief Marketing Officers (CMOs), their advertising teams, and the Brands they serve. Data privacy and protection laws, regulations and enforcement have exploded, globally. This has exponentially increased the rules and requirements for modern marketing solutions. Key pieces of law and regulation include the European Union’s General Data Protection Regulation (GDPR), Brazil’s General Personal Data Protection Law (LGPD), and five US state-level laws, including two in California alone – the California Consumer Privacy Act (CCPA) and California Privacy Rights Act (CPRA).
Kinesso and Matterkind, an operating division of Kinesso, are committed to the privacy of every individual whose information we process or maintain in our products and services. We provide people-first digital marketing and advertising solutions and technology services that help our clients connect and engage with people in more respectful and beneficial ways.
We believe that marketing and advertising done consciously, ethically, and fairly connects people to brand value, creates community, democratizes knowledge and access, and is a vital economic engine. This requires that marketing and advertising across connected digital channels is transparent, accountable, and trustworthy.
Google Marketing Platform (GMP) has been a longstanding innovator in digital marketing with powerful offerings across its entire product suite, including Display & Video 360, Google Analytics, Search Ads 360, and Campaign Manager 360. But in the product changes it has undertaken in recent years, one can observe an underlying idea that people shouldn’t have to accept being tracked across the web to get the benefits of relevant advertising, and that advertisers shouldn’t need to track individual consumers across the web to get the performance benefits of digital advertising. This idea aligns with our privacy vision as well, and our mutual commitment to privacy has been a driving factor in our growing partnership.
“As privacy and regulation evolves across the media ecosystem, how we continue to ensure we protect brands and consumers is at the forefront of all our thinking. As a component of our partnership, IPG and Google have an elevated focus on how we can bring privacy-centered solutions to our clients, protecting the data our advertisers value while continuing to elevate and protect the consumer experience,” said Vin Paolozzi, Chief Investment Officer at Kinesso. “The continued collaboration between our teams is inspiring thoughtful and conscious marketing solutions that will continue to transform our partnership and help lead the industry forward.”
In addition to its already strong suite of offerings, Google has become an industry leader in developing solutions that are privacy-preserving, often acting as first-mover to ensure the privacy of individuals’ personal data while giving marketers the necessary tools they need to be successful. This innovation trajectory has been both a proactive shift in product strategy to double down on privacy, as well as a response to similar initiatives by other actors across the digital marketing ecosystem that would have otherwise impacted the effectiveness of Google’s solutions. This privacy-by-design shift has occurred along core themes: respect for user choice, aggregation of data, gradual elimination of third-party identifiers, increased focus on first-party data obtained through consent and value exchange, probabilistic measurement to replace user-level tracking, and reimagining where necessary.
GMP’s Private-Yet-Performant Innovations
Privacy-by-design principles inevitably lead to signal loss, and over the past several years Google has rolled out a number of measurement and addressability innovations that aim to fill these gaps and provide the critical data granularity and powerful activation options needed for sophisticated marketers. Google has sought to balance developments that increase privacy and contribute to signal loss with others that accelerate performance and drive business results for marketers, reshaping their offering over the past few years to be private-yet-performant.
Ads Data Hub (ADH), Google’s clean room solution, was borne out of the privacy preserving trajectory of GMP. As a response to changing privacy legislation, in 2018 Google initiated the gradual deprecation of user identifiers in Campaign Manager 360 log file exports in various geographies. Understanding the limitations that this presented to marketers who used those identifiers for measurement and optimization, Google introduced ADH as a central analytics environment housing data from GMP where marketers could run queries to glean the insights that user IDs previously powered. ADH implements privacy-preserving controls that enforce the aggregation of data, preventing data from being generated for small groups of individuals. Since introduction, Google has deployed a number of product updates to increase usability based on user feedback, and has also collaborated deeply with key agency partners like Kinesso to develop complex custom queries that enable increasingly granular insights. ADH has also enhanced functionality by rolling out deeper integrations across the GMP suite and third-party measurement providers.
Google Analytics 4 (GA4) is the privacy-by-design successor to Universal Analytics, Google’s longstanding offering for measuring web analytics which will be sunset in 2024. Using an event-based framework that captures and reports on every interaction across websites and mobile apps rather than Universal Analytics’ session-based framework that focuses on web-based interactions within individual site visits, GA4 enables better understanding of how customers engage with brands, across all of a brand’s properties. By transitioning from third-party to first-party cookies, GA4 can retain the ability to accurately attribute events to unique individuals as third-party identifiers become decreasingly viable. GA4 employs a number of privacy-preserving features, including the cessation of IP address storage, options to limit data retention, and the ability to disable Google Signals, location, and device data. Deeply integrated with GMP activation platforms, GA4 offers predictive modeling capabilities both for reporting and audience creation in order to help marketers drive business outcomes against key measures like lifetime value, purchase propensity, and churn probability.
Modeled Conversions accounts for gaps in conversion measurement caused by deprecation of cross-site identifiers by browsers and device operating systems by using ad interaction and other bid stream metadata from observable environments to probabilistically predict the likelihood of conversion in unobservable environments. Used in conjunction with data-driven and custom attribution models in conversion reporting, marketers can tailor their measurement strategy to best reflect the nature of their unique business. As the amount of observable data declines across browsers, devices, and geographies, probabilistic methodologies are needed to generate the trends and informative data that can help marketers move toward sophisticated media measurement and optimization based on predictive techniques.
Enhanced Conversions enables marketers to further supplement measurement by passing hashed first-party data collected during the conversion process alongside the conversion event data. Google will match this data against its own first-party data in order to more accurately link conversions to the ad interactions that led to them. This feature leverages the power of first-party data to deterministically increase measurement fidelity, with careful security processes to preserve privacy.
Offline Conversions enables marketers to incorporate conversions that occur outside of their digital properties (such as sales closed in person or over the phone) into their conversion reporting by passing click and phone call information along with conversion data to Google. Using a similar methodology to Enhanced Conversions, this feature expands the scope of digital optimization to the offline universe and more directly drive real-world business outcomes.
Consent Mode is a feature of Google Tag Manager that adjusts the behavior of Google tags to adhere to individuals’ consent choices for ad and analytics cookies. The feature is integrated with Google Analytics, Google Ads, Floodlight, and Conversion Linker, and can integrate with a website’s Consent Management Platform. Consent Mode automatically adapts the measurement and targeting behavior of Google platforms to respect individual consent choice and augments with probabilistic methods to retain fidelity. For instance, GA4 will supplement data gaps stemming from individuals who decline consent for analytics cookies with behavioral modeling. By analyzing behavioral patterns of individuals who did provide consent, GA4 can estimate the unmeasurable behaviors in order to provide greater insight on key metrics like daily active users and volume of new customer acquisitions.
Customer Match enables marketers seamlessly to reach their existing customers across Google’s owned and operated inventory. Marketers can upload their consented first-party data to Google platforms in hashed form, either directly via the platform user interfaces or by working with approved Customer Match Uploader partners that can facilitate the process or even enrich the data. Google will then match the marketer’s first-party data to their own and generate audiences that can be deployed on campaigns, after which the marketer’s data is deleted from Google’s systems. With Customer Match, marketers can safely, securely, scalably, and deterministically reach their most important audiences: their existing customers.
PAIR (Publisher Advertiser Identity Reconciliation) is a new first-party data audience solution that Google recently released. Like Customer Match, it allows marketers to securely upload their consented first-party data to Google platforms for audience matching. But where Customer Match allows this data to be activated on Google’s proprietary inventory, PAIR matches the marketer first-party data with publishers’ own first-party data and enables these audiences to be activated across those publishers’ properties as well. PAIR enables advertisers to create data linkages with publishers on a one-to-one basis, which further enhances addressability powered by consented first-party data while aligning with Google’s stance against cross-site tracking,
Custom Bidding and Bid-to-Value are features of Google buying platforms that build upon existing automated bidding technologies that historically focused on driving Cost Per Action. These more powerful bidding solutions enable marketers to assign custom values to impressions and conversions that better reflect impact to their businesses, then allow trained machine learning algorithms to drive campaign performance toward more sophisticated outcomes like Return on Ad Spend, Lifetime Value, and higher value conversions. The algorithms leverage much of the machine learning power of Google’s probabilistic measurement solutions, aiming to drive performance despite the signal loss that occurs across increasingly unobservable environments. Google is also focused on enhancing the deployability of these solutions, with several options including through easily accessible user interfaces, advanced options involving custom python scripts, and even advanced integrations with sophisticated third-party algorithm specialists.
GMP has long adopted IFA (Identifier for Advertising) Audiences as a privacy-forward solution for reaching specific audience segments across Connected TV (CTV) inventory. Born out of the IAB Techlab’s CTV Working Group, IFA is a device identifier designed for CTV and OTT devices and meant as a privacy-preserving alternative to identifiers based on IP address. Enabling individual opt-out and not reliant on personal data, GMP was an early mover in adopting IFA across its own platforms and has also worked with major audience platforms across the ecosystem to proliferate support of it across third-party offerings.
GMP Innovations in Action
As a longstanding GMP partner, Kinesso and Matterkind have successfully deployed a number of these tools, demonstrating their potential to simultaneously preserve privacy and drive business results.
One of our longstanding clients is a gaming company with robust first-party data, with a primary KPI of driving online sales. They partnered with us to develop a comprehensive activation and measurement strategy for a high-priority, large budget new game launch.
Leveraging Customer Match to upload their CRM data to DV360, we not only unlocked their data for activation and suppression across display and video inventory, but also enabled advanced lookalikes using their CRM as high-quality seed audiences alongside tight controls on modeled incremental reach. Next, we deployed Custom Bidding across remarketing tactics, developing a custom python script that optimized toward “Buy Now” actions on their website.
This client’s customer base highly indexes toward devices where conversions are difficult to observe. To enable robust insights and optimizations across channels, we implemented a data-driven attribution model in Campaign Manager 360 powered by Modeled Conversions. To measure success, we leveraged Ads Data Hub for mid and post campaign reporting, using advanced custom queries to closely manage conversion rates as well as unique reach and frequency across tactics.
With this end-to-end strategy, the game launch generated very strong results. Remarketing on this campaign, traditionally their strongest performer for Buy Now Rate at a 0.12% benchmark, performed at an unprecedented 0.19% with the implementation of Custom Bidding. Customer Match audiences, implemented with a carefully designed suppression strategy, generated a 0.11% Buy Now Rate, nearly equaling the remarketing benchmark.
Significantly, these tactics were able to be deployed at scale while maintaining performance. Remarketing delivered a robust 23% of the campaign’s budget at this conversion rate. But because Customer Match was able to generate a 65% match rate against the client’s CRM data, Customer Match audiences were able deliver a whopping 77% of the campaign’s total budget at these performance levels.
To build on these successes, we are collaborating with the client to further leverage their strong first-party data assets to implement Enhanced Conversions for future software and hardware campaigns. We expect the increased fidelity to drive scale and performance to even higher thresholds and unlock greater test-and-learn opportunities.
“Google’s ‘privacy by design’ suite of products allows us to deliver the right message to the right audiences, while maintaining our digital responsibility values,” said Mike Miller, SVP of Product Partnerships and Activation at Matterkind. “Having a suite of applications that includes analytics and optimization, that seamlessly integrates into our business intelligence and data solutions, truly drives the results needed for multiple brands. It’s a great example of the value generated by our continued partnership with Google.”
The Privacy Sandbox
Google’s philosophy of preserving privacy while simultaneously driving performance can be seen in the innovations that they’ve made across GMP over recent years. However, the approaching 2024 deprecation of third-party cookies in Chrome represents arguably the most significant technological shift toward privacy-preservation ever undertaken by an ecosystem actor, and the biggest test yet of Google’s ability to deploy private-yet-performant innovations in GMP.
In the Privacy Sandbox, Google proposes to reimagine the measurement and targeting technologies that are foundational to digital marketing. Designed to replace legacy approaches that will become obsolete with the sunsetting of third-party cookies, the proposals consist of a series of APIs that serve remarketing, prospecting, measurement, and other use cases in a privacy-preserving manner. The design of the APIs embody the key themes seen in their previous privacy-preserving features but build upon them with additional themes: isolated processes distributed across individuals’ browsers, reduced reporting granularity, data sharing limited to only what each party needs to complete their part of the process, selective invitation of parties allowed to participate in individual processes, introduction of randomness and “noise”, and a strict adherence to user choice. Although Privacy Sandbox consists of a series of proposals that are aimed to address the entire digital marketing life cycle, three proposals have received particular attention from the industry.
FLEDGE is a Privacy Sandbox proposal that satisfies cross-site remarketing use cases without the use of third-party trackers. Core to this proposal is the concept of ad auctions being run by publishers on the individual’s browser, bid on by marketers according to interest groups that are also stored on the individual’s browser. Under the FLEDGE framework, marketers create interest groups relating to their products and services and assign these interest groups to their site visitors’ browsers. When those individuals visit a publisher site with ads, the publisher administers auctions where invited marketers can submit bids based on the interest groups that the browser is a member of.
Crucially, these auctions are distributed across people’s individual browsers, and are isolated to prevent interaction with entities outside of the bidding process. And although interest groups are created by marketers, they are unreadable by either marketer or publisher. This browser-hosted approach is designed to enable the informed auction process that is foundational to real-time bidding but eliminate the legacy need to share data with ad tech platforms who traditionally run auctions, prevents reidentification, and enables individuals to opt out of these processes altogether.
The Topics API is a newly introduced proposal in the Privacy Sandbox which enables prospecting (or Interest-Based Advertising) use cases in a privacy-preserving manner. In this proposal, the API assigns topics to individual browsers based on the sites they visit. The individual’s topics are assigned from a curated and trusted taxonomy of possible topics and are drawn from a limited period of browsing history (currently proposed at one week) then refreshed thereafter.
When users visit a publisher site, a browser-distributed auction process is administered in a similar fashion to FLEDGE. But the crucial difference with Topics is that marketers are only allowed to bid on topics where their code was present at the point of assignment. To preserve privacy further, individuals are only assigned one topic in each predefined period of browsing history, the assigned list of topics is refreshed every period, and topics can be randomly assigned 5% of the time.
In order to reduce the efficacy of invasive reidentification techniques like fingerprinting, the API is designed with limits in the number of available topics in the taxonomy, limits in the number of assigned and returned topics for each individual, limits in the number of parties that an individual’s data is shared with, and to incorporate random signals. And because these processes are hosted on the browser, individuals will be able to remove individual topics, clear all their topics, or opt out of this API altogether.
The Attribution Reporting API is built to support conversion measurement without the use of cross-site identifiers. Leveraging the same principle of distributed processes on individual browsers, this API enables marketers to serve traditional use cases of media spend optimization and insight generation.
When an individual visits a publisher site, the publisher can store ad interactions onto their browser. Later, when the individual converts on a marketer site, the marketer can similarly store conversion data onto their browser. Using a minimal amount of data generated from the ad and conversion processes, the browser can connect the conversion back to the ad interactions that led to it. Data is further obfuscated with a combination of data delays and random “noised” data being sent.Conversion information is provided in two types of reports: Event-level and Summary, each of which serve distinct use cases and have their own privacy preserving features. Event-level reports have unique IDs that can be used to optimize media, but significantly limit the amount of data exposed to eliminate cross-site tracking. Summary reports can be used for rich insights because they expose a greater variety of data points(such as spend and geography) but preserve privacy by only providing conversion data in aggregates and not exposing unique IDs.
Google, in partnership with its partners and customers, has found success in shaping GMP into a private-yet-performant product suite, retooling its platforms with enhancements that simultaneously preserve privacy but also drive performance.
But the deprecation of third-party cookies in Chrome and the accompanying Privacy Sandbox represent more than a retooling. The proposals represent a radical shift, and changes of this magnitude are inevitably accompanied by friction from impacted stakeholders. As such, Google is in active testing in collaboration with the entire digital marketing ecosystem, having already significantly altered or deprecated several proposals based on industry feedback since the Privacy Sandbox was announced in 2020. While there is much more development to come as we approach 2024, Kinesso and Matterkind are excited to work with Google and our ecosystem partners to explore the potential of this and other privacy-preserving innovations in the future.