One of the immediately interesting ideas I picked up from Reinventing Organizations (and since have discovered in other places, like Joy at Work) is the “Advice Process”. This is a decision model that works very well to replace both the top-down decision model and the consensus-based decision model, both of which have serious flaws in practice and theory both.

So far, we’ve been using the Advice Process at GrantTree for about 3 months now (since August 2014, in case you’re reading this much later), and it’s worked well (with some hiccups around decisions about pay – more on this further down). We’ve learned a few things about how to use it or fail to use it properly, and I thought I’d share these thoughts, three months in. First, though, a bit of background…

How do you end up with the Advice Process?

The advice process is a natural progression from the other two obvious decision models in startups. The default is the top-down model. It’s the dominant decision model in business today and most startups will default towards this unless the subject is given some thought. It can certainly work, and does in fact work for many large companies, but I believe it is inferior to both the consensus model and the advice model.

In the top-down model, decisions are ultimately made at the top of the company (e.g. by the founders and chief executives). Often, a single person has ultimate power to make every decision, and uses it regularly. By default, every important decision must rise to the top. Some less important decisions are delegated, but generally people avoid making decisions without their boss’s approval because they frequently don’t have enough information to make those decisions. This is very typical of closed cultures where secrecy is the norm – and is in fact the only decision model that can work there, since the lack of information sharing makes it a self-fulfilling prophecy that the lower strata simply cannot make the big, important decisions.

Some people are happy with that. Others really hate it. Some, like Tony Hsieh, will rather sell their company than continue working there once they’ve built this sort of culture1. The natural reaction is to say “rather than concentrate all the decision making at the top, let’s distribute decision power; everyone should have a say!” Enter the consensus model.

In the consensus model, everyone agrees before a decision is made. I believe this model is superior to the top-down model, for the simple reason that it engages everyone. It may be less efficient at making decisions, but as part of a culture of involving everyone in running the company (which requires a lot of transparency) it results in far more motivated people and therefore, in my opinion, it is better than the top-down process despite its flaws. What are its flaws? Well, it’s very slow, first of all – getting everyone to agree takes time. It’s also prone to politics – people spending energy building support for their proposals rather than getting things done.

One way to see the consensus model’s failure is that instead of giving everyone the power to make decisions (its intended aim) it gives everyone the power to stop everyone else from making decisions. Ouch.

Enter the advice process.

The Advice Process

Here is the way we have initially defined the advice process within GrantTree:

Anyone can make any decision they feel comfortable making. However, before they make that decision, they must ask those who will be impacted by that decision, and those who are experts on that subject, for advice. They are free to disregard the advice and make the decision the way they wanted to anyway, but they must first ask for advice.

Important: this is about getting feedback/input into your decision, not about building consensus. Do not use the advice process to try and browbeat people into agreement or to build political support for your decision. You don’t need people to agree. You don’t need political support. You just need input to make sure that you make the right decision.

It seems very obvious and elegant in retrospect, and it’s no surprise that many advanced open cultures have evolved similar decision processes in parallel. Lucky for us building companies in 2014, we don’t have to reinvent the wheel: we can just read books and blog posts about it and copy the best ideas!

The core is so simple: if you want to empower people to make decisions, let them make decisions. Boom.

However, in practice, there can be a few snags in the way you implement the advice process. I’ve noticed some, and they are outlined below. Hopefully they’ll help some people.

The right reasons

First of all, in order to work, the advice process must be brought in for the right reasons: i.e., it must be about enabling everyone in the company to make decisions. I’ve observed a case where a partner in a small firm brought in the advice process in order to justify making a decision which everyone else disagreed with. That’s not going to work.

Bring the advice process in to enable people who are already working well together to do so faster. If the consensus process is failing you because there are too many disagreements within the company, the advice process will only cause people to get even more pissed off at each other. Think of the advice process as a speed boost. If you were heading for a wall, you will only smash into it faster.

Advice process for compensation

We trialled this, and whilst I fundamentally believe it can work2, there are a few important caveats which are particularly relevant to startups.

The advice process risks surfacing conflict when a highly contentious issue is being decided, one where there is very strong, irreconciliable disagreement. For example, if people have taken comparatively low pay because they were joining a startup, and they feel they are significantly underpaid, then allowing them to change their pay via the advice process will set them on a collision course with others who disagree that the company is at a stage where it can afford to pay a market rate, there will be sparks. In theory, the advice process should be able to resolve this… but it is a bit of a test of fire. And most companies don’t have the ability to resolve that level of conflict.

I would recommend not applying the advice process to compensation until it is generally accepted that most people are paid mostly the right amount. At the very least, if you’re going to open that up to the advice process, do it a good many months into the process, at a point where both the advice process and the conflict resolution process (to be discussed in another article) are well bedded in, so this one conflict doesn’t sink the entire advice process endeavour.

Trusting people

As a founder, the hardest thing about the advice process is actually in clearly resolving, within yourself, to actually trust the people you’re delegating that power to.

People can feel if you’re only paying lip service to the advice process, doing it to “empower” people but not actually trusting them with important decisions. When you don’t really trust people, you will end up verifying and vetoing and changing their decisions after the fact. This creates a self-fulfilling prophecy (a perverse version of the Pygmalion Effect). Because people know they’re not really making the decision, they don’t pay attention to it all that much. They’re playing with a safety net, and that safety net is you, the founder. So they make whatever decision they feel like, without thinking too much about it, and let you take responsibility for which decision actually gets implemented.

Needless to say, that doesn’t work. The magic of the advice process is that the person making the decision is entirely responsible for the decision and its outcome. This concentrates the mind, and forces them to really seek out advice and pay attention to it, because ultimately it’s their decision and the company (which hopefully includes people they like to get on with) will have to live with the consequences of that decision.

When people make a decision via the advice process, let them actually make the decision. Trust them. That’s certainly the hardest part of the process as a founder.

Conflicts

Sometimes, nevertheless, there will be decisions that lead to a conflict. To resolve this, you need a well defined conflict resolution process. I’ll cover that in more detail in another post, but here’s an essential point to make here: let the conflict happen and resolve itself like any other conflict. In particular, don’t try and resolve the conflict ahead of time by undermining the decision process before the decision has even been made. Otherwise, you’re once again taking away responsibility from the person making the decision.

I made an exception at GrantTree for a salary decision conflict that seemed unresolvable – both sides were completely unwilling to budge and, knowing the people involved, I knew they were stubborn enough to drag it out all the way. I’m not yet sure suspending the advice process for compensation was the right decision (time will tell), but what I am sure is that if I suspend the advice process again, people will get the impression that the advice process is only there for unimportant decisions – and that will undermine it, lead to a self-fulfilling prophecy, etc.

Readiness and requirements

Finally, not every company is ready for this process. In particular, the advice process requires:

  • A team that has tried both the top-down and the consensus process and decided that they weren’t good enough, and that is therefore able to appreciate the evolution in the model;3
  • A company culture that is largely transparent – people cannot make good decisions without information, and if the information is locked down, the advice process will naturally be impossible;
  • Founders who are willing to take the big step of trusting people, letting go of control enough to actually implement this process whole-heartedly.

I hope some people who are currently mired in the consensus process read this and find the solution they were looking for. If you have any questions about how that might work, feel free to ask them (email, twitter or otherwise). I’ll do my best to answer.


  1. See the LinkExchange story in Hsieh’s Delivering Happiness.

  2. And, according to Dennis Bakke, it has actually worked in some parts of his company, AES back in the nineties.

  3. Can you skip directly from top-down to advice? Maybe… let me know if you’ve done that! Intuitively, though, I think the evolution might be necessary.