Antonio Garcia Martinez at Books Inc., Mountain View, Aug. 2016


Chaos Monkeys” by Antonio Garcia Martinez (hereafter, AGM) worked as a quant at Goldman-Sachs, came to Silicon Valley in 2008, did a small startup at Y-Combinator, and joined Facebook, where he worked on the advertising engine. He describes his experiences in his book.

Four friends emailed me (thanks!) to tell me the book came out (June 28, 2016) so I bought it, read it, and discussed it with a few friends. Here are my comments about his book, Facebook, and Silicon Valley.

FB, Ads, Metrics

For me, the best part of the book was AGM’s insider info (starting at p. 257) about Facebook (FB). When AGM arrived at FB, revenue numbers were abysmally low. I know this is true because throughout the 2000s, I ran two digital marketing agencies and in 2009, I was the director of Acxiom’s digital agency. We did extensive tracking (with an alphabet soup of KPI metrics such as CPL, CPA, LTV, CM, etc.) and saw that Google ads worked well but FB results were poor. AGM writes that FB’s ad team, starting with Mark Zuckerberg (CEO) and Sheryl Sandberg (COO), plus the ad management team, ignored advertising and revenue. The FB ad tool was junk and they didn’t care. They didn’t know what they didn’t know; they didn’t understand what they had; they didn’t understand their clients or competitors (see pages 279, 392, 394-5, 435, 446, 489).

Thousands of social startups and every social media VC, expert, blogger, and pundit said social data was important. “Wow, look at all those postings! All those likes! It’s gotta be valuable!” So their strategy was to get lots of users and somehow later monetize the data.

But AGM writes (298) that social data has no value because you can’t make money with it. Facebook (nor anybody else) has yet to figure out how to turn social data into revenue.

So what was FB selling? Promises. Dreams. Junk. Sheryl Sandberg (COO) was also on the board of McDonalds and got them to buy $10m of FB advertising (p. 362, 368, 370, 451) as “brand advertising”. AGM gets vicious when he describes how FB executives told nonsense to NYC ad agencies, which shoveled truckloads of their clients’ money at FB, and nobody had any idea how to measure the results. Everyone was making so much money that nobody cared that it didn’t work.

At p. 328, AGM reveals a major Silicon Valley secret: FB buys data. Yep, despite those trillions of likes and postings and all that wonderful Big Data, FB buys data from Acxiom (AGM mentions Acxiom four times at pages 384-9). I knew this in 2010 because at Acxiom, we sold big data to FB, Google, Yahoo, Microsoft, and many other companies (see more on this in the postscript at the end of this review).

So… if FB advertising didn’t work, is FB dead? No, FB is actually a huge success.

Which happened by accident.

In 2013, people switched from desktop to mobile. I remember many of us thought mobile was bad news for digital marketing. When users moved from Google on desktop to Google on mobile, ad revenues collapsed by 90% because on desktop screen, there were ten ads but on a tiny mobile screen, there were only three ads. People don’t click ads when they’re playing games or watching cat videos. FB dragged their feet for more than a year to keep users from switching to mobile. (Remember that? FB kept saying, “we want to ensure a good user experience” but in reality, they were delaying as long as they could.)

Here’s a bit of data that strikes terror in the hearts of Google, Facebook, etc. Mobile CPCs (cost-per-click) are generally less than half of desktop CPCs. Look at the table, especially the bottom three recent quarters: CPCs are less than half. When users switch from desktop to mobile, you make less.


Mobile CPC is half of desktop CPC. The data was prepared for me by Google and comes from one of my large clients.

But the oddest thing happened. AGM writes that nobody at FB, not Zuck nor Sandberg nor the heads of advertising predicted it. Users switched to mobile… and revenues went up!

How did that happen?

FB’s app has 1.7 billion users who are addicted to it for five hours a day. FB allows companies to place ads on primo real estate and charge beaucoup bucks. Ka-ching! That’s ad-sales critter talk. How much? FB’s Q2/2016 is $6.4B which will be +$25B for 2016 and growing 63% over last year.

What if the 14-year old kid next door invents a better Facebook? FB’s market cap is $364B (July 2016) so they just buy her app, her house, the city block, and her city.

FB isn’t the only one to trip over a gold mine. The same happened to Google when they added advertising. Apple too. Steve Jobs was fanatically determined there would be no apps on the iPhone (yep, read the Isaacson bio).

Is there Innovation in Silicon Valley?

AGM’s book makes a general point that few understand and none of the other reviewers noticed. Many people think Silicon Valley is about technology and innovation.

No, it’s not.

From the early 1900s to the mid-90s, Silicon Valley was indeed a place for innovation and technology, where Fairchild, HP, Varian, Loral, Intel, and many other engineering-led companies built oscillators, electronics, accelerators, satellites, transistors, semi-conductors, integrated circuits (chips), data storage on hard disks and RAM, and so on. Engineers could measure performance to see which device performed better.

But that’s over. What is Silicon Valley now? More than 95% of revenue at Google, Facebook, Twitter, Yahoo, Youtube, Pinterest, Instagram, etc. comes from advertising. These aren’t tech companies; they’re ad-driven media companies.

  • Advertising isn’t technology. It’s just ad management which is easy to build. It only took a few months for three FB engineers to build an ad tool that made $100m/year.
  • Advertising can be measured, but only in certain cases. Where it can be measured, it can be quite profitable. But there are many areas in advertising where it’s not possible to track results.
  • To measure advertising, you have to set up conversion tracking, but most advertisers don’t (as many as 80%), because they don’t know how or it’s not in their interest to show that it doesn’t work.
  • If there are no metrics, it becomes a matter of persuasion. The winner is the one who lies the most, pretends the loudest, has the most money, drowns out others, or spends money in self-promotion.

Silicon Valley is now a place where speculators (venture capital, investors, etc.) and entrepreneurs (people who want to get rich quick) use advertising and marketing to pump up valuations. Whether these companies actually do anything or even make money at all is irrelevant; the “wealth creation” or “value creation” (the motto of every Harvard MBA) is in the financial speculation. 150 unicorns have a total $500B in valuation and none make money.

That’s why much of upper management at nearly every large Silicon Valley media company spends their time in political intrigue, “managing upwards”, and backstabbing. They have little understanding of what is happening or how their own companies work. Or they know the company is fake so they’re looting as much as they can before it sinks. Look at how Yahoo collapsed. What about Twitter? Zuckerberg calls it a clown car, which is very funny and totally right.

There are indeed companies in Silicon Valley that build innovative technology. But they don’t turn into companies with $50 billion valuations, so nobody pays attention to them.

Okay, that’s the interesting part of the book. What else? Right, there’s AGM as a person and writer.

The Author as a Person and Writer

It’s hard to read this book because AGM is deeply angry about his employers (Goldman Sachs, Facebook, etc.), his father, his university, women in general, his girlfriends in particular, his children, executives, managers, co-workers, staff, and clients. The author betrays nearly everyone. AGM despises NYC, SF, and Silicon Valley. He races through small towns at speeds over 100 mph. He whines his $500,000 per year isn’t enough. In the end, he ends up alone on a sailboat somewhere off the coast of the Pacific Northwest. He thinks that’s cool; it’s pretty sad. What will his children think when they read what he wrote about them?

AGM adds irrelevant quotations from classical literature, which shows someone can be exposed to learning without learning anything. He uses so many vulgarities (including a few that I’ve never heard before) that he even starts using British spelling.

Do you need to be a jerk to succeed in Silicon Valley? No. I know many successful people and they’re nice. A few jerks stand out, but Silicon Valley is mostly cooperation and collaboration.

A number of reviewers think the first half (up to page 274) is interesting because he writes about how he built a startup at Y-Combinator. But others have written better books. For example, “Founders at Work” by Jessica Livingstone (co-founder of Y-Combinator) is a series of interviews with founders. If you’re doing a startup, you must read it. “The Business of Venture Capital” by Mahendra Ramsinghani (a VC) is an essential textbook on venture capital. “Zero-to-IPO” by David Smith, who has done several startups, is a definitive guide to building a startup. “Patent Practice Skills & Strategies” by Britten Sessions (Silicon Valley lawyer and a professor of law) covers IP law.

Does AGM’s book describe what it’s like to be in Silicon Valley? No, because he barely talks about actually living here. He mostly attacks people.

AGM writes there was no analytics or conversion tracking before he arrived in 2008. Yeah, right. Clicktracks, an analytics tool, came out in 2003. Urchin Analytics also appeared around that time and was soon bought by Google, which was renamed as Google Analytics. All the major analytics tools (Coremetrics, Omniture, Unica, Google Analytics, plus many others) were available by 2005-6. The same for conversion tracking: Google Adwords started in 2002 (I have one of the first accounts) and added conversion tracking soon thereafter. AGM makes this mistake because for him, something exists only if he knows about it.

If you work in startups, social, or digital marketing, you should read AGM’s book. Skip the first two parts (a summary? Part 1: Wall Street sucks. Part 2: Everyone is an idiot). Read Part 3 (“Facebook sucks”). Ignore his attitude and look at the business of Silicon Valley. The book is overall good because of AGM describes how FB was focused on the wrong issues yet blundered into success, which is the general business strategy at a number of large Silicon Valley companies.

Postscript (Extra Stuff)

Okay, you can stop reading. The review is over. Go back to your work.

Here’s a bit of extra stuff.

Several people asked me “but why would Facebook (and Google, Microsoft, Yahoo, etc.) buy data? Don’t they have enough? Isn’t their data valuable?”

Every day, FB collects 63 kinds of data from you. Multiply by the apps on your phone. You’re broadcasting more data than NASA satellites. That’s why your phone battery dies so fast.

Is there more data? What magazines do you read? Newspapers? What TV shows? Radio station?

Oh, we’re just getting started! What mail order catalogs do you get? What do you buy from those catalogs? How often? How much? What computer do you own? Air conditioner? Dishwasher? Clothes dryer? Blender? Color and year for each one? What credit cards? Bank loans (and history for each one, including every single late payment)? Credit card activity? Purchases? Stores? Where do you live now? Last address? The address before that? And every address for the last 25 years? Every phone number you’ve ever had? When did you buy your car? How did you buy it? Loan or cash? Type of car? Color of car? Miles per year? Has it been paid in full? What’s your driving record? What kind of car insurance do you have? How often you rent a car? Do you own your house? How many people live in your home? How many children? How much is the loan? Pay rent? How much? How many places have you lived in the last 30 years? How many pets? How did you do in high school? Grades in college? What is your job? Work history at every company you’ve ever worked for? What about your criminal history? Every arrest?

That’s 61 types of data. Acxiom has 1,500 types of data on you and every person in North America, South America, Europe, the Arab world, China, Japan, and just about every where. It’s a staggering amount of data.

None of the social companies, not even Google (or the NSA or the CIA), have this amount of information (for a number of reasons). So they have to buy it from Acxiom.

By correlating their tiny set of data with Acxiom rich data, they can build profiles of their users, which allows advertisers to place better targeted advertising.