2024 marked a significant year. AI became integral to our workflows, commerce and retail media networks soared, and Google did not deprecate cookies. Amidst these changes, ID bridging emerged as a hot topic, raising questions around identity reliability and transparency, which necessitated industry-wide standards. We believe the latest IAB OpenRTB specifications, produced in conjunction with supply and demand-side partners, set up the advertising industry for more transparent and effective practices.
So, what exactly is ID bridging?
As signals, like third-party cookies, fade, ID bridging emerged as a way for the supply-side to offer addressability to the demand-side. ID bridging is the supply-side practice of connecting the dots between available signals, that were generated in a way that is not the expected default behavior, to understand a user’s identity and communicate it to prospective buyers. It enables the supply-side to extend user identification beyond the scope of one browser or device.
Imagine you visit a popular sports website on your laptop using Chrome. Later, you use the same device to visit the same sports website, but this time, on Safari. By using identity resolution tools, a supply-side partner can infer that both visits are likely from the same user and communicate with them as such.
ID bridging is not inherently a bad thing. However, the practice has sparked debate, as buyers want full transparency into the use of a deterministic identifier versus an inferred one. This complicates measurement and frequency capping for the demand-side. Before OpenRTB 2.6, ID bridging led to misattribution as the demand-side could not attribute ad exposures, which had been served to a bridged ID, to a conversion, which had an ID different from the ad exposure.
OpenRTB 2.6 sets us up for a more transparent future
In 2010, the IAB, along with supply and demand-side partners, formed a consortium known as the Real-Time Bidding Project for companies interested in an open protocol for the automated trading of digital media. The OpenRTB specifications they produced became that protocol, adapting with the evolution of the industry.
The latest evolution, OpenRTB 2.6, sets out standards that strive to ensure transparency in real-time bidding, mandating how the supply-side should use certain fields to more transparently provide data when inferring users’ identities.
What’s new in OpenRTB 2.6?
Here are the technical specifications for the industry to be more transparent when inferring users’ identities:
- Primary ID field: This existing field now can only contain the “buyeruid,” an identifier mutually recognized and agreed upon by both buyer and seller for a given environment. For web environments, the default is a cookie ID, while for app activity, it is a mobile advertising ID (MAID), passed directly from an application downloaded on a device. This approach ensures demand-side partners understand the ID’s source.
- Enhanced identifier (EID) field: The EID field, designated for alternative IDs, now accommodates all other IDs. The EID field now has additional parameters that provide buyers transparency into how the ID was created and sourced, which you can see in the visual below:
Using the above framework, a publisher who wants to send a cross-environment identifier that likely belongs to the same user would declare the ID as “mm=5,” while listing the potential third-party identity resolution partner under the “matcher” field, which the visual below depicts. This additional metadata gives the demand-side the insights they need to evaluate the reliability of each ID.
“These updates to OpenRTB add essential clarity about where user and device IDs come from, helping buyers see exactly how an ID was created and who put it into the bidstream. It’s a big step toward greater transparency and trust in the ecosystem. We’re excited to see companies already adopting these updates and can’t wait to see the industry fully embrace them by 2025.”
Hillary Slattery, Sr. Director, Programmatic, Product Management, IAB Tech Lab
Experian will continue supporting transparency
As authenticated signals decrease due to cookie deprecation and other consumer privacy measures, we will continue to see a rise in inferred identifiers. Experian’s industry-leading Digital Graph has long supported both authenticated and inferred identifiers, providing the ecosystem with connections that are accurate, scalable, and addressable. Experian will continue to support the industry with its identity resolution products and is supportive of the IAB’s efforts to bring transparency to the industry around the usage of identity signals.
Supply and demand-side benefits of adopting the new parameters in OpenRTB 2.6
- Partner collaboration: Clarity between what can be in the Primary ID field versus the EID field provides clear standards and transparency between buyers and sellers.
- Identity resolution: The supply side has an industry-approved way to bring in inferred IDs while the demand side can evaluate these IDs, expanding addressability.
- Reducing risk: With accurate metadata available in the EID field, demand-side partners can evaluate who is doing the match and make informed decisions on whether they want to act on that ID.
Next steps for the supply and demand-sides to consider
For supply-side and demand-side partners looking to utilize OpenRTB 2.6 to its full potential, here are some recommended steps:
For the supply-side:
- Follow IAB Specs and provide feedback: Ensure you understand and are following transparent practices. Ask questions on how to correctly implement the specifications.
- Vet identity partners: Choose partners who deliver the most trusted and accurate identifiers in the market.
- Be proactive: Have conversations with your partners to discuss how you plan to follow the latest specs, which identity partners you work with, and explain how you plan to provide additional signals to help buyers make better decisions.
We are beginning to see SSPs adopt this new protocol, including Sonobi and Yieldmo.
“The OpenRTB 2.6 specifications are a critical step forward in ensuring transparency and trust in programmatic advertising. By aligning with these standards, we empower our partners with the tools needed to navigate a cookieless future and drive measurable results.”
Michael Connolly, CEO, Sonobi
These additions to the OpenRTB protocol further imbue bidding transactions with transparency which will foster greater trust between partners. Moreover, the data now available is not only actionable, but auditable should a problem arise. Buyers can choose, or not, to trust an identifier based on the inserter, the provider and the method used to derive the ID. While debates within the IAB Tech Lab were spirited at times, they ultimately drove a collaborative process that shaped a solution designed to work effectively across the ecosystem.”
Mark McEachran, SVP of Product Management, Yieldmo
For the demand side:
- Evaluation: Use the EID metadata to assess all the IDs in the EID field, looking closely at the identity vendors’ reliability. Select partners who meet high standards of data clarity and accuracy.
- Collaboration: Establish open communication with supply-side partners and tech partners to ensure they follow the best practices in line with OpenRTB 2.6 guidelines and that there’s a shared understanding of the mutually agreed upon identifiers.
- Provide feedback: As OpenRTB 2.6 adoption grows, consistent feedback from demand-side partners will help the IAB refine these standards.
Moving forward with reliable data and data transparency
As the AdTech industry moves toward a cookieless reality, OpenRTB 2.6 signifies a substantial step toward a sustainable, transparent programmatic ecosystem. With proactive adoption by supply- and demand-side partners, the future of programmatic advertising will be driven by trust and transparency.
Experian, our partners, and our clients know the benefits of our Digital Graph and its support of both authenticated and inferred signals. We believe that if the supply-side abides by the OpenRTB 2.6 specifications and the demand-side uses and analyzes this data, the programmatic exchange will operate more fairly and deliver more reach.
Latest posts
Published in AdExchanger. “Data-Driven Thinking" is written by members of the media community and contains fresh ideas on the digital revolution in media. Today’s column is written by Tom Manvydas, vice president of advertising strategy and solutions at Experian Marketing Services. The proliferation of connected electronics has spurred new interest in device-recognition technologies even though they have been in use since the 1990s. As we enter the “Internet of Things” era, device recognition will significantly impact the ad tech ecosystem. Many network advertising technologies are becoming obsolete as cookie blocking grows and the Internet becomes more mobile and device-centric. Device recognition will be yet another technology challenge for marketers but has the potential to overcome many key tracking, measurement and privacy issues with which data-driven marketers have struggled. By leveraging device recognition technologies, marketers can protect their investments in Web 2.0 ad tech, like multitouch attribution, and improve their overall digital marketing programs. Device Recognition Vs. Cookies Device recognition attempts to assign uniqueness to connected devices. By focusing on the device, you are able to “bridge” between browsers and apps, desktop to mobile and across OS platforms like iOS and Android. Device-recognition IDs function like desktop cookies for devices but with four important differences: 1. Coverage: Device-recognition methods are largely immune from cookie limitations. About half of mobile engagements on the Web do not involve cookies, while third-party blocking impacts up to 40% of desktop engagements. 2. Persistency: Device-recognition IDs can be more persistent and less fragmented than most desktop cookies. For example, Apple’s UDID or Android ID are permanent, and network node IDs like MAC addresses are near-permanent. Proxy IDs such as IDFA are persistent but can be updated by the device owner or ID provider. 3. Uniqueness: Devices are unique and cookies are fragmented. The digital media industry incurs substantial overhead cost and loss of efficiency when dealing with fragmented profiles and obsolete data caused by cookie churn. However, device-recognition methods are limited in their ability to recognize multiple profiles on shared devices. 4. Universality: Device-recognition technologies are universal and generally work across devices and networks. However, interoperability issues across device operating systems, such as iOS and Android, can limit the universal concept. There are many types of device-recognition technologies but two basic approaches to device recognition: deterministic and probabilistic, each with their pros and cons. Deterministic Approach: Accurate And Persistent But Complicated Deterministic device recognition primarily uses the collection of various IDs. While the mobile developer is familiar with the variety of IDs, it’s important that marketers become better-versed in this area. Examples include hardware IDs (including serial numbers), software-based device IDs (such as Apple’s UDID or the Android ID), digital data packet postal codes or proxy IDs (such as MAC addresses for WiFi or Bluetooth, IDFA for both iOS and Android and open-source IDs). Deterministic methods improve the accuracy of tracking, targeting and measurement over current cookie-based methods. They can improve the ability to more persistently manage consumer opt-outs. But the proliferation of device types limits the universality of deterministic device recognition. Without uniform standards across platforms, marketers need to account for multiple ID types. Also, deterministic device-recognition methods are not well developed for desktop marketing applications. The lack of interoperability across deterministic device IDs makes execution too complicated. Deterministic device IDs were meant for well-intentioned uses, such as tracking the carrier billing for a device. However, they present privacy and data rights challenges, leading to blocking or limited access by companies that control IDs. Probabilistic Device Recognition: A ‘Goldilocks’ Solution Probabilistic device recognition may be the ideal solution for a connected world that does not rely on cookies nor wants to use overly intrusive deterministic device recognition. Probabilistic device recognition is not a replacement for deterministic IDs. Instead, it complements their function and provides coverage when they are not available. The probabilistic approach is based on a statistical probability of uniqueness for any single device profile. This approach creates a unique profile based on a large number of common parameters, such as screen resolution, device type and operating system. This process can uniquely identify a device profile with 60% to 90% accuracy, compared to 20% to 85% accuracy for cookie-based identification methods. Probabilistic IDs are more persistent than cookies with better coverage, but less persistent than deterministic device IDs. The natural evolution of the device takes place over time and prevents persistent identification. Probabilistic device recognition can be universal and is not impacted by interoperability issues across platforms — the technology used to generate a probabilistic ID on one network can be the same technology on another network. Unlike some deterministic device recognition approaches, there is no device fingerprinting. Probabilistic device recognition accurately identifies profiles in aggregate, rather than a single device. That’s the inherent beauty of probabilistic device recognition: It can generate more accurate targeting results than cookie-based methods without explicitly identifying single devices. This is more than good enough for most marketers and significantly better than what’s available today. Another benefit is the absence of any residue on the device — no cookie files, flash files or hidden markers. Probabilistic methods can work on devices that block third-party cookies or connect to the Web without using any cookies. For example, you might have a hard-to-reach but valuable audience segment. Probabilistic device recognition could effectively increase your reach on this segment by 40% to 50% and increase the overall targeting accuracy by two times. Let’s say the actual population for this segment is 100,000 members. The typical cookie-based approach might reach 28,000 members but the typical probabilistic device-recognition approach could reach 65,000 members. A Decline In Hardware Entropy If you take a close look at the emitted data from today’s devices, it is not easy to analyze it for device identification. That’s because the data footprint of one device looks a lot like another. Device recognition augmentation methods can address this, such as device usage profiles, geo location clustering, cross-device/screen analytics or ID linkage for first-party data owners. In the short term, device-recognition technologies, particularly probabilistic methods, can greatly improve today’s digital marketing programs. Marketers should become fluent in their use cases and benefits. If 2013 was the year of mobile, I think we’ll see a surge in marketing applications based on device-recognition technologies in 2014. Follow Experian Marketing Services (@ExperianMkt) and AdExchanger (@adexchanger) on Twitter.
According to Experian Marketing Services’ 2014 Digital Marketer: Benchmark and Trend Report, social media Websites are playing an increasingly important role in driving traffic to other Websites, including retail sites and even other social networking sites, at the expense of search engines and portal pages. For instance, as of March 2014, social media sites account for 7.72 percent of all traffic to retail Websites, up from 6.59 percent in March 2013. Further, Pinterest, more than Facebook or YouTube, is supplying the greatest percentage of downstream traffic to retail sites. According to the Digital Marketer Report, more retailers are directing their customers to social media within their email campaigns. In fact, 96 percent of marketers now promote social media in their emails, and it shows. In 2013, for instance, email Websites generated 18 percent more clicks to social networking pages than the year prior. Social drives more traffic to other social Websites Social media Websites are driving more and more traffic to other social sites. In 2013, 15.1 percent of clicks to social networking and forum sites came from other social networking sites, up from a 12.5 percent click share reported in 2012. Despite driving the greatest share of traffic to social networking sites with 39.1 percent of clicks, search engines’ share of upstream traffic to social declined a relative 13 percent year-over-year. Among the other top referring industries to social, only the portal front pages industry — which includes sites like Yahoo!, MSN and AOL and is closely affiliated with search engines — showed a drop in upstream click share providing further evidence that increasingly all (or most) roads lead to social. To learn more about key trends in social media traffic, including downstream traffic from social sites and the share of consumers accessing social media across multiple channels, download the free 2014 Digital Marketer: Benchmark and Trend Report.
Mother’s Day may not exactly be right around the corner, but the time to send your Mother’s Day emails sure is! Based on our analysis of 186 brands that sent Mother’s Day mailings in 2013, 75 percent of email volume and 80 percent of email-generated revenue occurred between May 1st and Mother’s Day (May 12, 2013). The highest revenue-producing days were five days before the holiday (Wednesday, May 8, 2013) and Mother’s Day itself. This year, the sentimental holiday falls one day sooner than last year (May 11), but you still have more than enough time consider these quick tips for easy wins while planning and executing your campaigns. Tip 1: Give them what they’re searching for Last year, online searches five weeks before Mother’s Day were dominated by searches for the date of the holiday. As such, we recommend including the date of Mother’s Day in your email subject lines, particularly those early in the season, when customers are searching online for, and opening emails with, that information. Tip 2: Set the tone early with your subject lines A sample of early season subject lines that outperformed the overall unique open rate included: Remember Mom on Mother’s Day, May 12 Get a head start on Mother’s Day (plus a gift for you) Just arrived: Mother’s Day Gift Sets To Mother, With Love Tip 3: When it comes to timing, it’s the thought that counts Think through the timing of your emails depending on order delivery deadlines. On May 8th of last year, the largest revenue producers for email were orders for flowers and gifts placed in time to be delivered by Mother’s Day. Email subject lines on May 8th included reminders of the delivery deadlines: Last Chance: Free Shipping/No Service Charge for Mother’s Day! ENDS TODAY: Enjoy Complimentary Second Day Delivery in Time for Mother’s Day Tip 4: Let them treat themselves On Mother’s Day, the top email revenue generators were “self-gifting” (treat yourself on Mother’s Day only), Mother’s Day online sales and free shipping, as well as e-gift cards: Free Shipping Today Only! Happy Mother’s Day You deserve a treat yourself! HAPPY MOTHER’S DAY! Treat yourself to 30% off today only Last Chance: eGift Cards in Time for Mother’s Day Other email performance highlights: Note: All email performance highlights are based on comparisons to Mother’s Day mailings without the highlighted feature from matched brands. To all those in the midst of Mother’s Day campaign planning, good luck and happy sending!