Anthony J. Pennings, PhD

WRITINGS ON DIGITAL STRATEGIES, ICT ECONOMICS, AND GLOBAL COMMUNICATIONS

Hypertext, Ad Inventory, and the Use of Behavioral Data

Posted on | October 14, 2021 | No Comments

Artificial Intelligence (AI) and the collection of “big data” are quickly transforming from technical and economic challenges to governance and political problems. This post discusses how the World Wide Web (WWW) protocols became the foundation for new advertising techniques based initially on hypertext coding and online auction systems. It also discusses how the digital ad economy became the basis of a new means of economic production based on the wide-scale collection of data and its processing into extrapolation products and recommendation engines that influence and guide user behaviors.

As Shoshana Zuboff points out in her book, Surveillance Capitalism (2019), the economy expands by finding new things to commodify, to make into products or services that can be sold.[1] When the Internet was opened to commercial traffic in the early 1990s and the World Wide Web established the protocols for hypertext and webpages, a dramatic increase in content and ad space became available. New virtual “worlds” of online real estate emerged. These digital media spaces were made profitable by placing digitized ads on them.

Then, search engines emerged that commodified popular search terms for advertising and also began to produce extraordinary amounts of new data to improve internal services and monitor customer behaviors. Data was increasingly turned into prediction products for e-commerce and social entertainment. Much of the data is collected via advertising processes, but also purchasing behaviors and general sentiment analysis based on all the online activity that can be effectively monitored and registered. The result is a systemic expansion of a new system of wealth accumulation that is dramatically changing the conditions of the prevailing political economy.

The Internet’s Killer App

The World Wide Web was the “killer app” of the Internet and became central to the modern economy’s advertising, data collection, e-commerce, and search industries. Killer apps are computer applications that make the technology worth purchasing. Mosaic, Netscape, Internet Explorer, Firefox, Opera, Chrome were the main browsers for the WWW that turned Internet activities into popular and sometimes profitable pastimes.

In addition, computer languages made the WWW malleable. Markup languages like HTML were utilized to display text, hypertext links, and digital images on web pages. Programming languages like JavaScript, Java, Python, and others made web pages dynamic. First by working with the browser and then servers that could determine the assembly and content of a web page, including where to place advertisements.

Hypertext and the Link Economy

The World Wide Web actualized the dream of hypertext, linking a “multiverse” of documents long theorized by computer visionaries such as Vannevar Bush and Ted Nelson. Hypertext provides digital documents with links to other computer resources. What emerged from these innovations was the link economy and the meticulous collection and tracking of information based on the words, numbers, graphics or images what that people “clicked.”

Apple’s HyperCard in the late 1980s created documents with “hot buttons” that could access other documents within that Apple Macintosh computer. Tim Berners-Lee at CERN used one of Steve Jobs’ Next Computers to create the hypertext environment grafted on the Internet to become the World Wide Web. The HyperText Transfer Protocol (HTTP) allowed links in a cyber document on a web browser to retrieve information from anywhere on the Internet, thus the term World Wide Web.

The “click” within the WWW is an action with a finger on a mouse or scratchpad that initiates a computer request for information from a remote server. For example, online advertising entices a user to click on a banner to an advertiser’s website. The ability to point and click to retrieve a specific information source created an opportunity to produce data trails that could be registered and analyzed for additional value. This information could be used for quality improvement and also probabilities of future behaviors

All these new websites, called “publishers” by the ad industry, contained the potential for “impressions” – spaces on the website that contained code that called to an ad server to place a banner ad on the website. The banner presented the brand and allowed visitors to click on the ad to go to a designated site. Over the years, this process became quite automated.

Ad Metrics

When a web page retrieves and displays an ad, it is called an impression. Cost per impression (CPM) is one monetization strategy that measures an advertiser’s costs when their ad is shown. It is based on the number of times the ad is called to the site per one thousand impressions. Online ads have undergone a bit of a resurgence because they do more for branding than previously recognized.

A somewhat different strategy is based on the click-through rate or CTR. In the advertising world, CTR is a fundamental metric. It is the number of clicks that a link receives divided by the number of times the ad is shown:

clicks ÷ impressions x 100 = CTR

For example, if an ad has 1,000 impressions and five clicks, then your CTR would be 0.5%. A high CTR is a good indication that users find your ads intriguing enough to click. Averages closer to 0.2 or 0.3 percent are considered quite successful as banner popularity has decreased.

The Monetization of Search

An advertiser can also pay the publisher when they specifically drive traffic to a website. This is called Pay-per-click (PPC) or cost per click (CPC). PPC is now used by search engines as well as some website publishers.

PPC can be traced to 1996 when Planet Oasis launched the first pay-per-click advertising campaign. A year later, the Yahoo! search engine and hundreds of other content providers began using PPC as a marketing strategy. Pricing was based on a flat-rate cost per click ranging from $0.005 to $0.25. Companies vied for the prime locations on host websites that attracted the most web traffic. As competition increased for preferred online ad spaces, the click-based payment system needed a way to arbitrate the advertisers’ interest.

This led to the first auction systems based on PPC. A company called Overture.com was created at Idealabs, a Pasadena-based incubator run by Bill Gross. Later called GoTo.com, they launched the first online auction system for search in 1998.

Gross thought the concept of Yellow Pages could be applied to search engines. These large books were significant money makers for telephone companies. Businesses would pay to have their names and telephone numbers listed or purchase an ad listed under a category like bookstore, car insurance, or plumber.

Many words entered into online searches were also strongly connected to commercial activities and potential purchases. Therefore, it made sense that advertisers might pay to divert a keyword search to their proprietary websites. How much they would pay was the question.

Overture.com’s real-time keyword bidding system paid online publishers a specific price each time their link was clicked. They even developed an online marketplace so advertisers could bid against one another for better placement. They started with clicks worth only 1 cent but planning that valuable keywords would be worth far more. They invented PPC to emphasize that it was more important that the link be clicked than seen. By the end of the dot.com bubble in 2001, Overture was making a quarter of a billion dollars a year.

The tech recession in the early 2000s put new pressures on Internet companies to develop viable revenue models. Google had developed the best search engine with its PageRank system but wasn’t making enough money to cover its costs. PageRank ordered search results based on how many valid websites linked to a website. So a company like Panasonic would have many valid sites connected to them. Sites that attracted other search engines just because they listed the names of major companies would not get the same priority as with Google. But good search did not mean good profits.

The dominant strategy at the time was to build a portal to other sites. People would come to the website for the content, and the banner ads would provide revenues. Companies would license search capabilities from DEC’s AltaVista or Inktomi and build content around it. This is how companies like HotBot and Yahoo! progressed. So it was a mystifying surprise when Google rolled out its website with no content or banners. Just a logo with an empty form line for entering search terms.

Informed by Overture, Google rolled out a new advertising model called AdWords in late 2000. Initially a CPM (Cost-per-thousand-impressions) model, it developed into a subscription model that allowed Google to manage marketing campaigns. Then, in 2002, a revamped AdWords Select incorporated PPC advertising with an auction-based system based on the work at Idealabs.

Overture sued Google for infringement of their intellectual property but eventually settled. They had changed their name to Goto.com and were acquired by Yahoo! At Goto.com, advertisers “bid” against each other to be ranked high for popular words. When someone searches and clicks on a word like “insurance,” the sites for the highest bidders will appear according to the highest bids. They also automated the subscriber account process. A settlement was agreed to for millions of Google shares in exchange for intellectual property rights to their bidding and pay-per-click and systems. The move marked an offshoot of the digital ad economy. Emerging powerfully with keyword search and auctioning, and combined with MapReduce and Hadoop-driven “big data” processing, Google’s AdWords became an immediate revenue driver.

How Does the AdWords Auction Work?

From Visually.

Google also bought YouTube in 2006 and eventually created a new ad market for streaming videos. They used a new advertising product called AdSense that was offered in 2003 after Google acquired Applied Semantics. AdSense served advertisements based on site content and audience. It placed ads around or on the video based on what it sensed the content was about. Monetization depended on the type of ad, the cost of views or CPM, and the number of views.

Using Behavioral Data

Facebook’s social media platform started its ascent in 2005, but it also needed a way to monetize its content. It first focused on gathering users and building its capital base. That it used, in part, to acquire several companies for their technical base, such as the news-gatherer FriendFeed. By 2009, it had determined that advertising and data-gathering would necessarily be its profit-making strategy with Facebook Ads and Pages.

Facebook started as a more traditional advertising medium, at least conceptually. It would provide content designed to capture the user’s awareness and time, and then sell that attention to advertisers. Advertising had always merged creativity and metrics to build its business model. Facebook capitalized on the economies of the user-generated content model (UGC) and added user feedback experiences such as the “like” button. Also, sharing and space for comments to provided a more interactive experience, i.e., adding dopamine hits.

Facebook had the tools and capital to build an even more elaborate data capturing and analysis system. They started to integrate news provided via various feeders using coding techniques and XML components to move beyond just a users’ friends’ content. Facebook built EdgeRank, an algorithm that decided which stories appeared in each user’s newsfeed. It used hundreds of parameters to determine what would show up at the top of the user’s newsfeed based on their clicking, commenting, liking, sharing, tagging, and, of course, friending.

Facebook then moved to more dynamic machine learning-based algorithms. In addition to the Affinity, Weight, and Time Decay metrics that were central to EdgeRank, some 100,000 individual weights were factored into the new algorithms. Facebook began using what we can call artificial intelligence to curate the pictures, videos, and news that users visitors saw. This aggressive curation has raised concerns and increased scrutiny of Facebook and its algorithms’ impact on teens, democracy, and society.

Frances Haugen, a former Facebook employee testified in October 2021 to Congress about the potential dangers of Facebook’s algorithms. Legal protections allowed her to present thousands of pages of data from Facebook research.[2] The new scrutiny has raised questions about how much Facebook, and other platforms like it, can operate opaque “black box” AI systems outside of regulatory oversight.

Summary

This post discussed how the hypertext protocols created an opportunity to gather useful data for advertisers and also became money makers for web publishers and search engines. The Internet and the World Wide Web established the protocols for hypertext and webpages allowing for a dramatic increase in content to be made available and with it, ad space. The web’s click economy not only allowed users to “surf” the net but collected information on those activities to be tallied and processed by artificial intelligence.

Subsequently, information on human actions, emotions, and sentiments were technologically mined as part of a new means of economic production and wealth accumulation based on advanced algorithmic and data science techniques used to gather and utilize behavioral data to predict and groom user behaviors.

Notes

[1] Zuboff, S. (2019) The Age of Surveillance Capitalism: The Fight for a Human Future at the New Frontier of Power.

[2] Legal protections based on federal laws including the Dodd-Frank Act, a 2010 Wall Street reform law, and the Sarbanes-Oxley Act, a 2002 legal reaction to the Enron scandal give some protections to corporate “whistleblowers.”

Share

© ALL RIGHTS RESERVED



AnthonybwAnthony J. Pennings, Ph.D. is Professor at the Department of Technology and Society, State University of New York, Korea. From 2002-2012 he was on the faculty of New York University where he taught economics of media, and a course on New Technologies in Advertising and PR. He keeps his American home in Austin, Texas and has taught in the Digital Media MBA program at St. Edwards University He joyfully spent 9 years at the East-West Center in Honolulu, Hawaii.

Comments

Comments are closed.

  • Referencing this Material

    Copyrights apply to all materials on this blog but fair use conditions allow limited use of ideas and quotations. Please cite the permalinks of the articles/posts.
    Citing a post in APA style would look like:
    Pennings, A. (2015, April 17). Diffusion and the Five Characteristics of Innovation Adoption. Retrieved from http://apennings.com/characteristics-of-digital-media/diffusion-and-the-five-characteristics-of-innovation-adoption/
    MLA style citation would look like: "Diffusion and the Five Characteristics of Innovation Adoption." Anthony J. Pennings, PhD. Web. 18 June 2015. The date would be the day you accessed the information. View the Writing Criteria link at the top of this page to link to an online APA reference manual.

  • About Me

    Professor at State University of New York (SUNY) Korea since 2016. Moved to Austin, Texas in August 2012 to join the Digital Media Management program at St. Edwards University. Spent the previous decade on the faculty at New York University teaching and researching information systems, digital economics, and strategic communications.

    You can reach me at:

    apennings70@gmail.com
    anthony.pennings@sunykorea.ac.kr

    Follow apennings on Twitter

  • About me

  • Writings by Category

  • Flag Counter
  • Pages

  • Calendar

    October 2021
    M T W T F S S
     123
    45678910
    11121314151617
    18192021222324
    25262728293031
  • Disclaimer

    The opinions expressed here do not necessarily reflect the views of my employers, past or present.