We’ve seen a number of Idea Maze cases so far. What leaps out at you? What doesn’t?
In this piece, I want to talk a little about what’s leapt out to me while Guan Jie and I worked on these cases. If your conclusions are different from mine, don’t fret — it’s likely that your takes are just as (if not more) valid than my conclusions, especially when read in the context of your operating domain.
(One of the implications of Cognitive Flexibility Theory is that it’s necessary to inculcate an ‘adaptive worldview’ — that is, in ill-structured domains like business, there are many valid lenses and lessons to take away from cases. These lenses are mine; you will have yours; that’s perfectly alright.)
Here are the handful of things that I’ve noticed.
The Idea Maze is Fundamentally About Uncertainty
Most of these cases are about people building radically successful new products, which they had no hope of imagining before they started building. The analogy of the maze is one way of thinking about this experience — but another, equally powerful way might be ‘feeling one’s way through a thick fog.’
For instance, notice how — early on in the iPhone’s development in the iPhone Keyboard case — Apple’s leadership asked themselves after every major demo: “Did this demo close the prototype-to-product gap, even a little? Were they seeing positive change over the previous demo? Is this technology or app on track?”
Think about what that meant, though: it meant that Apple had to have some idea of the iPhone’s final form, but that so many of the details were still unknown to them that all they had available to evaluate these details were ‘squishy’ questions. At that point in time, Apple’s leadership knew that the iPhone would have certain constraints: it was going to have a multitouch-enabled capacitive touchscreen, it was to run a desktop-class operating system, and the user interface had to work with human fingers alone. But the only way Apple would figure out the rest of the details of the iPhone was through iteration … and the unit of iteration that Apple worked at was the demo.
The fundamental feature of the Idea Maze is that navigating it is about making decisions under uncertainty, which means that you can only really see a few decisions ahead of you. Nearly all of the cases involve some form of “they decided to do this, but they weren’t sure about that, and then they built this, which made them realise they had to decide between this and that.”
One way of reading this is to say that “feeling totally unsure about your decisions is the norm”. Many, many instances of the Idea Maze are basically stories of people doubting themselves (in fact, the one case where people were fairly confident they were building something revolutionary was General Magic, which failed). My point is that you shouldn’t be surprised when you start to doubt yourself.
Conviction Without (Strong) Evidence
Another thing I found pretty striking about the cases is the occasional need to develop conviction when faced with little to no evidence.
I’m thinking about Instagram, where Krieger and Systrom were unsure about their pivot to photos fairly late in the game. They built one photos-first prototype, Scotch, before abandoning it. But then they stuck with their second prototype (which eventually turned into the app we know today).
Or look at Amazon Prime, where Jeff Bezos held firm on the program, even throughout the two year period where the data did not show a predicted behaviour change.
Or TikTok, where Yiming judged that A.me’s execution was lacking; that there was still something in the Idea Maze worth chasing.
Or Microsoft Office, where members of the Desktop Application Division (DAD) had to debate the wisdom of pursuing a suite strategy (“did people really want to buy all of these apps?”)
As a counter example, consider that General Magic’s Marc Porat had seemingly high conviction betting against the World Wide Web — a decision that seems unwise with the benefit of hindsight.
What differentiates these cases?
I won’t be so cute as to suggest that you should always “develop conviction in the face of no evidence”. At most, I’ll suggest that “you should expect to make some decisions on the basis of little-to-no evidence when navigating the Idea Maze, and this is what that looks like.”
A slightly more opinionated take is that developing ‘conviction in the face of great risk and/or uncertainty’ demands a healthy respect for that uncertainty. Bezos was willing to wait a few years for Prime to succeed because: a) he knew that Super Saver Shipping had produced a similar behaviour change before Prime, b) that this was an existential issue for Amazon, and c) Amazon’s leadership knew any changes would take a few years to show up in the data, since people only shopped on Amazon a few times a year in 2004.
In other words, the argument for ‘no data for two years’ was sound. It’s worth asking if Bezos would’ve changed his mind if the data played out badly in three, or four, or five years. There’s no way to know for sure. Presumably he would’ve done so, assuming no compelling counter-argument emerged in those years. What we can be sure is that they had to make a decision to stick with Prime, even when they weren’t 100% sure it would work … which means that it shouldn’t be too surprising to you if you’re forced to do the same in your work.
Good Ideas Can Be Random, and are Often Fragile
One other thing that’s interesting is how often good ideas aren’t recognised in the moment. Or, more accurately, how reluctant people might be to embrace them.
For example, email had to be shoved down Max Levchin’s throat in the PayPal case — even though he was the one who implemented it in the first place.
It took a number of years for the suite strategy to gain traction within Microsoft — and it came from an unlikely place, the Macintosh marketing team. Heck, the notion that the suite could be a way to beat the competition only became clear after Lotus had pushed their ‘one integrated suite of products’ marketing message.
And finally, it’s worth remembering that it was Systrom’s girlfriend who provided the spark for Instagram’s adoption of filters — commonly said to be one of the reasons for their eventual success.
The most conservative conclusion here is that “good ideas can come from anywhere, look stupid, or take time to gain acceptance when they first emerge, and this is how it looks like” but a slightly less conservative conclusion might be “… and therefore you shouldn’t be too quick to dismiss ideas when you’re deep in the Idea Maze.”
Many Things Right
Some instantiations of the Idea Maze demand a large number of things to go right. My goto example for this is Instagram, where Krieger and Systrom needed to solve the a) slow image upload problem, the b) upload to multiple platforms problem (which later turned out to be a distribution boon), the c) low quality phone camera problem. They also launched Instagram with the help of Jack Dorsey’s massive Twitter following, right at the time when smartphone adoption was beginning to take off, when smartphone photography was beginning to be a thing; they kept the app simple right when incumbents were too oriented towards the web.
It would be disingenuous to say that luck didn’t have something to do with Instagram’s success. (“You can get lucky when navigating the Idea Maze, and this is what that looks like.”)
But one other conclusion you may draw is that only some forms of the Idea Maze are difficult in this ‘multiple things right’ sort of way; others are difficult in very different ways.
With Microsoft Office, for instance, the bulk of the difficulty was in the org design necessary to pursue the suite strategy (after the DAD developed conviction in said strategy).
With PayPal, the difficulty was in the many bad iterations before they stumbled onto the winning formula: sending cash over email.
With TikTok the difficulty was in executing on a copycat idea over a long period of time, even when it didn’t seem to be gaining traction. You could say that TikTok’s Idea Maze was as gnarly as Instagram’s, but TikTok’s success took much longer, and so I feel relatively safe in saying that it was difficult in a different way.
You’ll notice that my conclusions here are all very conservative. One of the nice things about collecting multiple concept instantiations is that you very quickly disabuse yourself of certain common-sensical but ultimately misguided beliefs. For instance, before I started on this series, I thought that world-dominating products grew quickly, and reached large audiences soon after launch. But that turns out to be merely one way that product success can look like. And even certain world-dominating products took a long time to get there (TikTok).
I also thought that it was always good practice to iterate in a ‘disciplined’ manner — which I believed meant clear statements of hypotheses at each turn of the Idea Maze. But it’s not so clear that this is required in every case … as you’ve probably already seen, some of the better successes relied on intuitive decision making … and a touch of luck.
As always, I’d urge you to read this piece not as a definitive take on Idea Maze patterns, but as some things that leapt out at me. The actual cases are more important than these takeaways, because your brain represents concepts as a cluster of cases. All of this is to say that you should have your own set of takeaways, assuming you’ve asked yourself, at the end of every case:
Compare this case with a previous case you've read. What is similar? What is different?
Does this remind you of a similar case? If so, what is different there?
These observations merely came out of my answering these two questions, repeatedly, over a period of weeks.
Cedric Chin 22 August 2022.