of view and the same matter is realised. To understand Google’s economic practices, then, the points of view from which such repeated practices create and rely on materialities need to be followed. Three such points of view will be examined: those of the searcher or user; of the advertiser; and of Google itself, particularly in relation to its algorithms and datasets and the demands on these created by its commitment to a for-profit corporate logic.
A caveat to this analysis is that Google search has changed and continues to change, becoming more complex over its development. This case study will simplify somewhat by focusing on Google relatively early in its advertising days, at the point broadly speaking when its two key ad programs – Adwords and Adsense – and personalisation through data were established.
In 2016 around 2 trillion searches were conducted using Google’s search engine. To explore user practices, I will follow one user finding an answer (Sullivan 2016). Imagine you are writing a book about the digital economy based on following economic practices in specific digital contexts, rather like this book. As part of this project, you wish to outline the various practices related to searching for information using Google, and, as you write that, you realise it might be helpful to locate those practices in a broader context, perhaps by establishing how many Google searches are made. Practices then follow.
First, your practices have a material context. This will include working on a desktop computer from home, rather than in the office your employer provides, or on a phone while moving around, or on a tablet while commuting to work on a train. There are a range of taken-for-granted practices here: using a mouse and the Microsoft operating system, using alt-tab to switch to an already open browser (Firefox) and knowing that typing a string of words – ‘Google search enquiries 2017 total number’ – into what looks like the address box of the browser will invoke the Google search engine. Underpinning even these unthought and semi-automatic practices are a range of things like electricity, broadband access, light and so on, which create a material context in which our user can sit and quickly bash out an enquiry, hitting return to initiate the search.
Taking most of this material context for granted, our user is now looking at a computer screen on which a window organised by the Firefox browser has opened up. The window is very familiar. At the top are buttons, an address window, and various customisations (Zotero for referencing, for example), which control the actions that can be taken in the browser. Just below this ribbon of buttons is the Google logo, the search box in which the words used in the search reappear, reminding the user what the initial action was. Below that is a list of responses to the query, with a blank space to the side of these answers, though the user knows that sometimes advertisements or summaries will appear there. The list of entries is each an address (a URL) linking to another site on the World Wide Web, with the first two to three lines of each site being shown. From this our user realises that their search words were not the best, as the second entry is for the UK’s Her Majesty’s Land Registry, which records land ownership in the UK and has nothing to do with the number of Google search queries. Scrolling over the entry box for a new search brings up a suggestion – ‘how many Google searches per day 2017’ – which our user clicks on to generate a second set of results. The result that was first on the first search is now second, but our user clicks on that result as it has now shown up twice, and in their experience (that is, in the practices they have developed to search for information using Google), this is a reasonable indication that the answer they want may be found there. Disappointingly, the article linked to is from 2016 but it is from a search engine analysis site and seems reasonably sourced, so our user lifts the 2 trillion a year figure and adds it to their text.
From here our user may move in several directions, perhaps diving more deeply by looking at the top result in the second search, which purports to record live how many Google searches are being made (www.internetlivestats.com/google-search-statistics). Or they may restrain the impulse to dive deeper into the topic and return to the writing at hand. The practice of searching is closely connected to other practices that make up this working life. Our user has one last reflection as they notice that accompanying the second site they looked at there are advertisements for paintings by indigenous Australian artists, and they remember that similar ads have been following them as they visited different websites at other times. These paintings indicate the practices of Google advertising, which we can turn to next.
Many readers will have guessed immediately what happened in relation to the searcher finding ads for indigenous Australians’ art on various websites that have nothing to do with such art, because ads that follow a searcher have become a common experience. The user must at some point have looked at or searched for such art, and tracking mechanisms on the internet have recorded this and used it to target ads. Similarly, some years ago, I booked a trip to Walt Disney World online, which led to Mickey Mouse and his friends stalking me across the Web amid the often noted, and ongoing, irony of being shown ads to go somewhere I could no longer afford to go because I had just paid to go there.
In following the practices of a user engaged in materialising an answer to a question, a second set of practices that can be followed has emerged with advertising, which implies an advertiser. For an advertiser, economic practices, at their crudest, involve trying to persuade people to buy the goods the advertiser has been paid to get them to buy (though it should be noted that advertising strategies may be complex, such as building brand loyalty or gaining attention). The question is, how does an advertiser end up in profit by being paid to boost a company’s sales? Let us assume this advertiser decides to work with online advertising and goes to Google. The magnet Google has to attract potential buyers is its search engine. In the period of Google’s development focused on here, two broad routes are offered. One is that when someone searches for a term related to a product being advertised, then the websites our advertiser has designated show up in the advertising sections of the Google page on which the search query is returned. The second is that Google facilitates ads appearing on other companies’ websites and, again, our advertiser can pay for their products and related sites to show up on sites other than Google’s.
To appear on the search page is to directly draw on the magnet of Google search. Ads appearing here are marked out in slightly different colours and with words indicating that they are ads – ‘sponsored’ often appears – ensuring that they remain distinct from the search results. If our advertiser works for a business that sells package holidays, they may want their site to be advertised to anyone searching for terms like ‘Disney’, ‘beach’ and so on. To do this they have to decide what kinds of words a user might type into search that would indicate they might be interested in holiday products. Google runs an auction on keywords and advertisers bid according to how much they are willing to pay each time someone clicks on the advertisement. The way Google’s auction is set up guarantees that the winner only pays just above whatever the second highest bid was. Once the auction is won, the ads our advertiser wishes to be seen will appear on Google’s search page when the words are used in a search. If a user then clicks on the ad, the advertiser pays Google. There are complications to this simple scenario – such as Google rating good or bad ads, the standards ads must meet, and the information Google offers advertisers to improve their ads and so on – but the fundamentals are in this practice of buying words at auctions which result in ads being served to users of Google search (Turrow 2011: 67; Levy 2011: 87–93).
The second process – in which ads appear on websites other than Google’s search page – runs in a similar way to the first. For example, a fan of model railroads might run a website and seek to gain some funding by signing up with Google to host ads. Our advertiser might then think that people who are interested in model railroads might also be interested in holidays featuring rail travel (and when I say ‘think’ that probably means research into the demographics and holiday habits of model railroaders). Having come to this conclusion, the advertiser bids for the relevant words and then pays in various ways. One key form of payment is what is called ‘cost per mille’, or what an advertiser will pay for 1,000 views of their ad on various sites. Google is paid by the advertiser and then pays the hosting website a percentage based on the ads
