Startups, company culture, technology, music, writing and life

Month: May 2015

An Open Letter to Tony Hsieh

Dear Tony,

First of all, let me start by stating that I truly admire what you’ve achieved. Your book, describing your path from LinkExchange to Zappos, from culture-by-accident to the deliberate creation of an awesome company based on culture, was an inspiration for us at GrantTree. The idea of optimising a company for employee happiness, as well as the knowledge that someone else had done this on a much larger scale than my cofounder and I, were invaluable guideposts on a road less travelled, towards cultures built upon openness and respect.

I also completely agree with where you’re trying to take Zappos. I believe that teal/open (or holacratic) cultures are a better way to operate, in many ways, with many advantages, at the very least for the people working there. I wholeheartedly agree with the objective you’re pursuing.

However, as someone who has, over the last couple of years, helped an open culture come to life, albeit on a much smaller scale at GrantTree, I find myself worried by your approach towards implementing Holacracy and the “Teal” model described by Frédéric Laloux.

I am not a culture consultant. I am not a social scientist. I am not a coach. But I am a founder of a company that, over the last couple of years, transitioned to a model that shares some features with Holacracy, and has a lot more in common with the Teal organisations described by Laloux, which I like to call open cultures. I don’t know whether you’re aware of the points below. You may very well disagree with them, and you may well be right and I wrong. I offer them up humbly hoping they might be of use to you.

My observations fall into a few key categories. First, an observation about basic respect for the individual. Then, concerning two of the key “breakthroughs” of Teal organisations according to Laloux: self-management and wholeness. Then concerning holacracy itself. Finally, concerning the agenda of open cultures itself.

As a final caveat, I should add that I don’t have any insider knowledge as to what’s going on inside Zappos itself. Perhaps the picture that emerges from reading the memo is incorrect. Perhaps things are going swimmingly and I’ve misinterpreted all the cues I’m sensing.

1. Respect for the individual

The first and perhaps most important point I observe is that the approach displayed in the memo seems, at least from the outside, to have little respect for individuals. I don’t doubt your personal commitment to creating a respectful workplace: it is a fundamental building block of happiness, and of course of trust.

And I think trust is the problem here. Open culture systems like Holacracy are fundamentally reliant on trust. It is the basic currency of an open culture company. And disrespect towards individuals does not breed trust: instead, it’s highly corrosive to trust.

In order for things like tension processing1 to function, you need trust. Without trust that they will not be needlessly punished for their opinions, people cannot be candid. Without candour, tensions are not brought to the circle’s attention and therefore cannot be resolved.

The message in the memo, “get on board with holacracy now or leave”, broadcast to the entire company and to the world at large, does not seem respectful of individual differences. Instead, it seems hostile to individuality, trying to cut everyone into the same mould without respecting or even listening to or acknowledging their differences. Starting from this premise, it can only alienate people and cause them, even if they were ready to progress to a better perspective, to regress back to a “me vs them” point of view which is toxic to Holacracy and any open culture agenda.

I respectfully urge you to consider approaching this important transformation in a more patient way, more respectful of individual differences.

2. Self-management

Holacracy is a model of self-management. It dispels the need for people managers, and distributes the responsibility for decisions to people throughout the organisations. This is great – when it works.

Self-management cannot work with people who do not want it. You can’t make people self-manage any more than you can make people love. Trying to impose self-management top-down is a bit like grabbing someone by the shoulders and screaming at them, “I want you to love me, now, spontaneously! Why don’t you love me?”

In order for people to want self-management, they need two things: to personally, viscerally believe that it is good for them; and to believe that it is possible. Both of those are things that take time. It took us two years to transition 10-20 people to self-management. We did not have the clarity of vision that you have, so you may be able to do this more quickly, but getting 1400 people to do anything will always be a challenge.

Even after 210 people have left, chances are there are many, many left who do not yet want self-management. No matter how much you try to impose it from above, they will not take responsibility for their work until they want to. And the more you try to force them to take responsibility, the less they’ll want to.

I hope that you will be willing to consider giving people more time and more examples of self-management to inspire them, rather than attempting to inspire self-management via top-down orders.

3. Wholeness

You quote Frédéric Laloux’s book quite substantially in your memo. The second “breakthrough” that he identifies as part of Teal cultures is wholeness.

What this means in the context of work culture is have the space, at work, to bring your whole self to work. This is something that has unfolded at GrantTree over the years and, for me, one of the most inspiring consequences of open cultures. When people feel free to be themselves instead of being cut down to size, they expand and swell to become superb versions of themselves. This touches every area of their life, from personal life to work life. It is truly inspiring to see people rise past their personal and emotional limits and fully grow into themselves. If I were to pick one single achievement in my life to be proud of, this is it: being instrumental in creating an environment that has enabled this to happen.

However, this breakthrough cannot happen without respect for the individual. When you push people into a “me vs them” mindset, you force them back into a fear-driven mode of operation. In that mode, they cannot be themselves: they are afraid of showing who they are for fear of being punished. Only a truly respectful, trust-based environment can achieve this breakthrough.

Once again, I request that you consider how to adjust your approach to open companies so it shows deep respect for individual differences rather than attempting to cut everyone down into the same mould.

4. Holacracy

Holacracy is a fascinating system, full of interesting ideas. I have felt its seductive pull myself, but I always resisted it. This was not because I was afraid of losing power or authority, quite the opposite. It was because I did not feel that it was well suited to GrantTree’s chaotic, freedom-loving ways. It seems too rigid, too structured. Yes, it promises that through this structure personal freedom will be achieved. Not all promises come through, though.

A first crack in Holacracy’s armour comes from noticing that it takes an all-or-nothing approach. Either you implement all of Holacracy, or none of it. This approach to open cultures belies the experience of all of the companies in Laloux’s book, all of which evolved open, teal culture systems progressively, over time. Skepticism turns to suspicion when I realise that the “all or nothing” argument enables easy evasion if Holacracy fails to deliver any improvement to company culture. It gives an easy way out of “well, you didn’t really adopt the whole thing properly, it’s hard, don’t beat yourself up for it”.

Another crack comes in practice, from noticing issues with it. GrantTree adopted some parts of Holacracy – specifically, something called a Holarchy, which is a way to organise work hierarchically without organising people hierarchically. Ironically, it was by following the language suggested by Holacracy that we fell into Holarchy’s biggest trap: importing people hierarchy via the “lead link” role.

We found that the word “lead link” was dangerous because it contains “lead”, which gives it an instant managerial sheen that is hard to rub off. This is multiplied when you combine it with a critical mistake that we made and that it seems you made too: which is to allow a “lead link” for the company as a whole, and to place the founder (formerly CEO) there. If the CEO of the company is the lead link of the anchor circle, then the lead links of other circles are naturally going to be understood as leaders of their circle. Holacracy also gives those lead links the power to invite and uninvite people from roles, which for many translates into the power to fire people. And in comes the people management hierarchy.

In our second iteration of Holarchy, we got rid of the anchor circle lead link, removed the word “lead”, and made sure the link role was understood as administrative/representative, not managerial, by voting mostly more junior people in those positions. I should add that after this Holarchy v2, I am not the link of any circle at GrantTree, and neither is my cofounder. This is a good thing if you are committed to self-management.

This is not to say that I don’t believe Holacracy is a good system. I think it is much better than many other systems for operating companies. However, I think it is important not to be blind to its flaws, and not to fall into the trap of thinking it is the only system available for open cultures, or even the best system for open cultures. Nor it is a necessary step towards open cultures. Many people may be very enthused by the concept of open cultures and yet put off by Holacracy’s rules-based approach.

Holacracy may be the best system for Zappos, or maybe it is not. I would like to humbly suggest that you consider that this may not be the right answer for Zappos, or that if it is, it may need substantial tweaks to adapt it to the Zappos people (rather than trying to adapt the people to Holacracy), and that even if it is the right system for Zappos, it is not perfect yet and you should evolve it when you see its shortcomings, rather than be beholden to a One True Way.

This take the courage to realise that in this step of the Zappos adventure, like in every other before, you are indeed alone: there is no ultimate guidance anyone can give, not even HolacracyOne. Ultimately the decisions of what to do and how to do it are yours. Or, alternatively, once you begin to respect and trust the people in Zappos, you will find, as I did, that you are not alone. There is never any certainty about any path, but when people are truly engaged with open cultures, when they truly embrace its ideas and practices, they take up more of the load than you could imagine before you see it happen.

5. The Agenda of Open Cultures

Zappos is the highest-profile open culture experiment happening right now, but it is not the only one. We are living in a time of change where many companies are adopting the principles that eventually lead to open cultures: respect rather than authority, trust rather than fear, transparency rather than secrecy, responsibility rather than obedience, and so on.

By becoming such a high-profile case, Zappos risks becoming the defining test case for open cultures. By trying to implement Holacracy in an authoritative, top-down manner, or at least by letting it appear that this is what is happening (if that’s a mis-reading), the wrong example is being given to countless companies looking for guidance on this.

This could cause the terrible harm of teaching hesitant founders and CEOs how to implement open cultures the wrong way. And even worse, should the high profile Zappos experiment fail, it may convince many to not even give it a try.

I believe that each company should be run in whatever way is suitable to it. There is no superior or inferior way. But I do believe that I much prefer to work in and with companies that have adopted open culture practices. And as I see the positive effects they have on people, how happy they feel working in environments where they feel respected and trusted and allowed to be themselves, I cannot help but wish that there were more companies operating like this in the world.

Obviously you cannot control the world press coverage of Zappos’s Holacracy story. It is obviously of interest, and many in the press will openly want you to fail so that they can write about a great disaster, which is always more juicy for page views than an untroubled success.

What I would humbly suggest in this case is that you use your time in the spotlight to shine the light on other successful open companies (like MorningStar, Buffer, Semco, Buurtzorg, Medium, etc), so the Zappos story does not become the defining story of teal/open cultures and Holacracy.

Yours Sincerely,

Daniel Tenner
Cofounder, GrantTree

  1. A feature of Holacracy focused on the resolution of conflicts.

Nonviolent Communication

The latest addition to the GrantTree books stable is Nonviolent Communication, by Marshall B. Rosenberg. It’s spreading rapidly through the company and having a very positive effect on our communications.

For the purpose of making it spread even faster, I decided to write a brief summary of the key points of the book.

I would like to start by emphasising that the book is incredibly worth reading, and this blog post is not a substitute. Go read the book!

4 Key Parts of NVC (Nonviolent Communication)

The book breaks down communication in four key elements: Observation, Feeling, Need, Request.

The key point is that when we express ourselves in this manner, people are much more likely to hear us, and to react favourably. This is not a guarantee – people may not be ready to receive this feedback. And they may disagree with our request for their own reasons. You can’t use NVC to compel or somehow manipulate people to do what you want – in fact, quite the opposite. You might change what you want when you listen using NVC.

And this is a first important point about NVC: it’s about both giving feedback, and about receiving it. It can be used both to deliver your own feedback to someone else, to disentangle what you’re observing, feeling, needing and requesting and deliver that in a way that’s not going to get the other person upset, and to disentangle other people’s requests. This is extremely powerful stuff, and means you can use NVC even if the other person has no knowledge of it, and it will still help.

But each of these steps has a catch, and can easily be misused. So here’s a quick overview of the “catch” in each step, based on my experience.

First, how to do it wrong

Here’s a great way to apply OFNR the wrong way. We’ll break down what’s wrong with this approach and fix it during this blog post.

Let’s say I’m in a meeting with John and he’s upset by something that’s being discussed. I’m not really sure what he’s upset about, because he is getting quite loud and, from my perspective, obstructing the progress of the meeting and perhaps even intimidating other people in the meeting. I find it difficult to pay attention to what someone is saying if they’re conveying it in what I perceive to be an aggressive way. I want him to speak at a normal level because he has a very loud voice and is a big guy so it’s scary to people when he shouts, and I don’t think that’s conducive for the meeting to progress. Here’s what I might say if I totally fail to apply the OFNR framework properly:

John, you’re being confrontational. It makes me feel you’re aggressing me in this meeting. I need you to behave more peacefully so I don’t feel so unsafe. I request you to stop behaving in this way, or else I will leave the meeting.

I could claim I’ve followed OFNR there – all the parts seem present, after all. But actually, this is a total disaster, a car crash of violent communication. Let’s look at why.

Step 1: Observation

The key with the observation step is that it needs to be non-judgemental, as objective and factual as possible. By throwing the label of “confrontational” at someone, I am not observing a behaviour, I am attacking the other person’s identity as retaliation for their behaviour. In some cases, that might get them to behave the way I want, through fear, guilt, or some other negative emotion, but it does not build a healthy relationship with the other person.

In the observation step, I need to be non-judgemental, to just observe a factual behaviour without categorising it. Here’s a better way to do it:

John, I see that you’re speaking loudly and sitting back with your arms crossed.

There. Either it’s true or it’s not. Either way, it’s not a judgement of the person and unlikely to be taken as such.

Step 2: Feeling

The second part is to state how I’m feeling. The problem with the example statement is that it’s not an expression of feeling – it’s yet another accusation. “It makes me feel” is a very poor choice of words here, because it pushes the responsibility for my feelings onto the other person. “I’m not responsible for how I’m feeling – John is!”

It’s quite uncomfortable to find yourself responsible for someone else’s feelings.  Generally, if the feelings are negative, it tends to elicit a guilt response. Again, John may react in a way that I find useful, but guilt-based relationships are, well, not ideal.

The other problem with “It makes me feel you’re aggressing me” is that it’s not a statement about my feelings at all. What’s the feeling? There is no feeling stated. There is just another accusation. Perhaps I am feeling angry, or sad, or frustrated… but “you’re aggressing me” is not a feeling.

Here would be a better way to phrase this, then:

I feel sad.

There, easy, isn’t it?1

Step 3: Need

The third part is to take ownership of the feeling. We don’t feel stuff because other people make us. We feel because we have needs that are being or not being met. This can be very difficult for many people to acknowledge, particularly in cultures or societal roles that are taught that it’s bad to have needs and acknowledge them2.

Stating the need is best done with a formula that combines it with the feeling, like “I feel X because I need Y”. When I say to John “I need you to do X” I’m not actually formulating a need, but a request. When I say “I need you to do X so that I don’t feel Y” I’m compounding the earlier error of pushing responsibility for my feelings onto John. Again, John is unlikely to react positively to this.

Another way that I could get this wrong while nevertheless using the “I feel X because I need Y” formula would be to say something like:

I feel scared because I’m being attacked by you.

That again would be an accusation rather than a statement of my need. I don’t know what’s going on in John’s head. Maybe I think he’s attacking me, but he probably doesn’t see it that way. Probably, he just thinks he’s being passionate. But once I accuse him of attacking me, what will he do? Defend himself, naturally. Whatever he does, it probably won’t be to calmly listen to my request and make a measured decision about it.

Here’s a better way to phrase my feeling and need together in a way that’s more likely to solicit a good response from John:

I feel sad because I need participants in a meeting I’m in to be civil to each other so that I feel comfortable that the meeting is going well.

Is that still good enough though? Well, maybe it’s as far as we’re able to go right now, and I want to commend that it’s a much better starting point than the previous projecting version (“I need you to behave more peacefully so I don’t feel so unsafe”). But there’s still a subtle trick going on there: the need is still being projected onto others, but instead of being projected onto a specific person (which would likely seem like a directly aggressive statement) it is now being projected onto a generic category of people (which happens to include the person we’re trying to communicate with).

Can we improve it further? Maybe we can’t. Or maybe we could dig even further, towards more vulnerability, if we feel safe enough to do so. Here’s what could perhaps lie at the end of that tunnel:

I feel sad because I get scared and confrontational when people raise their voice, and I would prefer not to feel that right now.

This is exquisitely vulnerable, both because the language is verbally turning the other cheek, and because it does not make the other person wrong in any way, or imply they are responsible for our feelings, and also it is vulnerably open to the possibility that the other person might refuse to meet us in this need (and while undesirable, that has to be ok for NVC style communications to work).

I’m not going to even begin to suggest that this is easy. If identifying how you feel is hard, identifying why you feel that way, what is the need that you have that is being met or not met, is even harder. Doing that without projecting that need outwards is tough. It often takes serious digging. I find that one trap I frequently fall into is to make the need sound more objective. For example, I could have said that I need participants in a meeting to be civil to each other so that the meeting can be productive. That may be a belief that I have, but it’s not the core, personal need I’m addressing when I request John to be civil3.

This part probably takes the most practice of the four.

Step 4: Request

What could be simpler than asking someone to do something? Well, it turns out that even in this step there are ways to get it wrong. In the example, I did it in just the right way to make sure I don’t get a good outcome from the exchange. The best way to screw up the request is to make it a demand, and that’s what I did by threatening to leave if my request wasn’t met.

When we threaten people with consequences, we take away their ability to decide to help us out. If I clarify my observation of John’s behaviour, and open up by explaining how I feel about it and what needs of mine are not being met, and then request that John alter his behaviour, chances are John will agree to my request, because John is a human being and us humans tend to prefer to be helpful to each other.

And if he doesn’t, he may have good reasons to disagree. By threatening him with consequences, though, I’ve ensured that even if he would normally agree, he will not feel like it was him that agreed it was me that forced him. And because of that, this exchange will not help build up the trust between John and I.

Or, even worse, trying to force John to do something may well elicit the opposite reaction. Perhaps John will refuse to help, and continue to speak loudly, precisely because I threatened him. When we threaten people, we take responsibility away from them, and it’s not surprising that they often start behaving like children – being contrary just for the sake of it.

It is my observation through years of working in the Open Culture that is GrantTree that left to their own devices, allowed to make their own decisions, people typically prefer to be good to each other. They tend to make decisions that I would consider to be right. However, when you try and force someone, you often push them into an oppositional, confrontational mode where they do not make the best choices they can.

So here’s a better way I could have made my request to John, one that doesn’t involve a threat:

Could you please speak more softly?

Obviously tone of voice is important here too. I could convey that question with a very strict tone and body language that implies threat. Or I could ask it calmly and openly, accepting that John may well decide otherwise. It’s his decision what he does, after all. Ironically, the latter is much more likely to convince John to speak more softly than the former. Sometimes exposed weakness is a strength.


There are many more subtleties at work in NVC, and so I really do recommend that you read the book. But in the meantime, if you’re going to take away anything from this concept, please take the following steps and caveats:

  • Clarify the behaviour that you’re observing and make sure it’s not expressed in a judgemental way.
  • Express how you feel, in a way that takes ownership for your feelings and does not imply that the responsibility for your feelings lies with the other person.
  • Connect this feeling with a need that you have – a personal need, not a falsely objective ideal, and a need which is about you, not about the other person. Try to use the formula: “I feel X because I need/want/like Y”.
  • Finish with a request which does not include any threat component, accepting that the other person may disagree with your request.

Using these concepts, you can turn:

John, you’re being confrontational. It makes me feel you’re aggressing me in this meeting. I need you to behave more peacefully so I don’t feel so unsafe. I request you to stop behaving in this way, or else I will leave the meeting.


John, I see that you’re speaking loudly and sitting back with your arms crossed.I feel sad because I get scared and confrontational when people raise their voice, and I would prefer not to feel that right now. Could you please speak more softly?

The power of this shift in language is impressive when you see it at work. I hope you do get a chance t practice it and see it for yourself soon.

  1. Actually it’s extremely hard and takes a lot of practice to work out how you’re feeling. I find myself stopping and starting constantly while using OFNR, as I do the work of figuring out what it is exactly that I’m feeling. This is extremely useful work though. It can also be very uncomfortable for people who have grown up or are living in cultures or societal roles that suggest that having feelings is somehow not appropriate – for example men in western cultures.

  2. For example, western culture tends to suggest that women in traditional, 19th/early 20th century standard gendered roles should let their needs be subservient to those of their husband and family.

  3. even the use of a term like “civil” can be very loaded

© 2023 danieltenner.com

Theme by Anders NorenUp ↑