The Philosophy of Business Analysis, Part I: Epistemology
I have an interesting pair of careers. For the past decade or so, I’ve been working in digital marketing and analytics. Simultaneously, I earned a Bachelor’s degree in philosophy and I’m about to start work on a Master’s. Now, that may seem like a diverse resume, but these two areas aren’t so different. They’re both concerned with understanding the truth of things. They’re both governed by clear, rational and precise thought. They both require you to see the big picture when your attention is focused on the tactical and technical aspects of what you’re doing.
This series of posts is going to look at business analysis from a meta-level, analyzing the process and delivery of analysis itself. I’m going to cover the five traditional areas of philosophy, which are epistemology, metaphysics, ethics, aesthetics and logic, giving a bit of explanation of each and where our profession stands. It should be a fun and wholly geeky adventure.
The Epistemology of Business Analysis
Epistemology is the area of philosophy concerned with knowledge. Its fundamental questions are as follows: What counts as knowledge? How can we acquire knowledge? Can we trust our knowledge? Closely related to epistemology is the philosophy of science, which covers a breadth of topics related to the acquisition of scientific knowledge.
As analysts, we like to believe that what we’re doing is scientific. Rightly so, I think. We’re concerned with understanding why the data we get is the data we get, and what data we will get next based on certain actions. When we’re doing things right, we follow a scientific methodology that involves testing our hypotheses and constantly revising our theory. All of this ties in closely with questions of epistemology.
Business Analysis as a Science
My remarks about science above echo those of 20th-century philosopher W. V. Quine:
From impacts on our sensory surfaces, we in our collective and cumulative creativity down the generations have projected our systematic theory of the external world. Our system is proving successful in predicting subsequent sensory input. How have we done it?
Quine’s question is an important one. Somehow, we as conscious beings, have been able to take our basic perceptions and turn them into a view of the world. We believe all kinds of things about reality beyond what we’re perceiving at any given time, and we use that information to predict what’s going to happen next. We’re pretty good at it, too.
Such is the case with business analysis: we collect data from all areas of a business, and with it we develop a theory about what’s going on with a business. We take the data we get, which come in the form of raw numbers, and interpret that to form propositional beliefs. We believe things like ‘the website is selling more goods internationally this month’, ‘the inventory can’t keep up with demand’ and ‘we need more salespeople’. We use these beliefs to predict what will happen if we put more resources into international marketing, order more inventory or hire more salespeople. We then watch the results and decide whether or not we were right.
Knowledge is thought, by many philosophers, to be a specific kind of belief; specifically, it is belief that is both true and that one is justified in believing. Much has been said and written about what exactly ‘truth’ means, and how we should cash out ‘justification’. But the idea boils down to this: we count ourselves as possessing knowledge when we believe something that is true, and we have discovered the truth of the matter through some reliable method.
In business analysis, does this ever happen? I’m comfortable saying that it does, at least with a loose definition of ‘justification’, but I think a fair bit of what we count as knowledge doesn’t quite fit the bill.
Prove Yourself Wrong
The philosopher Karl Popper made an important contribution to the philosophy of science. Popper wrote that in order for a statement (like a prediction or an assertion about a business) to be scientific in nature, it must be falsifiable. That is, we must have a clear understanding of what situations, or what data, would prove our statement to be false. For example, if I were to claim that a business will fail if we don’t invest in more salespeople, I know that I can be proven wrong if the business doesn’t fail despite a lack of such an investment. So, my statement is a valid scientific claim. If I were to claim that there are invisible, undetectable unicorns that live in our socks, that’s not a scientific statement — you can’t prove that there isn’t something undetectable if it’s meant to be undetectable.
For this reason, it’s of paramount importance that analysts never fear being wrong about something. An analyst must always be honest, of course, and I’ll touch on honesty when I talk about ethics. But in general, an analyst should welcome the case in which he or she is proved to be wrong in some prediction. After all, learning that you are wrong is learning something new about the business, and that’s a tremendously valuable thing. It makes your next predictions better. Like a good scientist or philosopher, a good analyst is always trying to prove his or herself wrong — that’s the scientific attitude that makes for good progress. Chances are, if your analyst is always right in every prediction they make, your analyst is either lying to you, or your analyst isn’t doing anything that’s worth doing.
The Limits of Business Intelligence
Epistemic concerns have always been a major part of philosophy. As far back as Plato, the question of how we, as humans, can grasp the truth have been central to philosophical questions. And in a science like business analysis, the question of whether or not we really know the things we think we do is central.
Let’s take a moment to think about the sources of our knowledge. Whether we’re pulling data from business intelligence tools, CRM, web analytics software or an order management system, we generally depend these days on one kind of technology or another. We also depend on processes of gathering and storing information, processes which are carried out by people.
Consider the list of things that could go wrong: technology can have bugs, databases can be corrupted, processes can be ignored, overlooked or improperly followed. Human beings are prone to errors which may come in by data entry, along with errors of interpretation and analysis, and so forth. To be honest, the idea that we’re getting a perfectly accurate picture of a business at any time is a little naive.
That’s not to say that business data isn’t valuable. But rather than thinking of data as either right or wrong, perhaps it’s better to think of data as having a resolution. In the same way that your high-definition television produces a much clearer image than a standard-definition one, better technology, processes and people bring the truth about your business closer into focus. Errors in the image, or the data, mean you never quite get to the level of reality. But that’s fine. Good movies look awesome in HD; your business becomes clearer with the right tools.
No Laws of Business
Physics is one of the most successful of the sciences. It has been extremely powerful in its ability to predict the future, in terms of what will happen when we control things. We can predict where in space the Earth will be a year from now, two years from now, ten years from now. We have predictions of what miniscule particles will do if we give them a jolt of energy. We know how much power we’ll get from splitting an atom. All of these things are possible because science is able to codify its findings into laws, such as laws of motion and thermodynamics. Those laws are in turn applied to future events, and boom! — predictive power.
Business analysis is an imperfect science. It falls into a group of sciences that follow scientific methods, but don’t quite have the predictive power of physics and chemistry. Sciences like this include psychology, social science, political science and economics — basically, anything that has to do with people. The reason is that we don’t really understand people at the level which we understand more basic physical processes. We haven’t quite figured out how consciousness works. We can’t quite predict what a person will do in a given circumstance with accuracy — there are always outliers, anomalies, rebels.
Business analysis isn’t about discovering laws of business, it’s about evaluating trends in business. While physics tells us how things work and will always work, business analysis tells us how things worked up until now. We can’t trust that things will always be the same. Recent economic troubles are a testament to that.
Consider this: every now and then, an email services provider will post some study that they’ve done on open rates and clickthrough rates, and announce some ludicrous claim like, “Fridays are the best day to send emails”. Now, it may be the case that they’ve seen the best open rates and clickthrough rates on Fridays. But that’s not a law of email marketing, that’s a finding for a specific set of emails. Every week of the year is different. Every company that sends emails is different. Every individual email is different. Never blindly rely on a rule-of-thumb like that — continuously try other things and measure which does best, then repeat. The same thing goes for all so-called ‘best practices’.
And that’s where business analysis has its grey areas. The subject of business analysis is business, but business ultimately boils down to human interactions. Businesspeople interact with consumers or other businesspeople. People interact with websites and other technologies. The technologies and the ideal processes are easy to understand, but people make it messy. World events, economics, social attitudes, psychological oddities, holidays and other variables each play a major role in how consumers of any kind make decisions and buy products and services. Even an analyst equipped with a calendar, a newspaper and a master’s degree in psychology will have trouble connecting all the dots. Ultimately, there are just too many variables to get a perfect image. The data isn’t perfect, so the predictions will be imperfect.
It Still Beats Intuition
It might sound like I’m trashing analytics. After all, our data is imperfect, and even if it were not, it wouldn’t be able to perfectly predict. If all this is true, why bother?
The answer goes back to what I said about justification when it comes to knowledge. Some ways of believing are better than others — if we have good reasons for a belief, and that belief is true, we can say that we know it. I want to argue that business intelligence and analysis is the best thing we have to predict future events and determine the best course of action for a business. Specifically, I believe that analysis trumps intuition.
Here are some things we’ve discovered that at some point were counter to our intuitions of the time:
At some point, all of these things were thought to be false, even paradoxical. But, through careful thought and analysis by those who did not merely accept their intuitions, we learned something new.
Don’t get me wrong — intuition can be a valuable thing. The problem is that it’s not reliable. Even analysts often use their intuition to guide their analysis, but this isn’t something a new analyst should do. Rather, an analyst’s intuition gets better the more ‘scientific’ they are in their methods. Eventually, you see patterns often enough that they occur to you before you run your tests. My argument is merely that you don’t know that they’re right until you perform the test and prove it. Intuition is a useful tool for guiding analysis; it is not a substitute for analysis.
Image credit: Jacob Bøtter
Doubt, but Act Anyway
The ancient Greek philosopher Socrates famously said that his wisdom came from awareness of his own ignorance. Doubt is a wise thing to have, and it’s an analyst’s job to be doubtful — doubtful of what they’re told by executives, doubtful of their own intuitions, doubtful about what’s best for a business and doubtful of what industry ‘experts’ call best practices. An analyst’s job is to test everything and only believe what the data confirms, with a healthy dash of doubt about the data, too.
That’s not to say that an analyst should not act. Things need to get done, and money can’t be made by quivering in a corner, overwhelmed by doubt. Write reports and write them confidently, but also write them accurately. Write only what you can reasonably be sure of. Write down your intuitions, but label them as such. Be honest about the margins of error in your analysis. A good analyst delivers competent, honest assessments of the areas of business they analyze, always knowing that each report could be improved.
There are always ways to lower the doubt, to increase the resolution of the data, to verify that processes were followed in capturing that data, and so forth. But intellectual honesty is what separates a good analyst from a great analyst. A great analyst uses the tools of science to provide business executives with the justification component of knowledge. By becoming an authority on what is and can be known about a business, the analyst becomes a source of truth.
This is just an introduction to what could be said about business analysis from an epistemic standpoint, but hopefully it highlights a few key themes and gets you thinking. Next up in this series is metaphysics, which deals with questions of being. That’s a lofty subject, but I’ll try to bring it down to Earth as best I can.
Until then, cheers!