Advertising on YouTube
https://www.theverge.com/2018/10/1/17924422/google-search-targeted-advertising-youtube-expansion
This article discussed a recent shift in Google’s advertising strategy given the popularity of YouTube (which Google owns), Amazon, and Facebook. For years, Google has been one of the technological giants of the web, making about $100 billion each year, a majority of which comes of the advertisements on Google. With the increasing popularity of Facebook and Amazon, especially in recent years with Amazon hitting a value of $1 trillion shortly after Apple last month, Google is potentially losing out on ad revenue to its opponents. As a result, Google is looking at YouTube, the second most popular search engine in America after Google Search, to better connect advertisers with viewers and hopefully stave off competition from Facebook and Amazon.
In class, we have analyzed the way that advertisers pay for and are given slots in the web to conduct advertising. For instance, in the case where the search engine, such as Google, knows all advertisers’ values for clicks, this can be modeled as a matching market where a perfect seller’s graph can be created. If values are not known, we learned about the VCG mechanism (Vickrey-Clarke-Groves) that encourages truthful bidding and is similar to a second-price auction for single-items. Rather than charging the “winner” of the bid the second highest bid for the item the bidder won, the bidder pays for the “harm” he caused by taking that slot. In the VCG case, bidding truthfully is the dominant strategy. On the other hand, there is the GSP (Generalized Second-Price Auction) which results in untruthful bidding and socially non-optimal outcomes. In the case of GSP, it is possible for bidders to get a higher payoff if they bid untruthfully and a less desirable slot. It is thought that Google’s advertising algorithm is similar to that of GSP. It would be interesting to see though how Google would adapt this algorithm to YouTube. In the GSP algorithm, it appears that the advertisements are based mostly on search history. For instance, when I look up “shoes” in Google, the first four links are advertisements that direct me to places to buy shoes: zappos.com, shoes.com, kohls.com/shoes, and justfab.com. In my experience with YouTube, however, the ads do not always seem to correlate with my search history and/or the video I’m watching. There always seems to be a period of a certain advertisement blasting on every video I click, i.e. the Wix designing a website, upcoming movie trailers, or car advertisements that do not seem to necessarily be related to my demographics or search history but are simply ubiquitous throughout YouTube throughout a specific time period. It would be interesting to see how Google handles these ubiquitous ad cases. Perhaps the advertising strategy on YouTube might have to be similar to that on Facebook/Instagram, for Facebook is not really considered a “search engine,” and most use it to simply scroll. While some of the advertisements on Facebook/Instagram are related to my personal demographics and interests (i.e. college ads on Instagram), those platforms also blast popular ads such as movie trailers to seemingly everyone that uses the media. It would be interesting to see how Google not only assigns advertisements to advertisers on YouTube, but also what kinds of data Google is using to decide which advertisements are being shown.