That Friend Who Always Knows What to Say: How Google is Helping Brands Talk to Consumers

I don't talk to my friends the same way I talk to my grandma. Neither does Google.

Google is building machine learning-based advertising products to help brands optimize content in real-time and develop more contextually relevant messaging for target consumers. A report by Harvard Business Review notes, “Research has shown that digital targeting meaningfully improves the response to advertisements and that ad performance declines when marketers’ access to consumer data is reduced,” [1].

In July 2018 Google launched Responsive Search Ads, which uses aggregated consumer data and machine learning to help brands determine the right messages for the right audiences more efficiently than ever. Google explains the product as follows:

Responsive search ads combine your creativity with the power of Google’s machine learning to help you deliver relevant, valuable ads.

Simply provide up to 15 headlines and 4 description lines, and Google will do the rest. By testing different combinations, Google learns which ad creative performs best for any search query. So people searching for the same thing might see different ads  based on context.

We know this kind of optimization works: on average, advertisers who use Google’s machine learning to test multiple creative see up to 15 percent more clicks [2].

Using machine-learning in advertising products changes the process flow of advertising from one where brands create a few pieces of content, test them, and serve highest performing content to their designated audience to a model where brands input a selection of copy and visuals and serve consumers unique permutations of any given message. However, using consumer data to customize ads has the potential to backfire, as consumers are increasingly worried about data protection [3].

Responsive Search is a short term improvement from traditional A/B testing and allows brands to be more nimble because they can quickly test multiple permutations of copy and images and optimize a given campaign in real-time [4]. This improves consumers’ experience with advertising because they are targeted with content that is more relevant to them, leading to higher click rates. Google’s Gmail already finishes writers’ sentences with Smart Compose; in the medium term, advertisers may not even need to come up with lines of copy to test, but rather Responsive Search will suggest copy based on a consumer’s interests and purchasing behavior [5]. For example, an advertiser could input a campaign objective and Responsive Search would deliver the best message to consumers who are most likely to be considering a purchase without the brand needing to input test copy and images. This would signify a shift toward what Harvard Business School Professor John Deighton calls “artificial general intelligence,” where Google’s machines are actually able to interpret advertisers’ needs and consumer behaviors rather than execute upon a strict set of rules [6]. Can machine learning help Google takeover the role of an advertising agency, and if so, what challenges may arise as a result?

It is important to note that access to consumer data is the input to machine learning that allows Responsive Search and other ad products to improve consumer response to branded content. If consumers see a customized ad and feel that their personal data was taken unfairly, there may be negative effects for brand perception [7]. Scandals such as Cambridge Analytica exploiting the personal data of over 38 million consumers and serving them targeted political ads have marred Facebook’s brand and shed a spotlight on the fight for transparency in data collection [8]. So, if Responsive Search serves someone to a message that seem too much like the advertiser has been watching them, Google’s brand may suffer. Professor Deighton suggests the onus of protecting data privacy is on companies, rather than regulation, asserting, “[a company] has an incentive to police the actions of the parties to which it sells the data, and to conduct its own interactions with a degree of civility, because its interests are aligned with the interests of the customer,” [9]. Because of this, in the medium term Google will need to determine how much it will allow its machines to learn, and how it will prevent its products from being used as a weapon. Further, Google must continually audit what advertisers are feeding into the responsive search software to ensure that consumers are being treated fairly and being given accurate and defensible information from brands.

The solution to the perfect advertisement lies somewhere between consumers wanting to see recommendations and content based on their preferences and not wanting so much customization that the advertisement feels uncomfortable. How will a machine find that line? (738 words)


[1] John Leslie, Tami Kim, and Kata Barasz. “Ads That Don’t Overstep: How to Make Sure You Don’t Take Personalization Too Far.” Harvard Business Review 96, no. 1 (January–February 2018): 62–69.

[2] Jerry Dischler. “Putting Machine Learning Into the Hands of Every Advertiser,” press release, July 10, 2018, on Google website, [], accessed November 2018.

[3] John, Kim, and Barasz. “Ads That Don’t Overstep: How to Make Sure You Don’t Take Personalization Too Far.”

[4] Susan Young. “Getting the Message: How the Internet is Changing Advertising.” Harvard Business School Working Knowledge, May 16, 2000,].

[5] J.D. Biersdorfer. “Let Gmail Finish Your Sentences.” New York Times, June 1, 2018, [], assessed November 2018.

[6] Gideon Lewis-Kraus. “The Great AI Awakening.” New York Times, December 14, 2016, [], accessed November 2018.

[7] John, Kim, and Barasz. “Ads That Don’t Overstep: How to Make Sure You Don’t Take Personalization Too Far.”

[8] Matthew Rosenberg, Nicholas Confessore, and Carole Cadwalladr. “How Trump Consultants Exploited the Facebook Data of Millions.” New York Times, March 17, 2018, [], accessed November 2018.

[9] Manda Salls and Sean Silverthorne. “Should You Sell Your Digital Privacy?” Harvard Business School Working Knowledge, August 25, 2003, [], accessed November 2018.


Machine Learning and the Future of Agriculture


C.H. Robinson – Predicting the Future in the Global Supply Chain

Student comments on That Friend Who Always Knows What to Say: How Google is Helping Brands Talk to Consumers

  1. I think it is very difficult to get to perfect advertisement and much of the power of the advertisement is lost by the current power we (customers) own today in just switching to another page or closing the video when it pops-up. Using machine learning to tailor advertisement helps increase the value of ads in a way of trying to get to a win-win situation in which websites are only showing me things I’m supposedly interested in. I agree the use of personal data can also bring threat regarding to privacy but also to distinguishing what is behind our behavior. The technology the way it is today for me is simply regurgitating our steps and brings no intelligence behind to support it. You search for Casper one single time in Google while reading a case for the class and boom now you become target for hundreds of mattress propaganda with zero intention of buying any of them.

  2. Google is leading the innovation in digital advertising. Given the full-stack solution provided, it could track consumers cross device, geo, platforms, etc. The perfect ad is to send the right message to the right audience, through the right channel, at the right time. Given search has very high purchase intent, Google is well positioned to outperform ad products by competitors. Advertisers are moving away from CTRs, how does the new responsive search change the way advertisers track, measure, and evaluate campaign performance?

  3. I completely understand the value of targeting advertising. As you said, “This improves consumers’ experience with advertising because they are targeted with content that is more relevant to them, leading to higher click rates.” However, you mentioned the question of consumer privacy and I similarly wonder if this new “responsive search” technology will hit a threshold of efficacy. Customer know when a brand is stalking them. Does this constant and/or hyper-personalized presence help close the sale or does it negatively impact the customer’s perception of the brand? I wonder if more research can be done to understand where that line resides. Are there lessons to be learned from traditional advertising – when does a brand become known for its ubiquitous advertising instead of as a brand? Is all news/presence good news? Thanks for an interesting read!

Leave a comment