Part 2 of the How Long Will Your Business Last blog series on the arc of organizational success

by Alex Lamb

In my last blog post, I introduced the idea of the arc: a developmental curve that appears to describe the rise and fall of a large number of companies. I explained that while I’d been talking about the arc for years without ever expecting to see it in reality, I found clear evidence for it on Google Trends, and then later in population data and a host of other proxy markers for organizational success. In this blog, I’ll demonstrate how to model the arc and discuss what real world factors might be behind it.

Why the arc?

Before I cover modeling, however, we should first ask why we might expect a characteristic arc for organizational development to exist at all? Why should company growth follow any specific pattern? Are we reading too much into the Google Trends data?

I’d argue that the arc exists because there are certain underlying realities businesses can’t avoid that determine how they evolve. For instance, I’d propose that a company usually needs two good ideas to get started. The first idea enables it secure preliminary funding and start operating. At this point it enters a highly plastic phase during which it alters its business model to try to secure market fit. If market fit is secured with a second good idea, growth begins.

Once growth kicks in, a company increases in size approximately exponentially unless market or logistical factors prevent it from doing so, and continues until its niche saturates. There can be many causes for that saturation, including adverse competition, change in market conditions, or dysfunction from within. Regardless of the cause, growth plateaus and the company enters its decline phase.

Decline from ‘process debt’

Organizations naturally resist decline. However, several factors impede lasting stability. The most significant of these is that a company has usually incurred what you might call ‘process debt’ during its rise. This means that even if there is still a niche to occupy, the company will struggle to adapt to changing conditions because of the infrastructure and hierarchy it has built up by virtue of its success. Attempts to maneuver the firm so as to capitalize on new revenue sources require reorganization. This incurs friction that scales with the company’s size. In effect, the larger you get, the more organizational mass there is to move if revenue sources slide out of reach, and the more pain and in-fighting is involved.

If a company is both smart and lucky, it can muster the required momentum to pull off another rise and the process repeats. However, secondary rises are rare for reasons we’ll explore later.

Modeling the arc

This rather uncontroversial picture of organizational development is enough for us to start modeling. We start with a collection of very simple agents and a resource well of market potential. In a given simulation round, agents attempt to draw resources from the well, which can be tapped with probability in proportion to how much potential remains. If they succeed in tapping the well, they reproduce or hire, whichever way you want to look at it, with probability pH. If they fail, then they die (or are fired), with some lower probability pF.

In the spirit of complex systems research, let’s examine the fit of this ultra-simple model against our data before we consider adding extra subtlety. After all, if a very general model reproduces the trend, then adding details will only pollute the result.

Here I’m using a very simple hand-rolled genetic algorithm to generate candidate parameters and aggregated batches of ten simulation runs to compensate for model noise.

Facebook. Source: Google Trends

This doesn’t strike me as too shabby for a first approximation. What we’re looking here is close enough to a skewed normal distribution that an analytical approach would serve equally well, but I find the agent-based approach more tightly explanatory. However, is this match a fluke? Let’s look at the model-fits for some of the other organizations I showed you in the previous post.

Flickr. Source: Google Trends
Second Life. Source: Google Trends
Tumblr. Source: Google Trends
MySpace. Source: Google Trends

Predicting organizational lifespan

These fits really aren’t too bad for a few hours of gentle investigation, which means that now we’re in a position to be able to start making predictions about organizational lifespan. For instance, how long from now would we expect it to take for Facebook’s core product to halve in social relevance? On the face of it, approximately six years. (Note that this does not reflect Facebook as a whole! You only need to look at the Trend curve for Instagram to see that they’ve made some wise acquisitions.) However, despite the fun of making reckless industry speculation, I’d propose that it’s more relevant to ask why the simple approach works as well as it does in the first place.

What the model suggests to me is that even though the internal dynamics of companies are highly complex and their fates propelled by human actions, from an aggregate perspective, the outcomes don’t really change. And this is because the number one determinant of organizational success is luck in matching an ever-moving landscape of opportunities.

I’d propose that the arc dominates organizational lifespans because good business models are rare and increasingly temporary. Compared to many organisms that occur in nature, successful modern companies are large, delicate creatures with respect to their environment, and the world in which they live is a volatile, highly-connected one. Most do not survive very long, having never secured a stable market niche in the first place. But this doesn’t mean that leadership choices are irrelevant. To my mind, what this picture tells us is when and how good leadership matters most.

In part 3, we’ll look at how good leadership can make a difference, and take a closer look at companies that have beaten the decline curve.