Playing To Win
Heuristics, Management & Strategy
I have been asked a lot of questions about heuristics in management generally and strategy more specifically. I think the topic is worth delving into. So, my 45th Year III Playing to Win/Practitioner Insights piece is on: Heuristics, Management & Strategy: Exploiting the Knowledge Funnel. You can find the previous 155 PTW/PI here.
There are many definitions of heuristic, but the one I use is an approach to problem solving that aids progress toward a solution but does not guarantee the solution will be as desired. In business, a heuristic would be a way of doing an M&A transaction, or a way of onboarding a new board member, or in my field, an approach to creating strategy.
Heuristic was a relatively obscure word that seemed to have entered vocabularies in the 19th century but didn’t gain much usage until the great Nobel Laureate Herb Simon used it in his groundbreaking work on “bounded rationality” beginning in the 1950s. Simon noted that people are not entirely rational, but they aren’t entirely irrational either. Humans use heuristics that are “boundedly rational” to guide their thinking.
Starting in the 1970s, Daniel Kahneman and Amos Tversky picked up the Simon trail, exploring heuristics and, among other things, helping to birth the academic field of behavioral economics. For a considerable time, the term stayed largely within the academic realm but burst out in force when Kahneman was awarded a Nobel Prize in 2002 (Tversky had since passed away and the Nobel isn’t awarded posthumously) and then truly hit the mainstream with Kahneman’s bestselling Thinking, Fast and Slow in 2013.
The biggest takeaway from the Kahneman and Tversky body of work has been cognitive biases. That is, heuristics that incorporate or are laden with a bias, such as confirmation bias, recency bias, availability bias, loss aversion bias and so on. One can easily find lists many pages in length of all the heuristic biases that have been identified out of the work that they started.
I have no quarrel with Kahneman and Tversky and in fact find their work to be very valuable. However, I am more partial to the work on heuristics of Gerd Gigerenzer, whose most famous book is Gut Feelings. His work doesn’t focus on heuristics as bias-laden but rather heuristics as helpful shortcuts. For example, the recognition heuristic. If you are asked to choose between something you have knowledge about and something you haven’t, it is a pretty good idea to choose the former. Is it a perfect idea? No. But it is better than flipping a coin.
My Take on Heuristics
I take a somewhat different lens on heuristics, a temporal one. I view heuristics as a temporal stage we go through. Sometimes we are in the heuristic stage for a long time, while other times we are just passing through.
I first drew the above image for a speech to an audience of designers in 2003 to put into context what designers do. I argued that everything we know in the world goes through a progression. Things start out as a mystery, which is characterized by us not knowing how to think about the thing in question. Consider the time in history when we didn’t understand why most objects fall to the earth — rocks, snow, etc. Some do so slowly, like a feather. But birds don’t. Why? In ancient times there were many theories — animal spirits; the desire for all things to be near mother earth; etc.
When something is a mystery and we don’t yet have a productive way of thinking about it, we have to consider everything. In the very early days of the new disease that was killing otherwise healthy young men, it was a mystery. Researchers had to consider whether it was a reaction to a certain class of drug, result of a weakened immune system, complications from a form of leukemia, a reaction to some aspect of the environment, an inherited trait, a virus, etc.
In each of the above cases, with enough work, we came up with a way of thinking about it that helped get us toward a useful answer — that is, a heuristic. Sir Isaac Newton figured out that there was a universal force — gravity — that pushes everything towards the earth’s surface. And in due course, researchers determined that the mystery disease was an acquired immunodeficiency syndrome (AIDS) caused by the human immunodeficiency virus (HIV).
Of course, not all mysteries get transformed into heuristics. Some stay mysteries — how to stop nations from attacking one another; how to cure ALS, what are the origins of Stonehenge, etc.
But for the topics we advance to heuristics, there is huge benefit. The biggest value is that we can stop thinking about all those features that we now know are irrelevant, which enables us to focus more intensively on a smaller number of features. Plus, we now have an idea of how those features work — e.g., how HIV causes AIDS. As with heuristics in general, it helps us get closer to the answer we want.
But a heuristic doesn’t consistently produce the desired answer. For example, we figured out that unprotected sex with someone not in a long-term stable relationship was a bad idea. But was that a perfect solution? No, we later found out that we needed to be careful about blood transfusions, too. That is why heuristic shortcuts are found be as imperfect as they turn out to be. But a heuristic is a huge advance in knowledge over a mystery.
In due course, some heuristics advance to algorithms: a formula for getting the desired answer. With experimentation, we were able to establish that gravity will cause objects to accelerate downward at 32 ft/s2, so we can calculate how fast an object will fall and its path of descent. For HIV, we created anti-retroviral drug combinations that reliably keep HIV from turning into AIDS.
Starting about 70 years ago, a fourth stage was added to the Knowledge Funnel because with the advent of digital computing, when knowledge was advanced to the algorithm stage, it could be coded and a computer could run the algorithm perfectly, quickly, and cheaply. For example, Honeywell figured out how to use the algorithms for gravity to create software that autopiloted planes safely to the ground.
In 2009, I published a book about the Knowledge Funnel called The Design of Business. The graphic advice I got was to keep it to three steps — because people like things in threes — and make it vertical because of the funnel metaphor. So, in the book, it looked like this.
I have always wondered whether that simplification was a good or bad idea. But it has gone on to be the second most cited publication in Design Thinking — which makes me happy.
Implications for Strategy
I think of heuristics like I think about teenagers. They are not good or bad. They are just passing through a necessary stage. In business, operating with guidance from a heuristic is more favorable than operating in a mystery. Working on a mystery tends to be expensive, because it is unclear to what to even pay attention. That is why that much mystery exploration is done in not-for-profit institutions, like universities and government research labs. Big pharma has figured this out and that is why more and more of them are cozying up to universities who can get governments and philanthropies to fund their exploration of mysteries, which big pharma attempts to commercialize when the non-profit develops a heuristic. In general, big companies struggle to be efficient in exploring mysteries.
However, there is huge value in the act of turning a mystery into a heuristic — from the idea of a printing press to the idea of electricity to the idea of telephony to the idea of a semiconductor to figuring out how Americans wanted to eat in the post-war, car & freeway environment in California — which the McDonald brothers determined was a restaurant format that came to be called the quick-service restaurant (QSR).
But strategically, there is even bigger value to taking the heuristic and driving it to an algorithm. Ray Kroc bought a four-store chain from the McDonald brothers and drove it to an algorithm, with components including the 57-step process for cooking a hamburger. That made McDonald’s a global QSR giant and Kroc a billionaire. Intel turned semiconductor production into an algorithm that enabled it to stamp out identical fabs across the world and dominate an industry.
But the biggest business phenomenon of the modern economy is in driving a heuristic to an algorithm so that it can be coded and then scaled immensely. Dan Brickman recognized that while a spreadsheet was then (in 1979) treated like a heuristic — a way of portraying, say, financial line items by year — with each cell hand-crafted, it was actually largely a set of algorithms, which he recognized that he could code — and that birthed VisiCalc, which led to there being a spreadsheet application on every smart device. In that case, as is too often the case, it was Bill Gates that benefited the most (with Excel), not Brickman. Not so for Larry Page and Sergey Brin who turned searching the web into an algorithm (perhaps the most famous algorithm in the world) that it converted to code to become two of the richest people in the world. Mark Zuckerberg turned social networking into algorithms that he could code. And so on…
The Knowledge Funnel explains much about the modern economy. If you can drive a heuristic to algorithm to code, you have a chance to make it big. That is why software isn’t an industry in the traditional sense — like retailing or specialty chemicals or auto parts. Instead, it is the destination for knowledge. All knowledge is heading in the direction of software. Lots of knowledge gets stuck in the mystery phase and others in heuristic, but much is charging through to algorithm and then someone figures out how to code it.
You use heuristics. But you also dwell in mysteries. And you run algorithms. And use the product of coding. You may even code algorithms, depending on your job. That is life on the planet in the modern world. You can’t exist in only one of the four states.
With respect to heuristics, use them but understand their limitations. Heuristics never guarantee the right answer — but they are the best knowledge we have in many situations.
You will often be told that you should use an algorithmic approach: crunch this data, this way, and it will tell you the answer. Be wary. Much of the business world is using algorithmic approaches that aren’t backed by the work necessary to push knowledge from heuristic to algorithm. You will be told that it will give you the ‘right answer,’ or ‘the truth.’ Audit them carefully to see whether that is hype or a valid promise.
The easiest thing you can do is to run your heuristic. It is valuable because it depends on your experience in understanding the nuances of the heuristic — whether your heuristic is about product management, M&A, or strategy. But recognize that you can create additional value by pushing your heuristic father along the knowledge funnel. Think Dan Brickman. Can you push it to an algorithm and code it to be the next great software company? Most times you can’t but sometimes you can, and it is worth it if you can.
The key is to think about yourself as being on a journey. Always be sure to be working on some mystery in your business — why customers act the way they do; why bottlenecks pop up in the strangest places, etc. Hone your heuristics, even as you acknowledge their limitations. They should get better and better with experience and reflection on them. And look for opportunities to hive off pieces of your heuristics to push them to algorithm — and then code the heck out of them. That is how to utilize the Knowledge Funnel to benefit you and the world.