The book is Ben Horowitz’s memoir with a particular focus on his company Opsware and the lessons he learned there.



From Communist to Venture Capitalist

Ben’s father was a communist, and he grew in the People’s republic of Berkeley. His first job was at NetLabs, which was run by professional management with little appreciation for the product or the technology. This job taught him the importance of founders CEOs. The job took a toll on his personal life, and to stabilize that, he left and joined Lotus. As soon as he came across Mosaic, he was mesmerized by the Internet and applied for Netscape. Marc Andreessen interviewed him, and is now his business partner at A16Z.

Leadership is the ability to get someone to follow you even if out of curiosity.

I will survive

Ben started LoudCloud, which was doing well before the dot-com bust. Its primary customers were all startups, and the customer bankruptcies decimated the revenue growth. The private funding was hard to come by, and Ben decided to take LoudCloud public. The Saas model of recognizing revenues only when they are earned (whereas the software revenue is recognized as soon as it is sold) and similar issues made the S-1 filing look bad. It got termed as IPO from hell. The reverse split to keep the stock price high enough further destroyed the fantasies of the employees. The IPO roadshow on the east coast was severely impacted by his wife’s west coast allergic reaction leading to the emergency room. The IPO happened at $6 per share. No one, not even the underwriting bankers were happy. The macroeconomic condition worsened, and the revenue guidance could not have met in the further quarters. So, the finance department suggested that let’s reset it completely, rather than doing that in steps.

If you are going to eat shit, don’t nibble.

The underwriters, Goldman Sachs and Morgan Stanley, dropped analyst coverage in response; the stock fell to $2. Then 9/11 happened. Then 26/11 Exodus, Loudcloud’s largest competitor, a $50 billion market cap company with $800 million raised filed bankruptcy. Ben realized that the operations are too fragile; he thought of acquiring Data Return and realized that neither’s future is bright. So, he decided to exit the cloud business completely and extract out the software business, which was built for cloud automation (Opsware). The stock slowly rose to $4, and Ben decided to raise $50 million from private investors. That’s when Atriax, the largest customer who owned $25 million, filed for bankruptcy. The plan to raise money was put on hold, and the acquisition was sought out. EDS and IBM were played against each other on an artificial deadline of eight weeks. EDS wanted to Loudcloud more than IBM. It eventually made the acquisition, excluding Opsware, for $63.5 million. It even licensed Opsware for $20 million a year. 140 employees got laid off, another 150 went to EDS, and 80 were left with Opsware. On Bill Campbell’s recommendation, Ben stayed with the employees on the west coast when the announcement happened in New York. In Bill’s words, “If you don’t treat the people who are leaving fairly, then those who stay back would never trust you again.”

If you don’t treat the people who are leaving fairly then those who stay back would never trust you again

This time with feeling

The stock price has fallen to $0.35, and NASDAQ warned that the stock would be delisted. Some members of the executive team who did not understand software accounting & sales had to let go of. The stock price rose to $7. But then EDS, which accounted for 90% of the revenue, wanted to cancel the contract since the deployment has not gone as smoothly as expected. EDS gave Opsware 60 days to fix all the issues. The head of the EDS servers’ team loved Tangram and hated CA products. EDS was going to get the latter for free in a deal. Opsware did the diligence of Tangram, found it to be bad, but even then acquired it for $10 million, which was a 67% premium to its market value at that time. Opsware offered Tangram for free in that 60-day window. The problem seems to have been solved. But then BladeLogic started to beat Opsware. The stock price was back to $2.90. The company went for another war-mode for the next six months. When things were going well, it was decided to go beyond server automation and look into network automation.

All decisions are objective until the first line of the code is written. And then they are emotional.

Ben could have built something internally or acquire another company. All decisions are objective until the first line of the code is written. And then they are emotional. Ben, therefore, decided to see if there is a possibility to acquire before starting an internal effort. Among the four automation companies, the cheapest one (Rendition networks) had the best architecture. Ben believed that the market was wrong. Opsware acquired Rendition networks for $33 million, three months later, Cisco paid $30 million to license its technology. Slowly the acquisition offers started to tick in. Virtualization was also taking over. The share price was about $7. Ben talked to the team, and almost everyone wanted to exit. He decided that the price would be $14. HP acquired it for $14.25 per share ($1.65 billion in cash).

When things fall apart

No Plan B

Startup CEOs should not play the odds. Statistics are against them, and they don’t change the difficulty of the task at hand. A good startup CEO can focus and make the best move when there are no right moves.

A good startup CEO has the ability to focus and make the best move when there are no good moves.

An excellent principle to follow is the first principle of Bushido: keep death in mind at all times. So, you may conduct yourself properly in all actions. Being a startup CEO is a fierce struggle, if you don’t like it, don’t do it, but if you do it, don’t take it personally, find the best move, persist, and convince the team to follow you.

Share all the news

As a startup CEO, share the news, both bad and good. It builds up trust, brings in diverse opinions, and allow bad news to be known faster. When people bring problems out in the open, they get solved. A culture that punished people from talking about the problems prevents the free flow of information. An example would be “don’t bring me a problem without bringing a solution”.

In any human interaction, required amount of communication is inversely proportional to the level of trust.


Loudcloud went through three layoffs but still got sold for more than a billion-dollar. Layoffs destroy the morale of the employees who are left. Unless they are done right.

Layoffs destroy morale of the employees which are left. Unless they are done right.

If the financial situation requires, the CEO should clear off his head of the past glory, execute quickly to avoid leaks, be honest about the reason that its company’s failures and not the individual’s performance issues, layoff must be completed by the manager since the employee is going to remember every last detail of this day, the CEO should address the entire company, and after the layoff, the CEO should be present and visible. Messaging and execution are essential for people who are staying.

The messaging and execution [of layoffs]  is important for the people who are staying.


Firing an executive is relatively more straightforward since they have been on the other side of the table.

  1. Root cause analysis – Figure out why the executive was hired in the first place. It could be a poor job role definition based on some abstract idea. Hiring could have been for lack of weakness. Hiring for the scale too soon, which might be attained in some hypothetical future. Hiring a generic “great” head of sales, ignoring the unique situation of your company. An Executive who had a mismatch of ambition or failed to integrate into the company.

Hiring should be for strengths not for the lack of weakness

There is no great head of sales. There can only be a great head of sales for your company for next 12-24 months.

2. Inform the board
Call board members individually and face them, then call the board meeting, give a good severance package to the executive, ensure that his reputation is preserved.

3. Prepare the conversation
State decisively (“I have decided” not, “I think”), be clear on the reason, and have the details of the severance ready. Let him decide how to communicate the decision to the company and the world. You cannot let him keep his job, but you absolutely can let him keep his respect.

You cannot let him keep his job, but you absolutely can let him keep his respect.

4. Inform the company
Inform their direct reports, other staff members, and the company, in that order, on the same day. CEO should ideally fully take over the role in the interim.  It will also help the CEO decide whom to hire for the position.

Demoting a friend

As the company grows, sometimes, it is crucial to demote an early employee who can no longer head the bigger unit anymore. It is essential to sacrifice his good for the good of other employees. This ends up messy. He might want to leave; if he stays on the same team and reports to a new boss holding his position, he might try to sabotage things. Acknowledge his contribution, be decisive in the language and admit the reality that if you as a CEO are under-skilled, it is even harder to develop him and that in itself is a bigger recipe for failure.

Loser Lies

Humans listen to leading indicators of good news. Revenue grows, immediately hire more. Revenue falls, no immediate action. When a company struggles, leaving employees are labeled with “they were bad anyway”, lost sales deals are deemed to be lost because competition gave the product away for cheap. These are all lies that we tell ourselves. It is important not to fool ourselves with them.

Lead Bullets

Sometimes, there are no silver bullets, and there are several lead bullets that are required to improve the product. This saying is true more often when the customers are buying a similar product from the competition because it is simpler, faster, and/or easy to use. At that time, pivoting or building a new feature is akin to looking for a non-existent silver bullet.

Nobody cares

Nobody [except you] cares about the underlying cause of why things are going wrong. It does not help them at all. Don’t waste time explaining why things are going wrong; spend it on trying to make them right.

Don’t waste time explaining why things are going wrong, spend it on trying to make them right.

People > Product > Profit

Ben wanted to hire a new head of sales, Mark Cranney aced the interview but was from a lesser-known university and did not look like the head of sales. The more time he spent with him, the more he was convinced that Mark is the right guy. He fought against the whole board, and the decision turned out to be great in the long run. Hiring the right people and taking good care of them is of the highest importance to an organization.

A good place to work

Being a good company does not matter when things go well but can be the difference between life and death when things go wrong. The only reason someone would stay at a company going through a bad phase is that he likes his job. When things go wrong, a bad company won’t be able to hold onto the employees, and that’s when the death spiral begins.


Even new workers at Mcdonald’s get the training. Training is one of the highest leverage activities a manager can perform. It sets the basis for performance expectations. It ensures consistent product quality. Training ensures employee retention, since people leave, primarily, because they hate their manager or aren’t learning anything new. The most basic training to start with would be the functional training for the job functions and the management training for the managers.

Hiring from Friend’s company

Hiring from a friend’s company is always tricky. Either you will hire mediocre people or their top-notch employees. It will never go down well since the trickle of employees leaving will begin after that. During the interview process, tell the candidate that you need to do a background check with that CEO friend and to let you know if you don’t want that to happen.

Big companies execs transitioning to small companies

In big companies, execs are interrupt-driven. They always have incoming requests to process; their job is not about taking new initiatives but fine-tuning the existing ones. In small companies, the requirement to take the initiative is much higher. Big company execs are good at following processes, at small companies, they have to create one. Screen them right at the time of the interview by ensuring that they have the self-awareness of this. Train them heavily by assigning them an initial set of tasks.

Hiring an Executive

The two biggest mistakes which can happen are “looking for the lack of weakness and not strength” and “hiring on some abstract look & feel”. Hire the right person for your company at this particular point in time. List the specific set of strengths you are looking for. The CEO must make a final solo decision without being influenced by anyone else.

When employees misinterpret managers

An organization should not be run as a black-box with the sole focus on metrics and quantitative goals. Such a focus usually leads to an incorrect prioritization of the goals. Run the organization as a white box, many questions to judge its progress and performance would be more qualitative; for example, long-term competitive edge,  customer satisfaction, etc.

Management debt

Short-term, expedient decisions with expensive long-term consequences can lead to a management debt. The three most popular ones are – putting two in a box (promoting two people to the same position and making them share responsibilities when only one is required), overcompensating someone (because they got another job offer), and no performance management. Most experienced CEOs opt for hard choices in these cases to avoid management debt. A good HR division provides management quality assurance; it designs a good process, works with managers to identify bottlenecks, and finds out unspoken problems inside the company.

Company politics

CEOs who avoid politics, accidentally, end up encouraging it.

The right kind of ambition is ambition for the company’s success and the individual’s own success being a by-product of it.

If one executive complains about another’s behavior, get both of them into one room to resolve the issue. If an executive complains about another’s performance and you already knew it, then it is already too late; you should fire that executive since improving skills is doable, but regaining the support of the organization is hard for him/her. If an executive complains about another’s performance and it’s new news, strongly disagree right there and then evaluate the complaint objectively. Individuals optimizing personal achievements over the company are bad. Worse if they are managers since that encourages employees to do so.

Titles and promotions

Employees love titles, and for an organization of beyond a few tens of employees, everyone needs to know who is who. The author believes that titles are one of the cheapest things which a company can offer, so if a better title helps to hire an employee, use it. Mark Zuckerberg has the opposite thinking, where he prefers leveling down the new hires to avoid giving them better titles than existing employees. People at any level compare themselves to the worst person at the next level and want promotion as soon as they reach that level. Such promotions lead to the law of crappy people. A good promotion process, to avoid that, must ensure the criteria of whether an employee is ready for the next level is an objective one.

When smart employees are bad employees

Some smart employees are bad.

  1. The heretic – The one thinks a bunch of morons runs the company. This thinking usually happens because he feels disempowered, is fundamentally a rebel, or is immature. It is difficult to fix this situation since once an employee has taken a public position, it is a credibility hit for them to change their stance.
  2. The flake – If an employee is not reliable, all the intelligence is a waste.
  3. The jerk – Bad behavior cripples communication. If the VP of marketing is a jerk, then no one will bring up the issue of marketing in the meeting, communication will slowly break down, and the topic of marketing will never come up. A few jerks who are making massively positive contributions are ok, but adding more of them will destroy the company.

If anytime someone raising a concern about marketing is attacked by VP of marketing then topic of marketing will stop showing up in the meetings.

Old People

Hiring senior people into a startup is like an athlete taking a performance-enhancing drug; if all goes well, you will achieve new heights. If all goes wrong, you will start degenerating from the inside out.


A one-on-one is an employee’s meeting where he can talk about issues, frustrations, and a manager must have regular one-on-one with their employees.

Company Culture

Properly designed culture might look bizarre from the outside, but that’s a coincidence, not the end goal. Amazon has desks made out of doors, A16Z fines its partners 10$ a minute for being late in a meeting with founders, Facebook has the motto to move fast and break things.  All these things reflect the core values of the company.

Scaling a Company

Scaling a human organization requires an organizational design (all have shortcomings, choose the one which suits best), and a corresponding [communication] process. The organization should dictate the individual leaders and not the opposite scenario, where the personal ambitions of the individuals dictate the organizational structure. If you want particular individuals or groups to communicate more, put them under the same manager. The further away people are in the organizational chart, the less they will communicate.

The further away people are in the organizational chart, the less they will communicate.

The Scale Anticipation Fallacy

Hire executives who will do the job for the next 12 months, don’t anticipate and try to hire executives who can handle scale longer than that.  Any judgment passed on an executive by comparing them to some hypothetical size of the company in the future is a bad idea. Scaling is an acquired skill, the act of judging them in advance creates prejudice. And even hiring a more experienced executive does not help since you would have to reevaluate eventually whether s/he is the right person or not. Rather than passing an absolute judgment, always think about whether you can hire a better person at this point and who will s/he be.

How to lead when you don’t know where you are going

After Ben sold Loudcloud business to EDS, the stock price of Opsware fell to $0.35 a share, putting the market cap half of their cash in the bank. NASDAQ threatened to delist them. Ron Conway told Ben to talk to Herb Allen (Allen & Company). Herb and their clients bought the share over the next few months, bringing the price to $3 a share. Herb later told Ben that even though he knew nothing about technology, he decided to invest in Ben’s courage and determination.

The most challenging CEO skill

In a human organization, things are bound to go wrong. A CEO usually takes two extremes, one where s/he takes the blame for everything and gets too stressed out and other who s/he tries to casually pass up every broken thing as normal, which frustrates the employees and turns the organization into crap.

The fine line between fear and courage

The most critical decisions test courage far more than intelligence. The right decision is obvious, but the pressure to make the wrong one can be overwhelming. Going with the crowd is always the safer choice but not always the best outcome. Every correct decision makes you more courageous and the wrong one more cowardly. The financial bar for starting a company has gone down; the courage bar remains high it has ever been.

The most important decisions test courage far more than intelligence.

Ones and Twos

There are two core skills for running an organization. Knowing what to do and getting the company to do what you know. The “ones” are good at knowing while the “twos” are good at getting the company to move. Ones gather data from many sources and make good decisions. They dislike execution-related tasks like process, goal setting, and performance management. Twos have a love for action and find it challenging to make critical decisions. An ideal CEO should have both skills. This leads to a tricky CEO transition problem. Most CEOs will be ones and have twos reporting to them, promoting the twos to the CEO could be harmful. Microsoft did that with Bill Gates to Steve Ballmer transition in 2000. Promoting one from deep within can cause the executive twos to leave, GE did with Jack Welch in 1981, and that turned out to be phenomenal. No easy way to do the transition.

Follow the Leader

There are characteristics of a good leader. Ability to articulate a vision (“Steve Jobs”), right kind of ambition  – ambitious but not selfish (“Bill Campbell”), and ability to achieve the vision (“Andy Grove”). A peacetime CEO maximizes and broadens current opportunity, wartime usually has just one bullet to hit the target. Google moved from peacetime (“Eric Schmidt”) to wartime (“Larry Page”) as of the writing of this book.

Making yourself a CEO

CEO has to make a lot of unnatural moves, constantly evaluating people is one of them. The end goal of your feedback is to open up rather than close down the discussion. Don’t try a shit sandwich, be authentic. As a CEO have an opinion on everything. Continuously give feedback so that people won’t take it personally and will focus on the content instead.

How to evaluate CEOs

Some employees make products; some make sales, a CEO makes decisions. That’s what a CEO is judged on. The best way to judge CEO is on his/her strategy, decision making, ability to convey that to the company and achieve desired results measured against a set of objectives.

The first rule of entrepreneurship: There are no rules

Right when Opsware was in a bidding war between BMC and HP, the  E&Y partner from BMC’s diligence process claimed that the reinstatement of finances is required unless the contracts were amended in 48 hours. Ben was able to gather the three [bank] customers and change the contract. BMC still backed out, and acquisition happened for $14.25 when Ben was expecting it to be $15. His learning was that there is no point arguing when things go south, accept it, and get on.

Accountability vs. Creativity paradox

If you punish people for not meeting deadlines regularly, they will avoid hard problems. If you don’t hold them accountable to the deadlines, then hard-working employees will feel bad. This paradox is hard to deal with. As a general rule, senior employees should be better at forecasting. Deadline slips are more probable if the task is tough. And always hold people accountable if they took a stupid risk.

Freaky Friday Management Technique

Read it at Ben’s blog

Should you sell your company?

If your company is getting an acquisition offer for its product or the business, consider the eventual market if you think that the final market can be much bigger than what your company has realized and can you be number one in that. If it is yes on both counts, then it is better to stay on the course. For example, Google rejected $1B offers, and they were pursuing a massive market.  Pointcast, on the other hand, was not pursuing a large market, and by rejecting $1B offers, it made a mistake.