Notes on Advertising

26 minutes read Jun 1, 2022 Sep 3, 2020

Ad Tech

2022

  • BleepingComputer - Vodafone Plans Carrier-Level User Tracking for Targeted Ads
    • TrustPID enables mobile phone carriers like Vodafone to share random subscriber unique IDs based on “network identifiers provided by network operators” with participating advertising providers.
    • Opt-in consent is obtained through traditional website cookie banners.
    • “The mobile carrier plans to assign a fixed ID to each customer and associate all user activity with it. The ID will be based on a number of parameters, so that the system will be able to maintain persistence. Then, the mobile ISP creates a personal profile based on that ID and helps advertisers serve targeted ads to each customer without disclosing any identification details.”
  • AdExchanger - SXM Media Intros an Audio Identifier (Because Cookieless)
    • AudioID seeks to be a widely used advertising identifier for audio content “across Pandora, SiriusXM and Stitcher” and “the wider AdsWizz network of audio and digital publishers” to enable targeted programmatic advertising.
    • “AudioID uses an algorithm to match and encrypt different consented listener signals, including mobile ad IDs and email addresses.”
    • “We want to offer whoever wants to integrate on the platform as a partner the capability [to do so].”
    • These identified users can be used to create look alike audiences for mediums which are traditionally identity poor such as podcasts.
      • “zip code, market, country/ region, metro device type and OS, language, and much more”, “Point of Interest targeting — coffee shops, stadiums, gas stations, and so on”, “Increased addressability – target by age, hundreds of behavioral segments” can be used for targeting in podcasts. [Source]
      • “Our unique AI Transcription Technology goes beyond traditional targeting parameters to dive deeper into what makes podcasts so strong – their content… which allow[s] advertisers to target or fend off specific content, ensuring brand safety at a granular level and a more contextual understanding of the content by unveiling unique patterns and eventually mood and sentiment”. [Source]

2021

2020

2019

  • Adelphic - Uniquely Identifying a Network-Connected Entity [Patent]
    • A software technology to identify users across devices by looking at device data and usage behavior. Feature data is collected from user devices. Rules are run against the feature data so that the same visitor can be identified across devices and linked to a single id. Rules consist of expert rules which are manually created, and probabilistic rules which are statistically generated (ex: machine learning).
    • Feature data may consist of…
      • “The deterministic data is data (e.g., a cookie, a device identifier, an MSISDN, a phone number, an email address, a user identifier such as an OpenId, or the like) assigned to a user by a provider. The device and system-specific feature data may be quite varied and typically is one of: operating system (OS) type, OS version, system clock value, execution speed, time of last installation, time of last boot, system default language, system local time, time zone offset, orientation, a display metric, a globally unique identifier (GUID), a model, a brand name, and a build version. The usage feature data also may be quite varied and typically is one of: HTTP headers, usage agent information, user query data, application launch time, application exit time, network type, page load time, page unload time, day and time of access, day of week, time of day, page referrer, plug-in data, geo-data, location data, URL view data, typing frequency data, and gesture data”
      • “typically is a cookie, deviceID (e.g., UDID, ODIN, Open UDID, Android_ID, or the like), email address, telephone number or the like.”
      • “Another encoding type is range, such as day of the week, hours of the day, and call duration.”
    • “For mobile applications, typically a device identifier, such as a UDID in iOS™ devices, and the Android_ID in Android™ OS devices, is used to identify a device. A major advantage of identifiers of this type is that they are unique to a device, and all the applications to which the device has access theoretically share the same identifier. This same advantage, however, creates privacy and security concerns, and these concerns have discouraged the use of such identifiers for identity purposes. Indeed, in some cases these device identifiers are being made inaccessible to developers. Other unique device identifiers, such as MAC address, IMEI, and ESN, are being proposed as substitutes. These alternatives, suffer from some of the same concerns as those that exist with the UDID, but they are also are difficult to use because there are no clearly-defined ways to access them. Another approach is to the problem uses application-generated identification. These approaches, which are available through initiatives such as openUDID, SecureUDID, and ODIN, overcome many of the deficiencies of the device-specific approach. Nevertheless, because these identifiers are assigned by applications, they are accessible only to the applications that have permission to access them.”
    • “Another common limitation of both the device-specific and application-generated approaches is from the perspective of usage monitoring in that they do not address or solve the problem of identification of an entity that uses different devices (such as a mobile phone and a tablet) to access the same content. HTTP cookies have long been a de facto standard for identifying unique users on the Web. However, cookie-based techniques suffer from several problems in that they cannot be used easily in non-browser-based applications, which now account for the vast majority of mobile applications. Further, cookies are device and browser-specific; two browsers (or distinct browser instances) on the same device receive two different cookies from the same domain. Conversely, different users of the same device and browser receive the same cookie when they visit a particular domain. Further, cookie support in mobile device-based browsers is not stable and, in general, the approach does not work well with mobile applications. Even when cookie support works, different devices are identified by different cookies. Moreover, because cookies are stored on the client, they are subject to blockage and deletion. Further, because of privacy concerns, browsers often implement functions that it makes it increasingly easy to disable and delete them.”
    • “For the mobile web, many techniques besides cookies are being tried for device identification. These include, without limitation, packet sniffing, web beacons, HTTP header information, IP addresses, and the like. Those variables alone often are not enough to serve as persistent device identifiers. These approaches may be combined with fingerprint-based device identification methods for fixed network devices, although the results have variable reliability. Browser fingerprints, for example, which work well with the fixed Web, become unreliable when applied to a mobile Web browser because little user and device-level customization are available to serve as distinguishing features. In addition, such information pertains only to the device, as opposed to the user of the device.”
    • Applications of this platform may include “an advertising (ad) engine, a product recommendation engine, a fraud detection engine, a conversion tracking engine, a database program, and the like.”
  • AdExchanger - The Companies Challenging LiveRamp’s Supremacy In Data Onboarding [Article]
    • “LiveRamp pioneered data onboarding, the technology to match anonymized online user cookies and mobile IDs to real-world consumers and offline data.”
    • “Neustar benefits from the legacy offline data set, which has personally identifiable information like names and home addresses that it can match to emails, cookies and mobile IDs, said Michael Schoen, SVP and GM of marketing solutions.”
    • “Throtle, for instance, is focused on adding healthcare brands, which Chachko said are more comfortable using first-party data now and value accuracy over scale for identity products. There are also healthcare data risks that discourage some tech companies from entering the vertical. For Throtle, that’s part of the attraction.”
    • “TransUnion has roots in consumer tracking for individuals to monitor their credit scores and for businesses that issue credit to appraise applicants, like insurance companies, mortgage financiers or real estate landlords. That archive of offline consumer activity is the foundation for TransUnion’s data onboarding service.”
    • ”[ZeoTap] has exclusive data licensing deals with 10 telcos and 70 other ecommerce or publishing partners. And… telcos own the “gateway” where browser cookies or advertising IDs can be deterministically matched to longer-lasting profile data like emails or phone numbers.”

2014

  • ArsTechnica - Ads use Ultrasound to link Phone, TV, tablet, and PC
    • SilverPush uses audio beacons embedded in ads to track users.
    • “The audio beacon enables companies like SilverPush to know which ads the user saw, how long the user watched the ad before changing the channel, which kind of smart devices the individual uses, along with other information that adds to the profile of each user that is linked across devices.”

2009

  • ACM Queue - What DNS Is Not
    • Missing DNS entries have been overriden by ISPs and Verisign (which operates the .COM TLD) to return ad revenue generating search results instead of a no-entry response (NXDOMAIN).

Advertising Techniques

2020

  • Dynamic post-production product placement is being used in film and television content.
    • See Ryft and TripleLift.

2019

2018

  • Wall Street Journal - Instagram’s Content Factories Are Huge—And That’s a Problem for Facebook [Article]
    • Instagram influencers and creators can command a large reach by growing their followers. Advertisers pay influencers to put out sponsored posts. Instagram’s rules require that sponsored posts are labelled but this is not always followed.
      • “On a typical day, 421’s staff of about 30 workers churn out dozens of posts from more than 100 Instagram accounts with a total reach of around 300 million followers”
      • “Advertisers on these sites pay by the eyeball—an estimated $373 million on influencer marketing in the U.S. and Canada in the first quarter of 2019, according to Instascreener, which tracks marketing on Instagram. About $265 million of that was on Instagram, up 62% from the same period a year earlier.”
      • “Instagram accounts for 22% of Facebook’s total ad revenue, according to estimates by Andy Hargreaves, an equity research analyst with KeyBanc Capital Markets Inc.”
      • “Aspiring influencers hope to match the success of such celebrities as Ms. Kardashian West, who charges several hundred thousand dollars for a single post.”
      • “Up-and-coming influencers can command $100 to $5,000 a post, according to Kamiu Lee, CEO of influencer platform Activate.”
      • “Salvatore DiBenedetto, 20, built a large following by buying and selling Instagram accounts, including one he created, @wastedoninstagram. Mr. DiBenedetto sold accounts to 421 for $120,000.”
    • This might be eroding public trust on the internet
      • “‘The perception that everything on the internet is fake is becoming more and more popular,’ said Sean Spielberg, co-founder of Instascreener, an influencer analytics tool.”
    • Some influencers indirectly pay to get more followers via giveaways
      • “The influencers pool money to buy a prize, often a coveted designer purse. Each influencer instructs their fans to follow the accounts of all the influencers who put up the prize money. That makes them eligible to enter the prize raffle. Ms. Faust’s followers grew to about 250,000 at the end of last year, from 100,000 followers in late 2016.”
  • The Outline - These are the People Paying Journalists to Promote Brands in Articles [Article] and How Brands Secretly Buy their Way into Forbes, Fast Company, and HuffPost Stories [Article]
    • Some smaller marketing agencies have paid authors to put brand mentions in tech news articles and blog posts they have written.
    • Some marketing agencies and clients may have unknowingly paid for these brand mentions by hiring agencies who use this practice.
    • Journalist paid brand placement in articles often goes against the site’s editorial guidelines and the content is removed when it is discovered.

In the News

2021

2020

2019

  • Wall Street Journal - You Give Apps Sensitive Personal Information. Then They Tell Facebook. [Article]
    • “HR Monitor, the most popular heart-rate app on Apple’s iOS, made by California-based Azumio Inc., sent a user’s heart rate to Facebook immediately after it was recorded.”
    • “Flo Health Inc.’s Flo Period & Ovulation Tracker, which claims 25 million active users, told Facebook when a user was having her period or informed the app of an intention to get pregnant, the tests showed.”
    • “Real-estate app Realtor.com, owned by Move Inc., a subsidiary of Wall Street Journal parent News Corp, sent the social network the location and price of listings that a user viewed, noting which ones were marked as favorites, the tests showed.”
    • “BetterMe: Weight Loss Workouts, was in its Android version also sharing users’ weights and heights with Facebook as soon as they were entered.”
    • “Among the top 10 finance apps in Apple’s U.S. app store as of Thursday, none appeared to send sensitive information to Facebook, and only two sent any information at all. But at least six of the top 15 health and fitness apps in that store sent potentially sensitive information immediately after it was collected.”
    • This data is used for analytics accessed by developers and for Facebook to do ad targeting and market research.
    • The data is collected as part of “App Events” in the Facebook SDK common to many apps. These events include a set of common events such as logins and purchases and custom events defined by the developed where this user data is often collected.

2017

  • Wall Street Journal - Vizio Settles FTC Lawsuit For Snooping On TV Viewers [Article]
    • Vizio was accused of “using its TVs to track what its owners watched, then selling that information to marketing firms, all without customers’ knowledge or consent. Vizio agreed to pay $2.2 million to settle the lawsuit filed by the Federal Trade Commission and the New Jersey Attorney General”
    • The TVs tracked what movies, TV, and commercial content was being watches on a per second basis.
    • “the FTC said the data that was sold included information such as a consumer IP address, which companies then used to track and target consumers across different internet-connected devices”

Laws and Policies

2023

2021

2020

2018

2017

2016

2015

2013

Terminology

  • Supply / Sell Side Platform (SSP) - a platform which provides ad placement opportunities across content providers (ex: sites, apps, etc) from multiple customers.
  • Demand Side Platform (DSP) - a platform which provides targeted ad content from multiple ad and data exchanges.
  • Customer Data Platorm (CDP) - a platform which unifies customer data across various different collection points in an organization

  • Ad Network - a company which connects advertisers to content providers (ex: TV, print, online).

  • Programmatic Advertising - software automated purchasing of digital advertising.
    • Real Time Bidding - a form of programmatic advertising where ads are purchased for placement immediately through an auction.
  • Linear Media - media content which is accessed in a fixed order without any interactivity (ex: cable TV)
  • Over the Top (OTT) - streaming media service provided directly to customers over the internet across multiple devices (ex: Netflix, WhatsApp, Skype)