Introduction to Differential Privacy

Privacy has become an increasingly hot topic in ad tech. From GDPR to ITP 2.0, marketers are becoming increasingly conscious of the importance of privacy, which they now have to actively balance against the need for transparency and accountability. Recently, industry leaders have started talking about differential privacy, and how this technology could be the solution to balance privacy with security. Digiday provides a good introduction here.

Before diving into differential privacy, it’s helpful to keep in mind how marketers actually consume data. It may seem counter intuitive, but a savvy data-driven marketer doesn’t actually care about any specific individual in their campaigns. Rather, the marketer is optimizing for the behavior (and results) from the entire group or segment it is targeting. (If you are a marketer, ask yourself this question: in your last analysis, did you care that User #123 converted or did you care how many users in your target population spent money?) This insight helps us realize a system that hides the behavior of any given individual but provides accurate user behavior can strike the balance between user privacy and transparency. Does this solution exist? It can with differential privacy.

Differential privacy is a set of statistical techniques that introduce noise into any given data set in order to protect user anonymity without changing your overall conclusion. Does it sound too good to be true?

Here’s an oversimplified example of differential privacy principles at work. If you wanted to ask a group of people sensitive questions such as “Have you cheated on your spouse?” you will likely get few people who to tell you the true answer. However, imagine before people answered, they were told to privately flip a coin. If the coin lands heads, they tell the truth – yes or no. If the coin lands tails, they then flip the coin again privately. If it is heads, they say “yes” no matter what the truth is. If it is tails, they say “no,” again despite the truth.

As a result of this basic obfuscation, any outsider who looks at the data won’t know if an individual participant’s recorded answer is the truth or not because it could easily have been an arbitrary answer. That said, there is a known statistical distribution of correct answers (50% of answers) versus arbitrary answers (25% no, 25% yes) thanks to random coin flips. For a large population sampled, you will be able to then reveal what is the true rate of spousal cheating without risking any individual’s privacy!

This example, of course, oversimplifies the actual mechanics of differential privacy. In reality, more complex techniques can be applied to each data set in order for more robust data security and greater transparency. But that discussion is better left for the Data Privacy 201 course…

Continue Reading

Truth in Measurement: Evolution of Digital Measurement

Brian Andersen of Luma Partners recently spoke at the Truth in Measurement summit, where leading brands and publishers gathered to discuss adopting a common approach for measurement that balances transparency, privacy, and consumer data protection. His presentation on the evolution of digital measurement touches upon the historical and current ways of measurement as background for understanding how things came to be, and what marketers want today. The full presentation is included below, but here are some highlights from his talk:

The Highlights:

  • Measurement started out focused purely on desktop website traffic, with metrics such as page views, click path, exit rates, etc.
  • The industry became increasingly complicated with the rise of mobile, programmatic, and walled gardens
  • Mobile became particularly complicated because 90%+ of time spent was in apps rather than on the mobile web. This led to the need for specific mobile analytics and measurement companies
  • The emergence of programmatic advertising led to more complicated processes, which created opportunities for bad actors to exploit
  • At the same time, walled gardens have become more ubiquitous. Unfortunately, each take a slightly different approach towards measurement
  • People-based measurement has emerged as the solution embraced by marketers, with a focus on real results (such as revenue) rather than proxy metrics (such as impressions, cookies)
  • The biggest challenge facing marketers today is apply these principles across platforms to get log-level data, which is exactly what Truth in Measurement is trying to tackle

Continue Reading

People-Based Measurement from TV to Digital

As more and more people consume media across multiple screens, device-based measurement has become increasingly inaccurate and incomplete. Thunder and TiVo recently partnered up to discuss some of the challenges with people-based measurement from TV to digital, and ways marketers are tackling this tricky problem.

Challenges with Measuring:

Marketers are scrutinizing their approach towards measurement to make sure they are truly understanding what goes into their media spend, and how this spend translates into results. Some of the areas they’re focusing on include:

  • Quantifying sales and brand impact
  • Increasing marketing ROI
  • Measuring omnichannel campaigns
  • Integrating in-store transactions with digital media data
  • Linking cross device data
  • Improving media attribution and optimize media mix

Applying traditional cookie- or device-based measurement approaches to these areas leads to imprecise, incomplete, and sometimes incorrect insights. As such, marketers have embraced People-Based Measurement as the new way forward.

What is People-Based Measurement

People-based measurement refers to the use of persistent identifiers to capture user behavior across channels and devices. This approach provides a more holistic view of user behavior compared with traditional cookie- or device-specific measurement. Here’s a simple video that explains the differences in approach.

There are two common ways people-based measurement is done: panel-based measurement and direct measurement.

Panel based measurement refers to the use of certain technologies that monitor how certain subgroups of individuals behave, and uses those observations to make general conclusions about the population. The advantage of this method is that you can make conclusions with far fewer data points. The downside, of course, is that your extrapolations are prone to sample bias and may be inadvertently distort reality.

Direct measurement, in contrast, provides far higher accuracy. Unfortunately, this approach requires collecting more more data via a persistent identity, which is a technological challenge for marketers who do not have access to this type of technology.

Both approaches of doing people-based measurement provide far greater accuracy relative to cookie- and device-based approaches. In a cookie-based world, ID’s are temporary rather than persistent, and impressions are subject to fraud, deletion, and blocking. In the device-based approach, each device may offer a persistent, unique identity, but one individual may have multiple devices.

Omnichannel with People-based Measurement

People-based measurement connects ad exposures across all environments – from open web to walled gardens, including linear and OTT video. Ad exposures can be connected to a persistent identifier, which can then be tracked against both online and online conversions.  For example, TiVo’s data set includes exposures from three million active households across 210 DMA’s. Using TiVo and Thunder’s people-based measurement, the marketer can combine the data from television with data from open web and walled gardens to provide a true view of the customer journey.

What is the Impact of Measuring by Person

When marketers evolve from device- or cookie-based measurement to persistent people-based measurement, they typically notice some startling changes in their reach and frequency. These observations include:

  • A decrease in reach (which happens when you connect multiple devices and cookies to a single person)
  • An increase an frequency (which happens because your audience may see the ads on multiple screens)
  • An increase in conversions that are ad attributed (which happens via events that are not tracked by cookies or devices such as offline purchases).

Looking to make the transition to People-based measurement?

There are two ways for marketers to embrace people-based measurement.

The quick and easy approach is a Wrap & Measure test, which uses a people-based ad server to track your ad exposures, person counts, and digital conversions for a particular campaign, and provides a report on a particular campaign to see the people-based difference.

The more comprehensive approach is the “Always-On” People-based ad serving, which uses a people-based ad server to track and personalize your message across all campaigns by person. Marketers using a people-based ad server switch from “per campaign” or “per channel” mindsets to “customer-centric” mindsets that focus more on customer journeys and lifetime customer value. Relative to the quick and easy approach, this more comprehensive approach can fill your data lake in real-time with people-based data.

To get the full scoop of the webinar, see the video below:

Continue Reading

People-based Measurement From TV to Digital

As people have begun using multiple screens, device-based measurement has become more inaccurate and incomplete. Thunder and TiVo are co-hosting a webinar to share insights on how brands are building people-based measurement stacks from the ground up to measure everything from TV to digital.

RSVP here to learn how new research from Tivo and Thunder, a people-based ad server, is uncovering the impact of adding TV and people-based to measurement methodologies and technology stacks.

Continue Reading

Neustar and Thunder join forces to deliver better customer experiences, powered by people-based intelligence

SAN FRANCISCO, Aug. 28, 2018 (GLOBE NEWSWIRE) — Thunder Experience Cloud, the leader in people-based ad serving, and Neustar Marketing Solutions (a division of Neustar, Inc.), the leading unified marketing intelligence platform for marketers, today announced the integration of Thunder’s people-based ad server with the Neustar Identity Data Management Platform (IDMP) and the Neustar MarketShare solution. The partnership will enable brands and agencies to quickly customize ad creatives to each customer, as well as measure performance for real-time optimization.

Thunder’s dynamic creative optimization (DCO) solution is a people-based, dynamic ad server that enables advertisers to factor in data signals such as CRM, weather, device type, time, media exposure, and now, audience data from large Data Management Platforms (DMP) like Neustar.

Customers of Neustar and Thunder will be able to target creative messaging for individual, real people and audience segments across digital channels such as display, video and mobile. By synchronizing people IDs on the open web, they can achieve a higher level of personalization, consistency and accuracy, eliminating irrelevant or redundant advertising.

In addition, Thunder’s people-based Experience Measurement solution tracks the performance of ads from exposure to viewer to conversion to allow for a high level of optimization. From there, joint customers can quickly and easily activate media by person tracked on the open web through the Neustar IDMP. This people-based data set will also be integrated within the Neustar IDMP and the Neustar MarketShare solution.

“Advertisers must be able to have a clear view of how their marketing performs across channels – which creatives and messages are being shown to whom, when and where. Neustar is dedicated to giving the industry access to independent and accurate media exposure data, ensuring brands and agencies have the tools they need for personalized, measurable experiences at scale,” said Steve Silvers, General Manager, IDMP, Neustar.

“There is no excuse for a bad ad,” added Victor Wong, CEO of Thunder. “This integration is another step toward ensuring every ad meets the highest standard of relevancy, frequency and impact, ultimately creating a better customer experience.”

About Thunder:
Thunder solves bad ads. Thunder Experience Cloud enables enterprises to produce, personalize, and track their ads cross-channel to achieve the right consistency, relevancy and frequency. Consumers maintain privacy, publishers safeguard data, and brands gain transparency through Thunder for a better ad experience for all.  To learn more visit: https://www.makethunder.com/

About Neustar Marketing Solutions
Neustar, Inc. helps companies grow and guard their business in a connected world. Neustar Marketing Solutions provides the world’s largest brands with the marketing intelligence needed to drive more profitable programs and to create truly connected customer experiences. Through a portfolio of solutions underpinned by the Neustar OneID® system of trusted identity and through a privacy by design approach, we enhance brands’ CRM and digital audiences, enable advanced segmentation and modeling, and provide measurement and analytics all tied to a persistent identity key. Neustar’s position as a neutral information services provider, and as a partner to Google, Facebook and Amazon, provides marketers access to the most comprehensive customer intelligence and marketing analytics in the industry. More information is available at www.marketing.neustar.

Continue Reading

How does Google’s Ads Data Hub Affect My Analytics? (Part III of the Ads Data Hub Series)

Note: We provided an overview of Ads Data hub in Part 1, and how Ads Data Hub will impact DMP’s in Part 2. This post covers data lakes and how analytics will be impacted in the Ads Data Hub world.

Many large brands today have set up “data lakes” where all their data gets stored and made available to other applications for processing and analysis. These data lakes combined with business intelligence tools such as Tableau have created powerful analytics environments where brands can answer questions such as:

  • What customer segment is most responding to my ads?
  • Which ads are leading to the most amount of lifetime customer value?
  • Do people who see my ads spend more with me?
  • Am I spending more money to reach my customers than they are spending with me?

Brands have staffed up data analysts and data scientists to make sense of all this data and answer these important business questions to improve strategy and validate what partners are telling the brand.

Data lakes ultimately rely on data to flow into them. Google’s recent changes with Ads Data Hub keeps data locked within Google Cloud and cannot be combined outside of Google’s controlled environment. As a result, data lakes for marketing are under threat by recent changes by Google.

Data Lakes without Data

Consequently brands with sensitive customer data are forced to decide whether to upload that data to Google to run in a Google-controlled data lake or keep it off the Google Cloud where they’ll need to find other vendors to solve their needs for tracking, analyzing, and modeling.

If you want to maintain control of your own data lake and preview it from drying up, talk to Thunder about our Experience Measurement solution. 

More on the Ads Data Hub series

Continue Reading

How does Google’s Ads Data Hub Affect My Data Management Platform (DMP)? (Part II of the Ads Data Hub series)

Note: We provided an overview of Ads Data hub in Part 1. In this post, we look at how Ads Data Hub will impact DMP’s in general.

Data management platforms (DMPs) power the marketer’s ability to track, segment, and target audiences across programmatic media. Leading DMP solutions include Salesforce DMP (previously known as Krux), Neustar IDMP, Oracle BlueKai and Adobe Audience Manager

If you weren’t paying close attention, you may not realize that the changes Google have announced have blown a hole in your DMP.

 

Two major capabilities are affected by the pending DoubleClick ID removal from logs and push toward using Google’s Ads Data Hub: (1) segmentation and (2) frequency capping.

First, marketers currently use DMPs to create new audience segments based on media exposure. A DMP can keep track of media exposure if its own tags/pixels can run with the ad, but on many publisher inventory such as Google’s Ad Exchange, DMPs are banned from running their code. These publishers are worried about data leakage, which happens when the DMP pixels proprietary audiences on media (such as sports lovers on ESPN.com) and purchased these users elsewhere without paying the publisher.

Historically, the DMP could still get a record of media exposure from the ad server such as DoubleClick, which would share data on who saw the ads running. Using DoubeClick’s data, the marketer could then still segment audiences within the DMP based on who saw the ad, who converted, etc.

Now that Google has discontinued the sharing of logs with IDs, DMPs are no longer able to see media exposure on either inventories on which they are explicitly banned and or inventories where they are allowed to operate but that Google’s DoubleClick ad server is used by the advertiser. If DMPs are to continue to be useful to the marketer, they will need a new source of data.

Second, some marketers use DMPs to create frequency caps across media platforms. By getting their pixel/code to run with an ad, or by ingesting ad serving logs, DMPs can count impressions exposed to a particular user ID and then send a signal to platforms like DSPs to stop buying a user after a certain amount of exposure. However, without log level data, DMPs will not be able to count frequency for inventory in which they are banned, leading to less accurate frequency measurement and therefore less precise frequency capping.

How do I keep my DMP running at full performance?

Marketers who have invested in a DMP and want to keep its capabilities at full power would be advised to either buy more digital media that allow DMP tracking or find an alternative ad tracking or serving solution that can data transfer log files to the DMP. A combination of these two strategies would allow a brand to continue using its DMP to its fullest by giving the DMP the complete picture of ad exposure tied to person.

If you want to add an independent ad tracker to your DoubleClick stack or to keep powering your DMP with data, talk to Thunder about our Experience Measurement solution. 

More on the Ads Data Hub series

Continue Reading

What is Google’s Ads Data Hub and is it right for me? (Part I of the Ads Data Hub series)

What happened to DoubleClick?

Most marketers today use DoubleClick Campaign Manager (DCM) as their primary ad server for delivering ads and tracking ad exposure to conversion. The largest advertiser and most sophisticated advertisers relied on DCM data to do analytics, attribution, and media activation.

These advertisers would “data transfer” log-level data (the raw data for each impression rather than the aggregate data that hides user-level and impression-level information) to their data management platform, data lakes, and vendors that do analytics or attribution modeling.

 

In April, Google announced it will no longer allow data transfer of DoubleClick log-level data with IDs. This decision effectively destroyed most of the value of the log-level data exported from DCM because advertisers wouldn’t know who saw the ads but only how often an ad in total was served. DoubleClick could be used only to verify that the total amount of impressions bought were actually delivered but all the other powerful use cases like analytics, attribution, and data management would no longer be possible with DoubleClick data.

In June, Google announced it was sunsetting DoubleClick as a brand and folding everything under Google’s brand.

R.I.P. DoubleClick.

Enter Google Ads Data Hub

At the same time, Google pushed forward its own solution to this new problem for marketers — Ads Data Hub. This product is essentially a data warehouse where ad exposure data is housed and can be connected to Google’s own solutions for attribution, analytics, and data management.

One new benefit is access to the Google ID, which is a powerful cross-device ID that uses data from users logging into Google services like Android, Maps, YouTube, etc. Previously, DoubleClick was only tracking and sharing a cookie-based DoubleClick ID, which neither connected cross-device ad exposure and conversion nor reconciled multiple IDs to the same person. For many advertisers doing log-level data analysis and activation, this new ID is a big upgrade because it provides more accurate measurement.

One major downside is that this data cannot leave Ads Data Hub. Consequently, you cannot do independent verification of Google’s attribution or analytics modeling. If Google says Google performs better than its competitors, you will have to trust Google at its word. In the past, you would at least have the raw data to apply your own attribution model if you so wanted, or to re-run Google’s calculations to verify its accuracy (since big companies are not infallible).

By extension, outside ad tech providers (such as DMPs, MTA, etc.) who may be best in-class will have a much harder time working with Google solutions. As a result, you will be dependent on Google.

To do matching of off-Google data such as other ad exposure or conversions that happen offline, Ads Data Hub now requires you to upload and store your customer data in the Google Cloud. In that environment, it can be matched with Google’s ID and tracking so you can build a Google-powered point of view of the consumer journey.

In a way, Ads Data Hub is for those who trust but don’t need to verify. It is a good solution for advertisers who today spend the vast majority (75%+) of their ad budget with Google because ultimately if their advertising isn’t working, no matter what Google says about how it is performing, it would be ultimately accountable for the results. You wouldn’t need to verify calculations to know if your ad budget is wasted.

What else can I do?

Another solution is to add independent ad serving and/or tracking in addition to or in replacement of Google. By doing so, you can still generate log-level data for Google-sold media but it will not be tied to a Google ID. Instead, you will be using your own ID or a vendor’s cross-device ID to understand who saw what ad when, where, and how often.

This approach is best suited for large advertisers who want best in class ad tech solutions to work together, and who cannot spend all their money on a single media platform to achieve their desired results. Typically brands large enough to afford data lakes, independent attribution providers, and data management platforms are the ones who will have the most to lose by moving to Ads Data Hub.

If you already realize you want to take a trust, but verify approach in your ads, talk to Thunder about our Experience Measurement solution. 

More on the Ads Data Hub series

Continue Reading

What CMO’s Say About Ad Experiences

Marc Pritchard famously said “It is time for marketers and tech companies to solve the problem of annoying ads and make the ad experience better for consumers.”

What do his peers think? The CMO Club has partnered with Thunder to publish a “Guide to Solving Bad Ad Experiences,” which includes a survey of over 80 CMOs and an interview with the CMO of Farmers Insurance on the impact of bad ads and how people-based marketing can fix them.

Some key findings include:
  • 74% of CMOs consider brand loyalty as most negatively affected by bad ads
  • 55%+ of CMOs consider frequency and relevancy as the top factors in bad ads
  • 78% of CMOs consider it “inexcusable” to serve ads for products the customer already bought from them
  • 71% of CMOs consider frequency capping important for ad experience but 60% aren’t confident in even their frequency counting!

Click here to download the full research report.

Continue Reading

What is a CMP?

CMP is the hot adtech acronym of 2018. There are actually two meanings to this term: (1) Creative Management Platform and (2) Consent Management Platform. Here’s an overview of both these products and why you may need one.

Creative Management Platform

Introduced in 2016 by Thunder, the CMP acronym original stood for “creative management platform,” a tool for producing and trafficking ad creatives. Rather than just a general purpose creative editor like Adobe Photoshop or Animate, which are applications built for a single designer to use by him or herself, CMPs are meant for an enterprise that has a scale issue with creative.

Many brands, agencies and publishers are increasingly needing to build ads in different sizes and versions for different audiences and media formats. Consequently, creative production demands have grown exponentially while most creative organizations can only scale linearly in their capability by adding more designers and programmers. Because traditional creative editors were built for highly advanced users, a creative bottleneck formed as demand went up and not enough talent or payroll existed to fill the void.

Creative Management Platforms radically simplified ad production by providing easier interfaces and automated production tasks like re-sizing. Forrester began recognizing CMPs in 2017 as part of their broader creative ad tech research which has been timed with the rise in enterprise demand for new marketing creative technologies.

Consent Management Platform

Introduced in 2018, the new CMP acronym stands for “consent management platform.” The European privacy laws known as GDPR required publishers and marketers to obtain explicit consent for certain tracking and targeting data. As a result, a new category of tools emerged to specifically help these enterprises collect and keep track of user consent.

The CMP then feeds that consent information tied to an ID to other selected partners in the digital advertising supply chain. As a result, every party in a publisher’s supply chain understands what data they may use and for what.

Which CMP do I need?

It depends if you’re looking to solve a creative problem or a data privacy problem. Talk to Thunder if you need help with your data-driven creative problems or digital creative production problems. Check out these consent management vendors if you’re looking to solve a privacy preference problem.

Continue Reading