by Brad Stone
No, advertisers couldn’t make vague claims in their ads. They couldn’t use exclamation points; that would be shouting at the customer. Their colors couldn’t garishly stand out because it might distract shoppers. They couldn’t use images that showed excessive skin. And on and on.
Advertisers accustomed to getting rich sets of demographic data about customers from Silicon Valley firms were also rebuffed. No, advertisers couldn’t access Amazon data about its customers’ age, ethnicity, and shopping habits. No, Amazon wouldn’t allow companies like Adobe and Acxiom to put their third-party software tags onto ads and track their performance, a common practice elsewhere on the web. Advertisers would have to settle for getting reports about their ads’ effectiveness directly from Amazon.
Inside the advertising group, a few of these battles became notorious. Over one holiday, Paul Kotas vetoed the particular hue of blue in ads by the Ford Motor Company, because the display campaign felt “Sunday circular.” Amazon also told the wireless carrier T-Mobile that its trademarked magenta-pink logo was too bright and distracting. And it informed Sony Pictures that a banner for the James Bond film Skyfall violated the policy on showing weapons. The studio “was like, ‘screw you!’ ” an Amazon ad executive recalls. “Who is James Bond in silhouette without a gun? Literally, he’s just a random dude.”
Amazon compromised in many of these disputes—and ultimately allowed 007 to hold his iconic weapon, reasoning that it wasn’t pointed at anyone in particular. But advertisers came to view the company as arrogant and remote. “We would be met with warm handshakes to start the relationship, and by the end they were tired of us,” said Steve Susi, an Amazon creative director for five years.
The stubbornness spoke to Amazon’s ambivalence about display advertising and a religious refusal to violate the trust of customers in any way—a philosophical departure from the approach of its Silicon Valley peers, such as Facebook. For Bezos, during the first part of Amazon’s journey into advertising, the sanctity of the customer experience took absolute precedence over any business relationship or incremental boost to the balance sheet.
There was also internal suspicion toward any ads that directed a clicking customer off the Amazon website and away from making a purchase. For years, the ad team offered a service called Product Ads that allowed other retailers, such as Nordstrom and Macy’s, to promote their wares on Amazon’s website. The goal, to allow customers to see more selection and competitive prices, was misunderstood and unpopular inside the company. “The retail team spent all their time trying to drive traffic to Amazon and we were driving traffic off Amazon,” said Colleen Aubrey, Amazon’s vice president of performance advertising. “I remember having regular meetings with retail leaders where they were like, ‘What are you guys doing?’ We would have to talk it through.”
By 2014, Amazon was close to retiring Product Ads, with the overall enthusiasm for advertising inside the company waning. The financial results were good but not great. Restrictive guidelines and inaccessible senior executives were alienating advertisers. The division was constantly fighting for resources, and its execs were putting in sixty-hour weeks but felt unappreciated and under attack. “We were bad for a long time and people held us accountable for that,” said one advertising executive.
* * *
In a shake-up that summer, Paul Kotas was promoted to senior vice president and placed in charge of the entire advertising group. Lisa Utzschneider, Amazon’s longtime global VP of ad sales, who had previously reported directly to S-team member Jeff Blackburn, quit in frustration six months later to work for Yahoo. Amazon’s nascent advertising effort was in tatters.
Kotas had joined Amazon in 1999 from D. E. Shaw, the Wall Street quantitative hedge fund where Bezos had conceived the original idea for an online bookseller. He liked to tell employees the story of how Bezos had actually recruited him to join Amazon in 1997. His bags were packed for Seattle when he decided to stay at the hedge fund at the last minute. The change of heart ended up costing Kotas millions.
Other than a fondness for punk and new wave music, Kotas was, like many of his colleagues, obsessed about metrics, like the length of time it took for ads to load, and fanatical about the leadership principles, such as frugality. “Tell me more about this dinner on your expense account” was a familiar refrain to his direct reports. When ad execs gathered in “war rooms” at the start of the peak season to monitor the performance of the holiday ad campaigns, Kotas could be relentless: “If any of you are thinking about going to grandma’s house and turning off your phone and not being available, think again!” an ad exec recalls him saying one year.
Kotas took over as sole manager of the beleaguered advertising group just as the possible answer to its perennial problems began to reveal itself. At the time, Amazon’s marketplace business was clicking into high gear. Third-party sellers—including the flood of merchants coming online from China—were eager to boost the visibility of their products on the increasingly crowded pages of search results. The solution was obvious: charge them for it, just as Google taxed web publishers to promote their websites in its search engine.
Amazon’s Google-style-search ad auction was called “sponsored products.” It allowed a third-party seller of bedsheets, for example, to place a bid to feature its linens whenever customers entered terms like “bedding” into Amazon’s search engine. At first the ads showed up at the bottom of the first page of search results; if a user clicked on a sponsored listing, they were transported to the product page and Amazon collected a fee.
As Amazon expanded sponsored products into more product categories and the ads migrated to the right-hand side of the page alongside search results, the advertising team had trouble developing the necessary technology fast enough. It had to build a search auction system to take bids from advertisers and a tool to track the effectiveness of the ads and report the results back to them. Amazon’s first search advertisers recalled that early versions of these services were flimsy. “The reports you would get back about the success or failure of the campaigns we ran were very, very poor,” said Jeremy Liebowitz, the former global e-commerce vice president of Newell Brands, maker of Sharpie markers and Elmer’s glue. “It was almost impossible to tell if something actually worked.”
Amazon also needed to make the right semantic connections between ads and specific search terms. Google had two decades of experience in the complex field of search relevance, whereas Amazon was a relative novice. Every time Bezos or Wilke saw a misplaced ad, they fired off an email to Kotas, who in turn forwarded it to his engineers in frustration. One ad technology engineer recalled a memorable ruckus when a sponsored listing for a sex toy popped up in the results for children’s toys.
Search ads were controversial inside Amazon and even in the ad group itself. Amazon ad sales execs in New York, L.A., and London, whose mission was to sell conventional banner ads, were given aggressive sales goals every year. Now their clients were being distracted by a whole new way to spend their ad budgets with Amazon. Engineers working in Amazon’s Silicon Valley–based search division, A9, loathed the new search ads; their job was to elevate objectively useful products in search results, not the listings of sellers who had paid the most money for an ad.
Yet sponsored ads clearly worked. Customers clicked on them, often failing to distinguish between sponsored placements and objective search results. While they grumbled about the expenditure, sellers and brands were already accustomed to search advertising on Google and seized on it as a way to stand out on Amazon. By 2016, the S-team was confronting the ads’ increased popularity and debating a key question: Should it allow search ads to appear on the top half of the search page—mixed with organic search results?
The debate over so-called “above the fold” placement was fierce and played out over countless meetings, pitting the sanctity of the customer experience against a promising new revenue stream. Ad execs argued that sellers and vendors would benefit from having their products appear
at the top of search results. Retail executives worried customers might be led astray by ads for low-quality products, have a bad experience, and decrease their overall spending on the site.
In one debate, Doug Herrington, chief of the domestic retail division, used the parable of the scorpion and the frog to frame the issue. In the story, the scorpion asks the frog if it can climb onto its back for passage across a river, then can’t help but sting the frog along the way, dooming both. His counterparts in advertising were the scorpion—they weren’t evil, per se, but it was simply in their nature to pervert the more egalitarian playing field of the authentic search results.
Eventually Bezos had to settle the argument. His predictable answer was to methodically test sponsored ads at the top of search results, first in a small percentage of search queries. The engineers who administered the tests never thought their instrumentation or data was very reliable, but the results were fairly consistent. When sponsored ads were prominently displayed, there was a small, statistically detectable short-term decline in the number of customers who ended up making a purchase. The longer-term effects were unknown. The scorpion wasn’t killing the frog, but lightly nicking it—and it wasn’t clear yet whether the bite was poisonous.
While there would almost certainly be collateral damage—fewer customers finding what they wanted—sponsored products also made money. A lot of it. And in that respect, Bezos’s decision whether to expand ads to the top of all search queries also consisted of a single word: Yes. Amazon should continue to expand the percentage of search results that included sponsored listings. Yes, it should increase the number of sponsored listings on each page of search results, even if that meant a small corresponding drop-off in customer clicks.
Back in the days of display advertising, Bezos had resisted compromising on the customer experience. But now, while he cautioned against alienating customers by serving too many ads, he opted to vigorously move forward, saying that any deleterious long-term consequences would have to be implausibly large to outweigh the potential windfall and the investment opportunities that could result from it.
Search ads had all of the business characteristics that Bezos loved. Customers weren’t transported off Amazon when they clicked but sent to individual product pages, where they made purchases and fed the flywheel. Few expense account–wielding mad men were needed to administer it; the system was largely self-service. And once the technology was in place, search ads would produce tremendous leverage—and a huge windfall that Bezos could use to finance new inventions.
“Moving ads to the top of search pages was the game changer,” said an Amazon computer scientist who worked on the ad business. “Sponsored products would be nothing close to what it is today if that decision wasn’t made, and Jeff was the one who signed off on it.”
* * *
Once Bezos showed a willingness to convert the relative meritocracy of search results into a domain that prioritized Amazon’s commercial interests, the possibilities were endless. For example, Bezos had received an email from a customer in Florida a few years before, who described visiting Amazon.com to buy a selfie stick. There were hundreds of choices and the customer had no idea which one to buy; then he went to a local store and got advice from a salesperson. Why couldn’t Amazon, the customer wrote, offer such a recommendation?
The S-team was already deliberating that question when Bezos forwarded the email, which kicked around in the retail organization and ultimately landed with a team working on voice shopping for Alexa. Their product was called Amazon’s Choice. It weighed variables such as customer reviews, prices, and shipping speed to bestow a recommendation on certain products in an otherwise crowded category when users asked Alexa to order it.
Voice shopping grew rapidly on Alexa, and the Amazon’s Choice badge started appearing prominently in search results in 2016 alongside sponsored products. Its meaning was ambiguous—the company did little to explain what the endorsement actually meant, but it seemed to fill the role of an informed salesperson at least somewhat. Customers nevertheless flocked to products that were assigned the badge and, by one independent account, lifted sales by a factor of three. Not surprisingly, when merchants weren’t trying to fool the system and earn the badge by generating positive reviews, they were desperate to know how they could pay for it. Amazon execs replied that it wasn’t for sale. “We said it’s fairly simple,” said Assaf Ronen, a voice shopping vice president. “Take your best products, make them cheap, and make customers happy.”
But Amazon’s Choice served the company’s interests in another way. The groups working on private-label products, such as AmazonBasics batteries, clamored for the badge when they saw it assigned to rival brands like Duracell. That sparked another round of tense internal debate, pitting private-label teams against A9 search engineers and even Paul Kotas and the ad team, who argued that giving house labels prominence in search results would undermine the company’s advertisers and corrupt the impact of the badge. Nevertheless, the badge gravitated toward many of Amazon’s private-label products, giving them another advantage over competitors in search results. A Wall Street Journal story in 2019 about Amazon’s Choice found that 540 AmazonBasics items were awarded the badge, more than any other brand.
Naturally, aggrieved vendors complained when they saw the Choice badge assigned to Amazon’s house brands. Though Amazon’s lawyers restricted the practice, particularly as European antitrust authorities investigated Google for the similar practice of privileging its own services in search results, private-label executives felt sheepish about it. “For a lot of people on the team, it was not an Amazonian thing to do,” said JT Meng, a former manager in the consumables division. “Just putting our badges on those products when we didn’t necessarily earn them seemed a little bit against the customer, as well as anti-competitive.”
Amazon’s search results had evolved from a straightforward, algorithmically ordered taxonomy of products into an over-merchandised display of sponsored ads, Amazon Choice endorsements, editorial recommendations from third-party websites, and the company’s own private brands. In some product categories, only two organic search results appeared on an entire page of results. Since brands and sellers could no longer count on customers finding their products the old-fashioned way, through the site’s search engine, they were even further inclined to open their wallets and spend more money on search ads. A bipartisan report by the U.S. House antitrust subcommittee would later disapprovingly conclude that Amazon “may require sellers to purchase their advertising services as a condition of making sales” on the site, since consumers only tend to look at the first page of search results.
By 2017, revenues from sponsored products had eclipsed those from display advertising like banner ads and would soon after leave them in the dust. That year, sales in the “other” category on Amazon’s income statement (the former home of AWS revenues), where the company parked advertising revenues, hit $4.65 billion—a 58 percent jump from the year before. Amazon had discovered a veritable gold mine in its own backyard.
* * *
But first Bezos had to prevent the consumer retail business from looting the sponsored ad business. By insisting during the OP1 reviews in the fall of 2017 that Amazon’s oldest business must stand on its own merits, without the camouflage of advertising, Bezos forced a reversal in the company’s longtime operating posture. An all-encompassing pursuit of revenue and market share growth was replaced by the quest for profit. A mandate to plant seeds in all corners of the business was replaced by a sense that only the largest trees mattered—often ones sowed by Bezos himself—while other expensive bets should be pruned back.
Over the next few months, Amazon did something rare in its history: it retreated. Websites set up to sell tickets to concerts and other events were shuttered in the U.S. and UK. The company slowed down the introduction of a service called Amazon Restaurants into new cities, where it was meant to compete with startups like Grubhub, DoorDash, and, in the UK, Deliveroo,
and two years later closed it altogether. It also cut the investment in its beleaguered Chinese marketplace, which was hopelessly behind Alibaba and JD.com and would also close for good in 2019.
Amazon also froze hiring in the retail division almost entirely. For years there had been few constraints on recruiting talented employees. Amazon’s headcount in Seattle had ballooned to forty thousand employees in 2017 from five thousand in 2010. Over the next year, it was basically flat. “We just decided that after years of rapid fixed cost investment, it made sense for us to slow it down at least for a year and digest the growth and make sure we were being efficient,” said Jeff Wilke.
Profitability, a foreign concept for years, was tried on for the first time, like a new suit. Retail executives were ordered to revisit relationships with major brands, like Coca-Cola and Unilever, to extract more favorable terms on products, like bottled water, that were costly to ship. They turned again to their search engine to advance the company’s commercial priorities, experimenting with factoring the profitability of items into the algorithmic equation that determined which products received an “Amazon’s Choice” badge. The ongoing “hands off the wheel initiative,” which replaced retail managers with software, also kicked into higher gear, with brands directed to tools on the Amazon website to run their promotions and manage their sales, instead of working with Amazon employees. The company still offered white glove service to its largest vendors but now would charge them for it.
