Being an entrepreneur is about building a company. Fast-growing companies may seem chaotic and frequently focus more on firefighting problems rather than on actually building up the company. 

“The real enemy of execution is your day job!” McChesney et al. argue1. “We call it the whirlwind. It’s the massive amount of energy that’s necessary just to keep your operation going on a day-to-day basis; and, ironically, it’s also the thing that makes it so hard to execute anything new. The whirlwind robs from you the focus required to move your team forward.”

Building a company at break-neck speed will always produce its own whirlwind. The challenge is that an hour spent on fighting fires is an hour that was not spent on building the business. As entrepreneurs, we, therefore, must distinguish between when we are building the business and when we are operating the business. 

The business as a collection of machines

In order to avoid the whirlwind (or Blizzard as he calls it), legendary investor, Ray Dalio2, recommends viewing the business as a collection of machines that are each based on logical cause-effect relationships. These relationships allow the designer of the machine to come up with rules for how to manage it which ensures that the outcome is predictable, repeatable, and can be performed by multiple people.

“Most people get caught up in the blizzard of things coming at them,” Dalio explains. “In contrast, successful people get above the blizzard so they can see the causes and effects at play. This higher-level perspective allows them to see themselves and others objectively as a machine, to understand who can and cannot do what well, and how everyone can fit together in a way that will produce the best outcomes.”

Designer versus operator

“If your business depends on you, you don’t own a business – you have a job.
And it’s the worst job in the world because you’re working for a lunatic!”
– Michael E. Gerber. The E-Myth Revisited 3

If the entrepreneur is required to be involved in the operations of each machine, she will be the bottleneck for the number of machines in the company and consequently for the scale of the business. She must free up her time and make each machine in the company independent of herself. 

To do so, she must first carefully recognize when she is working on the business (designing or improving machines) and when she is working in the business (operating the machines).

“Most people remain stuck in the perspective of being a worker within the machine,” Dalio argues. ”If you can recognize the differences between [when you design the machine and when you operate it] and that it is much more important that you are a good designer/manager of your life than a good worker in it, you will be on the right path.” 

It is in building machines that the entrepreneur adds more layers to the business. So how does one design and build a good machine?

Designing the machine

As a company scales, the knowledge that remains tacit (i.e. inside the head of the person doing the task) will eventually become a bottleneck. If that person leaves or if you need other people to execute the same task, it will no longer be possible to do so to the same standard. It also has the unfortunate effect that the person doing the task can never be promoted or moved to another team as the process depends on her (you can read more about that here). Building machines is a way to make tacit knowledge explicit and enables you to improve on it.

To my experience, designing a good machine consists of six phases:

  1. Design a winning process
  2. Document the process
  3. Create transparency on performance (and make it auditable)
  4. Hire and train an operator
  5. Audit the process
  6. Optimize the machine continuously

Step 1: Design a winning process

The most common reason for not building a robust process is that people either 1) are caught up in the operations or 2) try designing the perfect machine in the first go. The challenge is that you’ll never have the perfect process. Most processes can always be improved further. Accept that you won’t know how the perfect machine works without having tried it. Design the best you can and work from there.

The initial process

In order to design the initial process, immerse yourself – for the time required – into the project. Depending on the complexity of the process, you should be able to design an initial version in 15-30 minutes.

It’s important to include:

  • Objective – What are you trying to achieve and how will you know/measure whether you’ve achieved it (KPI)?
  • Process – The steps required to replicate the outcome. Here is an example from McDonald’s that illustrates the required detail

A machine for an email newsletter could for example include:

  • The goal 
    • E.g. people clicking through to read the full article on the website
  • The process for e.g. how to
    • Segment the subscribers
    • Design the newsletter to be consistent with the brand
    • Pick the right subject line
    • Write the content in the right style
    • Send at the right time
    • Split test variations
    • Measure and evaluate performance

In building a machine or company, Dalio sets out four ways to ensure you get the best machines:

  1. Seek out the smartest people who disagree with you and try to understand their reasoning
  2. Know when not to have an opinion
  3. Develop, test, and systemize timeless and universal principles
  4. Balance risks in ways that keep the big upside while reducing the downside

Stressing point 1, the easiest way to improve your process is to get help designing the process from the most believable and knowledgeable people in the field – both people who may agree or disagree with you.

As Harvard epidemiologist, Marc Lipsitch, explained in relation to understanding the Coronavirus 4, we must distinguish between three levels of information:

  1. What we know is true; 
  2. What we think is true—fact-based assessments that also depend on inference, extrapolation or educated interpretation of facts that reflect an individual’s view of what is most likely to be going on; and 
  3. Opinions and speculation.

While most entrepreneurs aren’t battling global pandemics, there is a key takeaway for designing a winning process: Even though no one knows the perfect way to do something (information that is true), some people will have a more qualified idea (what we think is true) based on their experience with and understanding of adjacent fields. Don’t let this hold you back. Don’t seek to get input from hundreds of people. Find a few critical people who can give you input and move forward.

Getting started

Getting started can often be daunting. The best way is to start by mapping out the different parts of your machine. If you are designing a machine for email marketing, for example, you may start by mapping out the components:

  • Database management and segmentation
  • Sending schedule and frequency
  • Email template design
  • Testing methodology
  • Performance dashboard

Once you have the high-level breakdown, you can start dividing each of them into subparts and define more granularly.

Step 2: Document the process

Imagine building a bakery. You would never let the baker play it by ear every time he would bake a new batch of bread. Even if you successfully did so, then firstly, you would be completely dependent on your current baker and unable to scale the size of the team as the next baker would start over in developing his recipe.

Secondly, if the recipe (knowledge) only existed in the head of the baker (-s), it would be incredibly hard to improve the result in the future. If e.g. the bread turned out too salty, how could you possibly improve on this without knowing how much salt was used in the first place? Without this knowledge, you are starting from scratch every time.

In other words, for a predictable outcome, you must standardize the way you bake bread. Author Michael E. Gerber 3 calls this orchestration: “Orchestration is the elimination of discretion, or choice, at the operating level of your business.”

This is not to say that the process is now set in stone and will never evolve, but more about this in step 6.

A word of caution: What may seem simple at first will grow complex over time. The complexity only scales with the number of tasks, so try to keep things as simple as possible so you don’t drown in complexity.

How to document the process

When documenting the process, don’t overdo it. I have seen too many entrepreneurs writing 10 or 20-page documents describing something that could be described on a one-pager. Keep this as simple as possible. Unlike this very elaborate guide, a list of bullet points will often do well at first. Once you have the list, get input to ensure it’s sufficiently detailed to be self-explanatory.

A common problem is that people start the documentation too late. They start when everything is up and running and therefore forget some of the parts involved in reaching the success or they become blind to some of the critical actions they are taking.

Step 3: Create transparency on performance (and make it auditable)

In order to do well repeatedly, we must start with the end: What can we measure to know whether we are doing well?

Depending on your machine, the metrics you need to monitor vary greatly. As for all KPI-setting keep this simple and focused on the essential metrics. It’s easy to overdo this and end up with too many KPIs. If everything is important, nothing is important.

You would often want to include both leading and lagging indicators in your dashboard. Leading indicators predict future performance. An example can be the number of sales meetings for a sales rep. While you cannot take that to the bank, there should be a correlation to future deals. 

Lagging indicators are backwards-looking and show how well you did. In my experience, lagging indicators can often be boiled down to one or two metrics. For performance marketing for example, I find gross profit to be an excellent metric as it encompasses both the volume aspect of marketing and the cost efficiency aspect. It’s important to include both aspects as most jobs entail balancing efficiency and scale, or other similarly opposing metrics. 

Step 4: Hire and train an operator

Entrepreneur-turned-venture-capitalist, Ben Horowitz 5 emphasises the importance of training with a perfect example: “I learned about why startups should train their people when I worked at Netscape. People at McDonald’s get trained for their positions, but people with far more complicated jobs don’t. It makes no sense. Would you want to stand on the line of the untrained person at McDonald’s? Would you want to use the software written by the engineer who was never told how the rest of the code worked? A lot of companies think their employees are so smart that they require no training. That’s silly.”

If we buy in to the idea that orchestration is essential (Step 1), then it must follow that we train our people to follow the process. People often mistake training for not utilizing people’s smarts. Contrarily, I see it as a way to speed up the process of value creation and the learning.

Take Google Adwords for example. If we don’t have a clear process for how the account should be optimized and expanded or don’t train people on that process, then we leave it to the employee to figure it out on their own. What are the odds that the one person you have hired finds the best solution for this task? 

As we will see in step 6, processes are not static and will evolve, but why not start from the best possible foundation rather than with a clean sheet of paper? Too many people mistakenly focus on “innovation and creativity” in the design of their initial processes. Rather, start with the best and most proven model as a baseline and innovate from there. Without a baseline, how could you know if your innovation is actually performing better? The importance of this can hardly be overstated.

By convention, even inside of our group of companies, only one solution can be the best for any given problem. In other words, if everyone else copied the process of that one task, they would immediately achieve better results. 

Processes + Training = Consistency + Repeatability

Step 5: Audit the process

Once you’ve got a solid machine and have trained people to operate it well, you must find ways to ensure that the machine runs smoothly and according to the recipe (if it doesn’t and the output is better, you should update the recipe).

My favourite example of this is peer review in coding where senior developers will look through the code written by other developers before it makes it to the final product. The principle can though be applied to even the most basic tasks (at McDonald’s, employees must mark on a sheet when they wash their hands so supervisors can see that they live up to hygiene standards).

A common mistake in auditing is to look at the operational output only. It’s easy to only look at the results, but you risk missing key insights that way. If you truly believe in your process, then people who don’t follow the recipe must have a big potential for improvement. 

By accepting output as the only metric, you may simply be setting the performance bar too low and thereby accepting performance below what should be possible.

Step 6: Optimize the machine continuously

The key to building a machine is to have a strong process. Without a process, there is no machine. At the same time, a common fallacy is a belief that the “perfect” process is just around the corner and therefore not defining the process now. “A little more experience and we will do so much better” is often the excuse. We may indeed learn more next week, but we shouldn’t let that hold us back when building a company. Define the process and improve it as you learn.

Optimization should not happen by chance. Improvement should not be organic. With orchestration we must take a scientific hypothesis-based approach to optimize our machine. What this means is that we should make changes only after carefully thinking about a) what are we trying to improve? b) what can we do to improve? c) why would this change improve the process?

Putting it all together

The opportunity cost of not having great machines is that you won’t be able to free up yourself. In my experience, machine-building skills is the single biggest predictor of success for entrepreneurs and leaders. The ability to ensure great performance while making yourself obsolete makes you promotable and allows you to free up time for other projects.

A few lessons before you jump into building your own machines:

  • Don’t build multiple machines simultaneously. Build one, make it work, move on
  • Train people to understand and differentiate between when they are operating and when they are building the machine
  • As you become more proficient, seek to spend an increasingly larger share of your time on designing, building, auditing, and optimizing machines rather than operating them
  • Trust the process. If your machines don’t produce the right outcome, the problem is with one of the six steps above, not with the idea of making yourself obsolete from the operations

The machines become a pipeline for talent in its own right. Operators learn to optimize the machine. Once they can optimize, it’s easier to learn how to design machines from scratch.

Build a company by designing and building great machines!


  1. McChesney et. al. The 4 Disciplines of Execution
  2. Dalio, Ray. Principles: Life and Work
  3. Michael E. Gerber. The E-Myth Revisited
  5. Horowitz, Ben. The Hard Thing About Hard Things

Image credits: and