“If you are unhappy, your system is broken.”Brandon Turner (The Book on Rental Property Investing)
In my last post, I talked about Gall’s Law and the different components of systems. In this post, I’m going to discuss ideas surrounding analyzing and improving systems. There are going to be some terms I reference from my last post, so I highly recommend checking that out first.
Josh Kaufman beautifully and simply explains how to do this in his book, The Personal MBA, and I also highly recommend checking that out too. The ideas in this post are mostly from that book as well as my own personal insights.
How to Analyze Systems
When analyzing a system is difficult to know what to look for, especially when dealing with complex systems. It’s also crucial to keep in mind that we all have a bias which could get in the way of analyzing systems. Whenever possible, we need to take steps to mitigate our bias.
Here are some other things to look out for when analyzing systems:
This is an excellent first move when it comes to analyzing systems. As discussed in Gall’s Law, all complex systems arise from simpler systems, therefore every complex system is capable of being deconstructed into simpler systems. Deconstruction is simply breaking up complex systems into their interdependent parts and understanding how each of those parts works. It’s also helpful to try to identify triggers (what kicks off another system) and endpoints (what makes a system stop). Diagrams and flowcharts are also lifesavers when it comes to deconstructing systems. Recognizing if-then and when-then relationships are also invaluable.
After breaking the complex system up into it’s smaller parts, we can then break down those simple systems even further into their components. Identify inflows and outflows, stocks, interdependencies, and so on. I talk about what those components are in my last post.
Deconstruction makes understanding systems possible. Without deconstruction, we are sure to experience confusion.
How does the system collect data? What kind of data is it collecting? Measurement describes the process of data collection in a system. If we can understand the information related to the system, then we get an insight into the system itself.
Paying attention to the measurement is fantastic for dealing with absence blindness — the idea that we have a hard time seeing things that aren’t there.
Let me give an example, measuring someone’s blood glucose levels tells us if someone has too little or too much blood sugar. This number gives us tremendous insight into what is going on inside the body even though we wouldn’t be able to visibly see changes in someone’s blood sugar.
Measuring something is the first step in improvement. I wrote a post, Tracking vs. Loss Aversion, that talks about the importance of measuring ourselves and our progress.
Don’t sleep on measurements. Trust me.
KPI (Key Performance Indicator)
I know I just emphasized how important it is to measure things, but there is such a thing as too much data. If we measure too many things, we end up having a bunch of junk data that weighs us down and doesn’t show us anything. In order to prevent this, we try to keep our measurements to only KPIs or Key Performance Indicators.
KPIs are measurements of the essential parts of a system.
Identifying KPIs can be tricky, but try to limit them to only 3-5 KPIs per system otherwise, we risk measuring too many things.
Garbage In, Garbage Out
This is one of the more straightforward ideas – what we put in is what we get out. The quality of the output is only as good as the quality of the input. You get what you give. There are so many cliche phrases that express this idea.
I found this to be especially true when I started cooking. I used to watch Gordon Ramsey cooking videos to try to learn how to cook (I even watched his Masterclasses), but I could never make my food taste good until I spent the extra money on fantastic ingredients. Now, I try to only cook with fantastic ingredients. It really makes all the difference.
One of the best ways to improve the quality of a system is to pay attention to what we start with.
This can be thought of as the range in which the system is working normally. If the system is performing within that range, then it’s within tolerance.
Tolerance can either be loose or tight. A loose tolerance is when there is a considerable amount of leeway and small mistakes don’t make a huge difference while a tight tolerance is when there’s little room for error or change, this is usually the case for essential components of a system.
In order to properly analyze a system, we must acknowledge our propensity to make things look better than they are. We have to be able to apply objective judgment to our data which means that the best analysis of a system will come from someone who isn’t personally invested in it.
As I mentioned in my post, Our Unconscious Filters, human beings view the world through their bias and it’s hard to shake them, even if we’re aware of them. Having an outsider provide analysis is the only way to completely prevent our bias from contaminating our observations.
Most measurements are meaningless without context. Context is all of the information we use to understand if measurements are favorable or not. Setting goals for arbitrary numbers like a 20% increase or 3 new deals are meaningless if we don’t know the performance of the system in the past and it’s projected performance in the future.
Trying to oversimplify how a system operates by judging it off one measurement will blind us to other important changes as well. Context is crucial for an accurate understanding.
Sampling is what we do when we try things. We’re simply taking a small part of the output and using it as an example for the entire system. Sampling is great for catching errors without needing to check the entire system.
Just like all other methods of analysis, we have to consider our bias, and sampling is prone to bias. A way to control for it is to make sure the sample is random.
Margin of Error
Of course, not all samples will be perfect representations of the entire system. The margin of error is how much the sample deviates from the whole. The higher the margin, the more inaccurate the sample. The more samples we have, the smaller our margin of error.
Ratios are fractions. Somethings divided by something else. It’s a simple way of measuring multiple variables at once. Ratios are also great for letting us know how a particular measurement changes.
For example, ROI is a percentage ((Returns/Investment)*100%) or comparing MPG (miles/gallons) or unit price of groceries. You know a ratio is involved in you hear the words per.
In order to analyze a system properly, we need to know how it would operate normally. Kaufman suggests that we can measure typicality through calculating mean, median, mode, and midrange or various measurements.
Correlation & Causation
Causation comes from the idea of cause and effect. One part of a system is causing another part of a system to act. Correlation, on the other hand, is not always causation. Sometimes variables may seem the act like one causes the other, but that won’t necessarily be the case. For example, 100% of people who drink water die. Does the water cause the death in this case? Probably not. Water and death are simply correlational.
So how can we determine is something is correlational or causational?
By adjusting for known variables. If we control for as many variables as possible, we can see the relationship between each more clearly. As systems grow more complex, this becomes more and more difficult. The more we can isolate a variable, the more confidence we have that the changes are causational.
Proxies are measurements of something by measuring another thing. A proxy is useful when we cannot measure something directly. The closer the proxy is to the original, the more accurate the measurement. We have to be mindful about correlation and causation when measuring a proxy.
Segmentation is grouping data into separate subgroups to get a more comprehensive context. I do this with all of my blog posts! That’s why I have titles and headings and subheadings. It gives all the (seemingly) random information I’m spewing a more detailed context.
Segmentation plays a huge part in how we understand complex and large amounts of information.
When looking at data is easy to see it as an inanimate object, but when analyzing systems we have to keep in mind that the data tell us information about human beings. They are insights into real people — their behaviors, experiences, and thoughts. It’s easy to disconnect from data about a system because it seems so abstract and inanimate, but it’s quite the opposite. If we pay enough attention, the data lets us understand people on a deeper level.
When I worked at Kohls, they always emphasized selling to “her.” Her being a personified collection of the average data on their customers. They used average household income, gender, family size, and other variables to create their typical customer and found ways to satisfy that person.
Our data tells us what’s up with other people if we look hard enough.
Other Things to Look Out For
“If something in your business is causing you stress, most likely, you either don’t have a system for that issue, or you are not following your system.”Brandon Turner (The Book on Rental Property Investing)
Pay attention to environmental changes and selection tests. I talk a little bit about these in my last post. These changes give smaller players a chance to outperform larger players. Identifying selection tests gives us a competitive edge.
Some questions to ask while looking out for these things could be: How is the environment changing? Who is unable to adapt to these changes? What can I do differently from those who cannot adapt? Who is taking advantage of these changes? What can I do similar to those who are doing well?
Always keep an eye out for the “black swan.” I first heard about this idea from Chris Voss, an ex-FBI terrorist negotiator. The “black swan” is any information that if discovered would change everything. Back in the day, people would say swans are white and if anyone said otherwise they would be crazy because swans are white. Eventually, someone discovered a black swan and everyone had to change how they saw the situation. Systems are the same way, try to keep an eye out for the information that would change everything. There’s always a piece of information that, if known, would change everything This is excellent for accurately identifying and balancing risk and uncertainty.
It’s also helpful to keep in mind that we process the unknown the same way that we process threats. We literally see and respond to what we don’t know as a threat. Expect to encounter threats, but instead of responding to it, we can respond with curiosity to learn more.
The last thing I want to mention about analyzing systems is to analyze close-calls when they happen to minimize accidents. Sometimes shit happens, but most of the time we can prevent it from happening. If we can notice when things almost go wrong, then we can take the steps to make sure that it doesn’t happen again or prevent the conditions that allowed it to happen in the first place without having to deal with the fallout of the accident.
How to Improve Systems
“Anyone who understands systems will know immediately that optimizing parts is not a good route to system excellence. For example, let’s build the world’s greatest car by assembling the world’s greatest car parts. We connect the engine of a Ferrari, the brakes of a Porsche, the suspension of a BMW, the body of a Volvo. What we get, of course, is nothing close to a great car; we get a pile of very expensive junk.”Donald Berwick (1946 – )
Now that we’ve discussed analyzing systems and have a base framework for understanding systems, let’s talk about some of the ideas useful for improving systems. Most of these ideas are also included in Kaufman’s The Personal MBA.
This is the idea that human beings tend to add changes to a system just to feel like we have more control. So when we set out to improve a system, we have to entertain the thought that we might be implementing a new change just to feel in control. If we don’t, we risk adding unnecessary complexity to the system.
The best way to account for intervention bias is to analyze through a null hypothesis – what would happen if we did nothing? What if the situation was simply an error?
If the null hypothesis experiment determines that we’re better off doing something than nothing, then we will have minimized our chances for intervention bias to take hold. Examining the null hypothesis isn’t our natural reaction, especially since human beings have a proclivity to doing something rather than nothing, but it’s crucial for actually improving systems.
When improving systems, first think about what would happen if we did nothing.
Optimization is what people usually think of when it comes to improving systems. This typically involves maximizing output or minimizing an input. Optimization is usually focused around the KPIs.
Kaufman suggests when optimizing a system to focus on one variable at a time. Optimizing a system across multiple variables will almost always lead to disaster. System interdependencies and second-order effects make it challenging to change more than one thing at any given time.
This refers to changing a system’s process so that it can perform the exact same result but in a more efficient way. This is most obvious in coding. Some programmers will pride themselves on performing the same actions in fewer lines of code.
To the average person, refactoring may seem insignificant, but more efficient systems run faster and require fewer resources which could be redirected elsewhere.
Some questions to ask when refactoring a system could be: What are the essential processes to achieve the desired objective? Do these processes have to be completed in a certain order? What are the constraints of the system?
If you’ve heard of the Pareto Principle (a.k.a. The 80/20 Rule), then you understand the concept of the critical few. Essentially, 19th-century economist and sociologist, Vilfredo Pareto, discovered an interesting pattern when analyzing data regarding land ownership and wealth distribution.
He discovered that 80% of the land was owned by 20% of the population. Pareto didn’t just find this pattern in wealth and land distribution, he also saw it in his garden. 20% of his pea pods produced 80% of the peas.
Today, we can see his 80/20 split in almost everything. In systems is useful to know that 80% of the output is from 20% of the input. In businesses, typically 80% of revenue usually comes from 20% of customers. 80% of the work is done by 20% of the people. 80% of our time communicating is with 20% of people we know.
Focus on the 20%. That is where the biggest changes will happen. Identify which parts are critical and give it attention or starve it of attention, whichever is required. The idea is to not try to focus on the whole thing, but the smaller parts that matter.
I do this with my students. 80% of my income comes from 20% of my clients and my attention and efforts are split accordingly. I give the clients who matter more attention and I starve the ones who don’t. After practicing these methods, I’ve eliminated a lot of headache clients and I’ve strengthened my relationships with the ones I do like. 80% of the problems came from 20% of the clients.
Focus on the critical few.
This is the idea that after a certain point, adding more starts to cause more harm than good. This is common when optimizing high performing systems, people tend to try to push the system even more to the point where the system breaks.
“The last 10 percent of performance generates one-third of the cost and two-thirds of the problems.”Norman R. Augustine (Aerospace Executive & Former U.S. Under Secretary of the Army)
A way to control for diminishing returns is to apply Ramit Sethi’s infamous “85% Solution” from his fantastic book I Will Teach You To Be Rich. Simply get 85% of the problem right and move on. Yeah, we can really hone in on getting that extra 15%, but then we risk diminishing returns.
Is it worth doubling the effort just to squeeze out that extra 10-15%?
It might be, but not every time. It’s better to spend our energy getting the big wins, than trying to squeeze out every little bit.
Friction is something I pay a lot of attention to. I spend a lot of time dedicated to removing friction from my life because it stops me from doing so much. Friction is any force or process that removes energy from a system. Remove the friction, increase efficiency.
Amazon Prime is a perfect example of a company removing friction to make a system more efficient. If you have amazon prime, then you know how easy it is to purchase things. That’s intentional. The ease of use creates more cash flow for the business.
If a system has a lot of friction, it can still perform but it will require much more energy. If we don’t add more energy, then the system will eventually slow and stop.
For me personally, when I encounter friction while doing an activity that I don’t enjoy, then I won’t do it at all. So if I can help it, I try to remove as much friction as possible whenever I’m doing something difficult or something I don’t want to do.
Sometimes introducing friction is what’s needed to improve a system. When I want to prevent myself from performing certain actions, I introduce friction because I know it will stop me. Some business makes it cumbersome for a customer to return their product so they are less likely to return it.
The gold standard of no friction. Automated systems operate without human intervention. Automation is best for repetitive tasks.
Be mindful that automating a system tends to magnify the efficiencies and inefficiencies. If the system is already efficient, then automation will make it faster. If the system is not efficient, then automation will slow it down.
When it comes to understanding automation, we want to be familiar with the Paradox of Automation: the more efficient an automated system is, the more critical the human inputs are. While automate reduces the need for human intervention, the small amount of human intervention that occurs becomes increasingly significant.
Automation makes our actions count more, not less.
On that note, I also want to mention the Irony of Automation: the more reliable a system is, the less attention humans pay to it. Reliable systems train absentminded operators. This is dangerous because if something goes wrong, we aren’t likely to notice and the automation will propagate that error.
The best way to avoid automation errors is to perform consistent sampling and testing.
Standard Operating Procedure
SOPs are predetermined processes for completing certain tasks or solving common problems. We save cognitive load and cut down the number of decisions we have to make in a day if we have a preselected method that’s known to work.
Using SOP helps us spend our energy on improving a system, rather than solving repetitive problems over and over.
Kaufman recommends reviewing the SOPs every two to three months to keep it running effectively.
SOPs can look many different ways. For example, I have a set price for certain students, and certain times I will tutor. But it can go further than that, I have predetermined phrases that I say when talking to clients to make communication easier when it comes to scheduling or other common tasks. I also have predetermined methods for dealing with certain kinds of students so we have a simple system for us to start with and build upon.
Focus on creating go-to methods for things you encounter often. You’ll find that it can seem like a lot of work upfront, but it will streamline the process in the long run and it’s so worth it.
I can’t talk about checklists without referencing Dr. Atul Gawande’s The Checklist Manifesto. That book beautifully describes the power of checklists. Checklists have been a vital part of pilots take off routines and are the reason for their high success rate. Checklists have also played their role in minimizing infection rates in hospitals all over the world. The secret to repeatedly completing complex tasks perfectly is writing it down as a checklist.
Checklists are simplified SOPs for specific tasks. They’re fantastic because they create systems for processes that haven’t been articulated and minimize our chances of skipping critical steps.
I always make a checklist for my students who are struggling to “manage the chaos.” Transforming the glob of craziness that is school work into a list that can help us narrow our focus works for every single student I have ever worked with. Seriously, I haven’t come across any academic situation that a checklist could not solve.
Checklists are so critical for entropy management that I use them whenever I’m feeling overwhelmed. Whenever I’m feeling stressed and swamped with work, I ask myself “What’s the 80/20 I need to tackle here?” then I made a checklist to conquer the critical few.
I’ll probably write another post on checklists because they’re so damn powerful, important, and useful.
Checklists are also great because once we have a good one, we can delegate or automate it — which frees us up from doing the work! Checklists tend to be the first step to freedom.
This refers to the idea of stopping something intentionally. Sometimes a system may be going haywire and the best thing to do is to stop a process. As I mentioned earlier, humans have a proclivity to do something to improve a system, but sometimes the best choice may be to not do anything or stop altogether.
Cessation is not our natural reaction when we want to improve systems and it’s usually an unpopular choice when dealing with a group, but keep in mind that it’s a valid option.
When analyzing and improving systems, I entertain the idea of cessation after I’ve tried the null hypothesis. If both of those options are determined to be ineffective, then I’ll start doing something to improve the system.
The resilience of a system is determined by how much change it can withstand. The ability to weather change and adjust plans means the difference between disaster and survival.
Resilience usually comes at the price of optimal performance. A system can increase its resilience through leverage. For example, if a business needed some money to weather the storm it will be more resilient, but that money can’t be used more efficiently. Resilience comes at a cost.
Another way to boost resilience is to prepare for the unexpected. Having plans for different/unexpected scenarios or extra supplies on hand are great ways to make a system more resilient. Fail-safes and backups are great for that also. A fail-safe is a backup system designed to prevent or recover from the original system failure.
This is the process of identifying the boundaries of a system by changing the environment. Testing different extremes on a system can help determine which variables affect which processes.
When I stress test my systems, I try to break them. This is the part when we want to try and test out our “what-if” scenarios. Scenario planning is at the heart of any effective strategy. Rather than trying to predict the future, we can prepare for a handful of imagined scenarios and be ready for what comes next. A proper stress test can really help with a system’s resilience.
Sustainable Growth Cycle
This cycle is a pattern that systems follow when undergoing consistent growth without any major issues and it’s split up into three phases:
The Expansion Phase – this is when the system is focused on growing. This is a creation phase. New components and strategies are implemented and dedicated to growing the system and collecting data.
The Maintenance Phase – this is when the system focuses on executing the strategies and maintaining the functionality of the system. Pressing play on the system, so to speak.
The Consolidation Phase – this is when the system is focused on analysis. All the data that was collected is now put into context. Things that work are given more resources and attention while things that don’t are cut back or reworked.
The Middle Path
This idea comes from the fact that the balance between too much and too little are constantly changing. Balancing what systems need requires constant reevaluation. The best approach usually lies somewhere between too much and too little.
No one has everything figured out and determining the best choice when it comes to improving a system is a difficult task. This is when experimenting comes in handy. Frequent experimentation is the only way to accurately determine what improves and system and what doesn’t.
I like to treat experimenting is play. I love trying to new things, changing stuff up, and seeing what happens. The more we experiment, the more we learn about our systems.
More Methods of Improvement
“A man with a surplus can control circumstances, but a man without a surplus is controlled by them, and often has no opportunity to exercise judgment.”Harvey S. Firestone (Founder of the Firestone Tire and Rubber Company)
A personal note for improving systems – I like to have stock built up for projects with continuous deadlines like blog posts and beats. Having more stock makes me less anxious about meeting a deadline and I can focus on making good music or writing a good post. To increase stock, simply increase inflows and decrease the outflows. In this case, I increased how many beats I made in a week, but released only 1 (I usually release 2). After a while, I started to build a stock and the beats I made later were of higher quality. My end goal is to make high-quality music while enjoying the process, so I modified my system to make that happen.
In every system, there’s always a limiting reagent. Finding that constraint and removing it will improve a system’s efficiency. Israeli author, Eliyahu Goldratt, suggests using the “Five Focusing Steps” to identify and eliminate constraints:
- Identification – examining the system to find the limiting factor
- Exploitation – ensuring that the resources related to the constraint aren’t wasted
- Subordination – redesigning the entire system to support the constraint
- Elevation – permanently increasing the capacity of the constraint
- Reevaluation -after making a change reevaluating the system to see where the constraint is located
When dealing with systems that involved other parties, we introduce counterparty risk. The best way to deal with counterparty risk is to have a plan of action in the event that the other party doesn’t deliver on their end of the deal.
These ideas are foundational for analyzing and improving systems but the methods are endless and I recommend that you go out and find concepts and methods to build upon your knowledge of systems. Remember Gall’s Law, all complex systems evolve from simple systems, and these ideas are the components of creating a simple system to analyze and improve other systems. How meta.
However, the most important concept for analyzing and improving systems is understanding that we can always learn more and education never stops. Systems can be complex and there is always something more to learn about a system or systems in general