19 February 2022 |

Amazon's $31B ad business, explained

By Trung Phan

Jeff Bezos long shunned ads, but the opportunity is massive and the e-commerce platform has gone all in on it.

Hey, thanks for subscribing to SatPost.

Today, we’ll be talking about Amazon’s massive $31B ad business (and the meme dump is back).

PS. If you were forwarded this email, I’m happy to report your internet is working and that you can hit the button below to join the mailing list.

Amazon Ads

One of the wildest running themes for Big Tech is the size of their side hustles:

  • Apple’s Wearables business (AirPods, Watch etc.) is on a $40B a year run rate. In comparison, Tesla’s revenue hit $54B in 2021.

  • Alphabet’s YouTube is currently on a larger annual revenue run rate ($34B) than Netflix ($30).

These huge numbers are obviously part of a larger antitrust conversation. And it’s hard to treat these businesses as truly standalone (to wit: you can’t pair Apple Watch with an Android phone).

Having said that, Amazon is the latest Big Tech company to officially unveil a “sizeable side hustle”. Two weeks ago, the company broke out its ad business (for years, it was classified in company filings under “other revenue”).

The punchline: Amazon’s ad platform generated revenue of $31B in 2021, which is nearly 3x the combined sales of Twitter, Pinterest and Snap (all ad-based social platforms).

On a semi-related note, here are the first 8 image results when you Google “Bezos laughing”:

The success of its ad business is a 180-degree turn for Amazon. In 2009, Bezos famously said “ads are the price you pay for an unremarkable product or service”.

Here’s how the pivot went down:

  • The state of ads on Amazon

  • Amazon’s original position on ads

  • The man running Amazon ads

  • Why Amazon embraced ads

  • What ads mean for the future of Amazon

The state of ads on Amazon

In February 2021, Marketplace Pulse published an article titled “Everything on Amazon is an Ad.” A quick scroll through Amazon confirms the headline:

  • Search results: On the left is a search for “toothpaste”, with the first 4 results returning sponsored products (depending on the product, the first 3-7 search results can be ads).

  • Product page: On the right is a product page for speakers. Everything shaded in blue is ad inventory.

A year later in February 2022, Amazon revealed the size of its ad business for the first time: revenue hit $31B in 2021, per CNBC.

That figure is a significant jump from $1B that Amazon reported as “other revenue” in 2015. As highlighted in the intro, here’s how Amazon’s ad revenue stacks up with Twitter, Snap and Pinterest [insert Bezos laughter].

Amazon’s ads have grown so much that — in 2018 — it passed Microsoft to become the 3rd biggest digital advertiser. Now, Amazon’s $31B in annual digital ad revenue only trails Google ($200B+) and Meta/Facebook ($110B+).

And unlike Meta/Facebook — which reported that Apple’s new iPhone privacy tracking policy will cost its ad business $10B in 2022 — Amazon isn’t at the whim of another Big Tech rival.

Amazon’s original position on ads

In 2003, three Amazon researchers published a cornerstone paper for the company: “Amazon.com Recommendations: Item-to-Item Collaborative Filtering.”

For Amazon users, the ultimate e-commerce experience was to be built on a recommendation engine. Not ads.

Per the paper (bold mine):

At Amazon.com, we use recommendation algorithms to personalize the online store for each customer. The store radically changes based on customer interests, showing programming titles to a software engineer and baby toys to a new mother. The click-through and conversion rates — two important measures of Web-based and email advertising effectiveness — vastly exceed those of untargeted content such as banner advertisements and top-seller lists.

In recent years, Amazon’s organic recommendations…

  • “Customers who bought this also bought this”

  • “Customers who viewed also viewed”

…have been replaced w/ ads:

  • “Sponsored products related to”

  • “Brands related to this category”

One of the few organic recommendations that still display is the trusty-old “Frequently bought together” feature.1 Sometimes they make sense…

…and other times, less so:

The man running Amazon ads

On August 22nd, 1994, Jeff Bezos posted the first job listing for Amazon (which was originally called Cadabra; as in abracadabra 🤷 ): “well-capitalized Seattle start-up seeks Unix developers”.

Befitting someone that also considered the name “Relentless” for his startup (the relentless.com URL redirects to Amazon), Bezos is hilariously blunt about the type of colleague he wants:

  • “You must have experience designing…complex systems…and should be able to do so in about 1/3rd the time that most competent people think is possible

  • “Expect talented, motivated, intense and interesting co-workers”

The person running Amazon’s ad business — Paul Kotas — is extremely aware of Bezos’ exacting standards: they worked together at hedge fund D.E. Shaw in the early 1990s.

Bezos recruited Kotas to Amazon in 1997. He declined but joined 2 years later after a phone call that probably went something like this:

Bezos: Paul, you gotta come to Amazon.

Kotas: I dunno man. Things were pretty intense the last time we worked together.

Bezos: C’mon. We’re changing the world.

Kotas: Can you be slightly less relentless?

Bezos: No, but I can give you a ludicrous amount of $AMZN stock.2

Kotas: OK, I’m in.

The Information has some great details about the 61-year old Kotas, who is one of the most powerful people in all of digital advertising but flies under-the-radar:

  • Low key: When he meets with ad execs, Kotas tells his team to identify him as someone who works in “product” (his real title is Senior Vice President). Part of his stealthy approach is to “avoid drawing unnecessary attention” to Amazon’s ad efforts.

  • Technical background: He’s an “engineer by training” and has ~30 patents under his name. He’s very analytical and “wants arguments backed by data” (kind of like Bezos).

While Bezos hated ads, Amazon has such valuable digital real estate that — in 2005 — he tapped Kotas to explore ways to monetize it.

Per The Information, the first ads were plain vanilla display ads on the product pages. But there was a big problem: competing retailers bid for the ad spots and when users clicked on them, they left the Amazon website.

The project was soon shut down. The next Amazon foray into ads would take a few years and came amidst a rapidly changing tech landscape.

Why Amazon embraced ads

You know the saying “when it rains, it pours”?

That basically describes Jeff Bezos’ angel investing portfolio. In 1998, he cut one of the most baller angel checks ever: Bezos gave two Stanford PhD dudes $250k for their startup.

Those dudes were Sergey Brin and Larry Page.

Here’s how the deal came together per Business Insider (citing Brad Stone’s The Everything Store):

“…the investment story starts with Amazon’s 1998 acquisition of Indian delivery service Junglee, which eventually tanked. But the acquisition brought Ram Shriram to the Amazon team, which proved to be a fortuitous meeting for Bezos. 

Shriram had been discretely advising two Stanford PhD students, Larry Page and Sergey Brin, who were trying to invent a new way of searching the internet. In February of 1998, Shriram became one of the first early investors of Google, with a $250k investment. 

Six months later, Bezos and his wife were vacationing in the Bay Area when he reached out to Shriram with a request to meet the guys behind Google. Shriram invited the Bezos,’ along with Page and Brin, to his house for breakfast with a demonstration of how the search engine would operate. Bezos immediately told Shriram he wanted to invest.

It took some convincing on Shriram’s part, since the early funding cycle had closed, but Bezos’ status as a CEO with a then $1.6 billion net worth swayed Google’s founders to let him in.”

Bezos later said of the deal: “There was no business plan. They had a vision. It was a customer-focused point of view. I just fell in love with Larry and Sergey.”

When Google went public six years later in 2004, Bezos had 3.3m shares of $GOOGL worth ~$285m. He’s since unloaded his stake, which would otherwise be worth $8.5B today. In 2005, he launched his family office — Bezos Expeditions — and has written early checks to Twitter, Uber and Airbnb among others.3

Making the very unrealistic assumptions that he’s entirely held his positions and faced zero dilution, here are guesstimates as to how much those bets could be worth:

Anyways, I bring this up because Google is the reason that Amazon ultimately went all-in on the ad business.

In the early 2000s, Amazon was becoming overly dependent on Google search slots to drive traffic. Per The Information, the company even created a “Google Reliance metric” to track this dependence…which could potentially be existential.

Google, of course, occupies a lucrative part of the customer’s buying journey. It sells ads at the point that someone shows intent for a purchase.

However, Amazon owns the last foot of the transaction funnel: the purchase. Amazon’s transaction data is particularly valuable for 3rd-party merchants, which have grown from 3% of Amazon sales (2000) to ~60% of its $300B+ retail sales (2021).

And by one account, more than 60% of product searches now start on Amazon.

This is yet another example of Amazon turning what is an expense into a revenue opportunity. As highlighted by Social Capital in a viral presentation slide, Amazon has methodically turned various expense lines into money makers:

  • Product costs —> Amazon Kindle, Amazon Basics

  • Fulfilment costs —> Fulfilment by Amazon

  • Technology costs —> Amazon Web Services

  • Marketing costs —> Amazon Prime

  • Payment costs —> Amazon Payments

What ads mean for the future of Amazon

Amazon ads might already be the company’s most profitable business line.

Using Google’s 68% margin for its core search ad business as a comparison, tech analyst Benedict Evans writes:

“Given [Amazon Ads] margin structure and incremental cost base, it’s highly likely to be generating similar absolute profits to [the Amazon Web Services cloud business].”

To stand out, sellers are wiling to pay for sponsored placements (just like brands buying shelf space in grocery stores). And Amazon sells a number of ad products:

  • sponsored products (search)

  • sponsored brand

  • sponsored display ads

  • sponsored posts

  • sponsored videos

While Amazon says that ads are “optional”, getting buried on page 5 of search results is bad for business. The average cost per click (CPC) on Amazon ads is climbing. And it’s not uncommon for merchants (there are ~5m on Amazon) to spend up to 50% of a product on listing fees and ads.

As the the main Amazon e-commerce platform gets more saturated, Amazon is finding other forms of ad inventory like:

The economics of running a 3rd-party merchant business is trending towards scale. A number of companies are raising big money to “roll up” Amazon merchants: Thrasio ($3.4B raised), Perch ($909m), Heyday($800m), Razor Group ($560m), Elevate Brands ($373m)

The profitable siren call of ads is invading other e-commerce platforms, too:

  • Walmart’s ad business cleared $2B last year

  • Instacart has poached a number of Facebook execs and is projected to have a $1B ad business in 2022

  • DoorDash and UberEats are building ad platforms

What’s next?

Marketplace Pulse writes that “advertising has distorted Amazon’s customer-obsession” and “clouds the retailer’s ability to innovate on discovery, personalization, and any form of interactive shopping. At a certain point, every decision to improve the experience competes with lost revenue from advertising it would replace.”

That’s pretty depressing. Thankfully, I can drown my sorrows away by purchasing some pressure cookers and toilet seat washlets.

Meme dump

I’m trying to squeeze as much juice as possible out of this $4.5B Bitcoin / Razzlekhan heist. You can read my full breakdown of the situation here.


The most recent news around Amazon’s recommendation engine is quite awful: the platform was apparently recommending preservatives that can be combined for suicide (based on other user purchases).


And Kotas definitely got a lot of stock. The current CEO of Amazon (Andy Jassy) made a base salary of $160k — which was the max for a corporate employee — when he ran Amazon Web Services (but obviously had a shit ton of stock). Amazon recently bumped the max base salary to $350k.