A few days ago, I discussed how systemic bottleneck analysis could be a reliable way to make market predictions in a high uncertainty context. While I’m often baffled that consulting firms never really share how the prediction sausage is made, I truly believe it’s important to demystify this work. And let’s not be fooled; making predictions is easy until you have to justify a few down the road why you were predicting things that never happened!

Predictions don’t work, and yet…

Yes, this is my mandatory copy and paste disclaimer when discussing forecasts and market predictions… What’s the point of these forecasts if you can’t trust them and want to disprove them?

Forecasts and predictions give you reference points looking ahead. The hypotheses that are framed precisely set a vision of the future that you can assess. The hypotheses are the real value you want to extract; predictions themselves are just a by-product.

And I’m sorry to say that customer companies are mainly to blame for the poor state of the “prediction market”.

In the early 1980s, AT&T asked McKinsey to estimate how many cell phones would be in use in the world at the turn of the century. They concluded that the total market would be 900,000 units. This persuaded AT&T to pull out of the market. By 2000, there were 738 million people with cellphone subscriptions.

Andrew CHEN

Companies want to know where to skate next, never really why. As such, they never really build up their own 6th sense or compression radar-like detection capabilities and need on-going perfusion of weak signals, trends, and next big bets. Paying big money to Gartner or Deloitte to get the same sectorial info as your competitors seems to be OK if your exec committee is happy with the reinsurance.

This prediction fallacy is not only due to cultural laziness or short-term corporate political games. It’s also about being able to break out from your incumbent mindset.

Sizing the market for a disruptor based on an incumbent’s market is like sizing a car industry off how many horses there were in 1910.


When you understand that the value is in how you build your predictions, not so much the output, how do you proceed?

I know that very often, market forecasts are just made by trying to extrapolate how weak signals and trends would scale up within a few years from now. To put it as nicely as possible, this is a very candid way to work at it (akin to playing investing billions in the stock market just using linear regression).

But it doesn’t have to be. And as already discussed a few months ago, there are many (many) tools and strategies to make predictions and unmask the underlying hypotheses about the market.

The difference between linear times (low market uncertainties) and non-linear times (high market uncertainties).

Seeing systemic bottleneck

In that regard, systemic bottleneck analysis is rather interesting when markets are all over the place, and no one really knows what’s going to happen six months from now.

So what are we talking about?

A systemic bottleneck is a turn point event that is building up in your market that, when resolving, will impose a 10 years directionality impacting every actor in the said market.

Incumbent players often disregard systemic bottlenecks because they will reshuffle the cards so much that it’s genuinely difficult to face them. If you want one of the most spectacular examples of these last 10 years, it’s for sure the shift from the combustion to the electric engine in the automaking industry.

And if you think that what leads to this ‘catastrophic’ market turn point is the rise of consumers buying electric vehicles, you’d be wrong. If anything consumers behaviours is always is a laggard metric, and the worst possible one to base forecasts on. The underlying systemic bottleneck is much simpler to see coming: it’s massive, it’s been years in the making, and as soon as it reaches critical mass no one in the value chain will be able to ignore it anymore. I’m talking about worldwide electric battery manufacturing capabilities.

Governments investing in macro-infrastructures are the easiest to spot systemic bottlenecks in the markets.

This is the real systemic bottleneck for the car industry. And it reached critical mass around 2018, pushing China first to massively adapt its automating culture to EVs, while the US and Europe are still slowly shaking out of their state of denial).

In my latest webinar, I briefly discussed 3 systemic bottlenecks that will impact most markets:

  1. Tech platforms (GAFAM and BATX) reaching market saturation.
  2. Covid-19 pandemic being a new normal for the coming 3-5 years.
  3. Hardware (not software) platforms being the next trillion dollars businesses.

Understanding them, monitoring their build-up (or not), and already forecasting the directional outputs they will force upon markets is certainly where the big money is for most multinationals, but also… startup.

How systemic bottleneck analysis work

If we consider this first systemic bottleneck: 1. Tech platforms reaching market saturation we’ve been seeing its build-up for quite some time now. Even recently, the TikTok ban discussion in the US had much less to do with Trump than with incumbent platforms unable to innovate their way up anymore, which will undoubtedly lead to more and more cold war-like regulatory fights.

Distinguishing between a market bottleneck and its possible consequent.

Is this bottleneck really building-up? Just monitor these events, like you’d monitor the first waves preceding a major earthquake. And being smart about it would mean monitoring high and wide, to be sure that you don’t miss what is also happening in other markets as well…

The second order of consequences of tech platforms reaching market saturation: moving to tier 4-5 cities in China.

Forecasting the outputs

While seeing the bottleneck coming up is the core added value, the less essential by-product is forecasts and predictions. You have to view pushing the logic to the next step and forecasting the second order of consequences of the bottleneck as a bonus :

The difference between seeing a systemic bottleneck coming up and following up with predictions.

In the case above, the core strategic value is when you understand that all consumer markets will be constrained by increasingly aggressive platforms gasping for oxygen. If you see this bottleneck coming, the growing importance of tier 4 and 5 cities in China will become an obvious input in your next strategic planning. Predicting that Apple or Starbucks will first deploy regional networks from these tier 4-5 cities is non-essential. You might be wrong in the sense that Tesla will be the first mover here or another western brand entirely. Your prediction would be wrong, but you got the huge market shift right and maybe had a chance to prepare (or even leverage it).

It’s just like predicting that houses number 5 to 10 will be toppled off when you know that a major tsunami is going to hit. The core value is in seeing the tsunami coming. Said differently:

When dealing with high uncertainty markets, 80% of the total Innovation ROI is the Scouting ROI, not the financial ROI.

How bottleneck analysis can help you?

To conclude this post, let me share 3 simple and immediate deliverables of understanding and applying systemic bottleneck analysis in your business:

If you’re a startup:

  1. Center your business plan on a clear market hypothesis: What is the systemic bottleneck that builds up in the market you target within the next 3 years, and how are you going to leverage it to bypass incumbents’ slower to adapt to it?
  2. Present rational pivot strategies from the get-go: Even if your predictions on how the market will adapt to the bottleneck are wrong because you’ve seen it ahead of time, how are you prepared to use it differently if need be?
  3. Target your first key customers more ambitiously: Rather than aiming for more accessible customers for your startup, can you already target some of the big ones in your market? They will have to face harsher consequences going through the bottleneck you leverage. Of course, you might not convince a lot of them, but getting only one onboard will validate your strategy way more than a dozen of unknowns.

And if you’re a corporation:

  1. Rebalance your innovation strategy: How many of your innovation projects (internal or external) are actually centered around clear 3 to 5 years systemic bottlenecks building up ahead of your cash cows? The answer might be “not enough” or worse, “it’s unclear.” This is the easiest to fix critical flaws of most corporate innovation programs.
  2. Have projects explore different bottleneck outputs, not only a big win hypothesis: Consecutively, how many different outputs are you exploring around a key bottleneck? The usual mindset is to try to find one big win and get all in. Again, in a high uncertainty context, this is flawed. You cannot predict where the next big win will appear, but you can prepare to react quicker than everyone else and leverage a systemic market turnout.
  3. Identify the scouting ROI for each innovation project: For each innovation project you’re dealing with (again, internal or external), can you define the ROI of scouting the market ahead of time for your core business units? With a metric?