How Google Works
I recently read How Google Works and found it really interesting. Usually I don’t take notes when reading (maybe I should…), but I found myself marking a lot of pages to come back to later. These notes aren’t at all a comprehensive representation of the book, just a collection of ideas that I found most striking.
How Google Works
Foreword by Larry Page
Larry wanted to become a professor or start a company
Goals were: autonomy and thinking from first principles rather than prevailing “wisdom”
Think about things that aren’t possible (what if you could download the entire Web and just keep the links)
Hard to get teams to be super ambitious—people assume things are impossible rather than using real-world physics to figure out what’s actually possible
Attract smart creatives and create an environment where they can succeed at scale
- Deep technical knowledge, hands on experience
- Analytically smart, can deal with data and makes data driven decisions
- Thinks about competition
- Thinks about the user
- Firehose of new ideas
- Curious, questioning, unsatisfied with the status quo (possibly overbearing)
- Takes risks
- Open, collaborative, and judges ideas on merits and not provenance
- Thorough and knows the details
- Communicative, charismatic
No one has all of these characteristics—key commonality is work ethic and willingness to question the status quo and attack things differently => impact
Culture—Your Own Slogans
Messiness is a virtue
- Randy Pausch—“If your kids want to paint their bedrooms, as a favor to me, let ‘em do it.” (He had covered his childhood bedrooms walls with handwritten formulas)
Don’t listen to HiPPOs (Highest Paid Person’s Opinion)
The rule of seven
- In many organizations this states that the max number of direct reports is 7, at Google the minimum number of direct reports is 7
- This leads to flatter organizations
Bezos two-pizza rule—teams should be small enough to be fed with 2 pizzas (small teams get more done, spend less time politicking, more efficient per person, require less communication)
fun, not Fun
- Fun is manufactured, fun is organic—nothing wrong with organized events, but they need to be fun For offsites, forget “team-building” and have fun
Strategy—Your Plan Is Wrong
Foundational blueprint: Bet on technical insights that help solve a big problem in a novel way, optimize for scale, not for revenue, and let great products grow the market for everyone.
Combinatorial innovation—the availablity of components that can be recombined to create new inventions
- New technologies often come to the world in a primitive condition, for very specific problems (steam engines were used to pump water out of mines long before they powered trains)
Default to open, not closed—examples: arpanet and TCP/IP, IBM PC was open and had probably 100 times the engineering compared to Apple
Eric’s Notes for a Strategy Meeting
- Start by asking what will be true in 5 years and work backwards—what will change quickly?
- Consider disrupters—nearly perfect market information and broad availability of capital, think about product and platform
- As an incumbent, when there is a disruption you can acquire, build, or ignore (for some time).
- As a challenger to an incumbent you must: understand the tools an incumbent will use as obstacles to stop you (business relationships, regulations, lawsuits, etc.), consider the role of other players whose incentives can be aligned to help you, growth matters most
- Growth matters most
- Articulate a rough time frame and end point
- Slides kill discussion (don’t use market research and competitive analyses, get input from everyone in the room)
- Iteration is the most important part of the strategy—it needs to be very, very fast and always based on learning (data driven)
- Gather your strategists carefully—shouldn’t be based on tenure or titles (the best smart creatives and people who will have a good perspective on the changes to come)
Talent—Hiring Is The Most Important Thing You Do
Google’s founders modeled hiring off of academia (not corporate America)
- Universities don’t typically lay off professors so they invest lot’s of time in getting faculty hiring and promotion right (normally using committees)
- Hiring should be peer-based, not hierarchical
- Focus on bringing the best possible people into the company, even if the fit isn’t a perfect match for one of the open roles
The herd effect
- Having a group of brilliant people leads to more brilliant people wanting to join
- Early recruiting ad said “You’re brilliant, we’re hiring”—Marissa Mayer saw this on a recruiting flyer in the CS building when she was a student at Stanford. She brought hte phrase with her to Google.
- Jonathan would keep a stack of resumes of the people he had hired and would show people the resumes of the group they would be joining
Passionate people don’t use the word
- People have learned that passion is a strong indicator and will describe themselves as passionate. Passionate people demonstrate their passions without needing to label themselves as passionate
Expand the aperture
- Google’s Associate Product Manager (APM) was started by Marissa Mayer when she was given the prime directive to hire the smartest computer scientists they could find straight out of school
- Google hasn’t always been successful at expanding the aperture. Kevin Systrom was a promising marketing associate (and self taught programmer). Salar Kamangar was impressed and wanted to transfer him into APM. Execs fought back because he didn’t have a CS degree. Eventually Kevin left Google and cofounded Instagram.
- When struggling to hire a head of HR to round out executive team Eric blurted out “Find me a Rhodes scholar who is also an astrophysicist”. Sergey asked a lawyer to write a contract that is well done, comprehensive, and funny. Then the search partner suggested McKinsey partner + Rhodes Scholar => Shona Brown.
- Once, when hosting a group of Rhodes Scholars Jonathan was trying to decide who would receive offers. Sergey said “Why decide at all? Offer all of them jobs.”
- Expanding the aperture brings risks: startup costs for a brilliant, inexperienced person are higher and failures may happen. The book claims that these concerns should be set aside for the greater good and that brilliant generalists are far better for the company as a whole.
Everyone knows someone great
- Why isn’t it everyone’s job to recruit the great person they know?
- At >500 employees often managers start worrying more about headcount than finding brilliant people. You hear the phrase “fighting for headcount” a lot more often than “finding great people”.
Interviewing is the most important skill
- Develop a process for pre-interview and interview
- Trusted-reviewer program—an elite team of people who were actually good at interviewing and enjoyed doing it
Friends don’t let friends hire (or promote) friends
- For several years Larry reviewed every offer—a process designed for quality, not efficiency and control, not scale
- To this end the hiring packet was designed to be comprehensive (contain everything known about the candidate by Google), standardized/systematic, and data driven
- Packets include summary statistics on a reviewer’s past scores in order to normalize for differences between interviewers
- Promotion is also done by committee (which has a calming effect since denial/lack of promotion comes from a more faceless entity than an individual’s decision)
- They plug Laszlo Bock’s book Work Rules
Urgency of the role isn’t sufficiently important to compromise quality in hiring
- Candidate packets are designed to that they can be reviewed in 120sec or less
Trade the M&Ms, keep the raisins
- Google had several rotation programs with execs. Jonathan had chief of staff for 6 month stints out of the APM programs. Also side programs (so-called chain gang tasks) for APMs e.g. helping Larry Page learn more about how projects got done at Google. Gives exposure to executives and makes employees’ lives more interesting
- Create positions for key contributors that are ready to expand their responsibilities
- Institutionalized rotation programs are great, but hard with non-entry level employees. Encouraging internal mobility can be a proxy for rotations
- If you do this, make managers trade away their high performers (M&Ms) and not just their raisins
Career—Choose the F-16
At the start of their career many people prioritize the company, then the job, and last the industry. This is the opposite of what you should do. It is difficult to change industries so choosing the right one at the start is very important. From a compensation standpoint, equity can be somewhat limited early in your career and it is more lucrative to develop expertise in the right industry than to bet on a particular company. Later on you can leverage this expertise to become well-compensated.
Decisions—The True Meaning of Consensus
Beware the bobblehead yes
- Many people will agree in a meeting and then complain about the decisions made afterwards
- The goal is not to be unanimous, it is to come to the best idea for the company and to have everyone rally around it
- “Be interested in finding the best way, not in having your own way”—John Wooden
Meet every day
- When a decision is crucial for the future of the company make decision makers meet every day
- “At first they didn’t make much progress. But the sheer drudgery of repeating the same argument every day helped spur the team to…”
Every meeting needs an owner
- Meetings should have a single decision-maker/owner
- The decision-maker should be hands on
- Even if a meeting is not a decision-making meeting—for example it’s designed to share information or brainstorm solutions—it should have a clear owner
- Meetings are not like government agencies—they should be easy to kill
- Attendance at meetings is not a badge of importance
- Timekeeping matters
- If you attend a meeting, attend the meeting (don’t be on your laptop multitasking)
Spend 80 percent of your time on 80 percent of your revenue
- It is easy to focus on new things because they are shiny and interesting, but your core business is what pays the bills for now and requires attention
The World’s Best Athletes Need Coaches and You Don’t?
- Get an executive coach. Eric and Jonathan had Bill Campbell.
Communications—Be a Damn Good Router
Default to open
- Larry (and previously Eric) post their OKRs and hosts a company-wide meeting to discuss them. The grades of the previous quarter’s OKRs are on display
Know the details
- While Eric would talk to managers about what is going on with their teams, Larry used snippets (weekly status reports that cover a person’s most important activities in a short, pithy format) from the engineers directly. This allowed Larry to cut more directly to the truth.
Start the conversation
- Most companies have “tribal elders”. At Bell Labs these were called “the guys who wrote the book”, often because they had. People should be directed to these experts, but warned to be prepared so that they don’t waste the star’s time (or embarrass themselves).
Have a playbook For 1:1s Bill Campbell does
- Performance on job requirements
- Relationship with peer groups
Press interviews should be a conversation not a playbook. Corporate messaging can be taken too far. Insightful conversations are more important and better than memorizing a script.
Relationships, not hierarchy
- The steady state of successful Internet Century venture is chaos. When things are running smoothly the processes and infrastructure have caught up to the business. This is a bad thing. The business should be outrunning the processes (and you should be constantly adapting).
- Three-week rule: When you start a new position, don’t do anything for three weeks. Listen to people, understand their issues and priorities, get to know and care about them, and earn their trust. Rather than taking immediate, disruptive action (as an outsider) you are establishing a healthy relationship.
Innovation—Create the Primordial Ooze
The CEO needs to be the CIO
- It’s ridiculous to think that you can delegate or create a “uniformity of command” for innovation. A CIO (Chief Innovation Officer) is doomed to failure by the very need to create this position. You must create innovation ooze (innovative culture), or, better yet, don’t lose what you had when you were small.
Focus on the user
- How will we monetize Google Earth? Sergey wanted to buy Keyhole, asked Jonathan to answer the question. He made up an answer on the spot (not how they monetized it). Eventually Sundar Pichai thought of a way. The innovative product was more important than a path to revenue.
- “You aren’t thinking big enough”
- “Think 10X”
- 10% better gas mileage requires iteration. 500mpg requires you to start from scratch.
- There is a symbiotic relationship between big challenges and highly smart, skilled people.
Set almost unattainable goals
- OKRs should not be all green at the end of the quarter
- 70% core business
- 20% emerging
- 10% new bets
- Larry created a hacky setup to copy books to see if digitizing all books ever written was feasible. Possible for a team of funded engineers, but restrictions spur creativity.
20 percent time
- Check and balance on imperial managers
- Often turns into 120% time
Ideas come from anywhere
- Marissa Mayer ran meetings that were like The Gong Show: People presented their ideas and could keep talking until they got gonged.
- These meetings morphed into “Beer and Demo” meetings. You drink beer and vote on projects you like
Ship and iterate
- Larry scrawled “These ads suck” on printouts and posted them on the wall
- Marissa posted negative product reviews on the wall outside her office for scrutinizing with PMs and engineers
- Forgetting sunk costs is tough and important
- Any failed project should yield valuable technical, user, and market insights to inform your next efforts.
- Don’t stigmatize teams that fail. They should land good internal projects and jobs
- “Good judgement comes from experience; experience comes from bad judgement”
It’s not about money
- People are not rewarded for successful 20% projects
Conclusion—Imagine the Unimaginable
Ask the hard questions
- Ask yourself what could happen in your industry in the next five years. What could change most quickly, and what will not change at all?
- How would a smart, well-capitalized competitor attack the company’s core business (or some other problem)
- Go through your pipeline and determine what percentage of them are built on unique technical insights.
- Find out how many product people are on the senior leadership team.
- Are you in the right place? There are hubs for finance, tech, fashion, diamonds, biotech, energy, shipping, cars, and most other industries. Do you go to the smart creatives in these hubs or try to get them to come to you?
Big problems are information problems
- Computers push humans to be better at chess. Why not in other pursuits?