Four Ways to Scale Up: Smart, Dumb, Forced, and Fumbled

Modularity and speed are key to smart scale-up. In contrast, bespoke plus slow equals dumb.

Bent Flyvbjerg
8 min readJun 22, 2021
Source: Author,

Figure 1 illustrates four ways to scale up a venture, based on a simple two-by-two matrix. On the horizontal axis, the figure shows modularity, with ventures grouped as either modular and replicable or bespoke and one-off. On the vertical axis, the figure shows speed, with ventures grouped as fast or slow. This results in four main types of scale-up, of which smart scale-up and dumb scale-up are the most important.

We define scalability as the capacity to change something in size or scale, for instance, the potential for an entity to be able to handle a growing — or diminishing — workload. Contrary to common under­standing, being big is not the same as being scalable. Scalability is the capacity to change in size, includ­ing the capacity to grow big at speed, but scalability is not identical to size in itself. True scalability resides in the smart quadrant of Figure 1 and dumb scale-up is not scalable, it’s just big.

For instance, the Channel rail tunnel between France and the UK has a fixed capacity that is not scal­able, and because the tunnel has only had demand at just over half of this capacity the investment proved financially and economically nonviable. Twice the tunnel has had to be bailed out of insolvency and to be financially restructured. Similarly, consider a large hydroelectric dam. It cannot scale beyond its capacity and is therefore not scalable.

Now contrast this with a wind farm where you can add, or remove, turbines to adjust capacity as needed, up or down. The wind farm is scalable. And if you need more capacity than one wind farm can provide, you can add another wind­farm in the same manner, so that windfarms are scalable not only at the level of turbines but also at the level of farms. The same holds for server farms, solar farms, energy storage and transmission, and many other phenomena. Wind is just an example. We call the ability to scale at any scale “scale-free scalability.”​ It is a core feature of truly scalable systems.

In what follows we account for each of the four types of scale-up, one by one.

Figure 1: Four ways to scale up, based on modularity and speed

Source: Author,

First, smart scale-up is placed in the upper right-hand quadrant of Figure 1. Smart scale-up entails developing something big based on the principles of modularity, replicability, and speed. In “​My First Megaproject: 20,000 Schools in Nepal”​ I describe how my team and I used smart scale-up to plan and deliver 20,000 schools in Nepal, successfully getting children, and especially girls, to attend primary school, starting a virtuous circle of better education and health, falling birth rates, and economic growth. But smart scale-up is more common in the tech industry than in construction. Computers and smartphones, e.g., are built using smart scale-up, as are commercial satellites and cloud-computing

Second, dumb scale-up is placed in the lower left-hand quadrant of Figure 1. It is the opposite of smart scale-up. With dumb scale-up you slowly build something big and bespoke, for instance a nuclear power plant, a large hydro-electric dam, a big-bang enterprise resource planning (ERP) system, a chemical processing plant, a new mine, a national health or pension IT system, a big defense system, a space shuttle, an aircraft, or a new airport.

Dumb-scaled ventures are often binary, in the sense that they are either completely on or completely off, with no middle ground. For instance, a dam or a nuclear reactor cannot be completed in incremental fashion, in contrast to a server farm or the schools in Nepal . Even a 95 percent completed dam or nuclear reactor is of no use; it must be 100 percent completed before it can deliver its benefits. That’s a big drawback for anyone who needs to impact the world now, or get to the cash flow fast to grow their business.

Dumb scale-up is common in conventional business and government, with devastating consequences for investors’ and taxpayers’ money, and for the wealth of nations. Dumb scale-up strongly correlates with high costs and large cost overruns, long schedules and schedule delays, and sizable benefit shortfalls. The combination of cost overrun and benefit shortfall makes for bad business, needless to say, which is why this approach is called “dumb”: it wastes money, talent, and other resources.

The result is your classical boondoggle, like California’s high-speed rail system, Japan’s Monju nuclear power plant, the New South China Mall, the F-35 Joint Strike Fighter program, Kmart’s enterprise resource planning system, the UK National Health Service IT system, the International Space Station, Pakistan’s Tarbela dam, or Berlin’s Brandenburg airport. In the terminology of Taleb (2012), dumb-scaled ventures are fragile whereas smart-scaled ones are robust, or even antifragile (Ansar et al. 2017).*

Source: Author,

Third, forced scale-up, is shown in the upper left-hand quadrant of Figure 1. It designates the combination of a one-off, bespoke design and high-speed delivery. This will typically result in a product of low quality, because it is difficult to deliver bespoke designs at speed. For ventures with forced scale-up, speed is often imposed from the outside, for instance by pressure from top management or politicians for monumental prestige projects, or by the project running up against an immovable deadline, as in the case of mega-events.

Hosting the Olympics or the FIFA World Cup are examples of forced scale-up, with their typical combination of bespoke signature architecture — notorious for being late and over budget — and a deadline written in stone that hosts almost always have difficulty meeting.

Cost overrun for the Olympic Games is higher than for any other type of megaproject, at 172 percent on average in real terms (Flyvbjerg et al. 2020). Many host countries and cities are littered with low-quality stadiums and other facilities that are underutilized or boarded up. Facilities often begin falling apart immediately after events are over, many ending up as full-fledged white elephants. Athens, Beijing, and Rio de Janeiro are recent examples, as host cities for the Summer Olympics. The Athens 2004 Games, hosted by Greece — a small country with a fragile economy — became a contributing factor to the country’s 2011 debt default. The cost overruns and incurred debt from the Games were so large that they negatively affected the credit rating of the whole nation, thus weakening the economy in the years before the 2008 international financial crisis. This resulted in a double-dip recession in Greece — with financial and economic disaster still rippling through the economy a decade later — where other nations had only a single dip and recovered faster.

China’s construction bubble — driven by a policy of nation building through accelerated development of infrastructure, buildings, and whole cities — may also be considered an instance of forced scale-up (Ren 2017). To illustrate just how accelerated the speed is, consider that China poured more concrete in the three years from 2011 to 2013 than the US did the entire 20th century. For decades, China has been driving what The Economist calls “the biggest investment boom in history,”​ recently accelerated by President Xi’s so-called Belt and Road Initiative, exporting China’s model for infrastructure building on a global scale. Global infrastructure spending has never been this high in both absolute and relative terms, i.e., measured as a share of world GDP.

Source: Author,

Fourth and finally, fumbled scale-up is shown in the lower right-hand quadrant of Figure 1. This is the least bad of the three types of non-smart scale-up.

Fumbled scale-up designates a combination of modular, replicable design and low speed in delivery. In principle, modularity would allow for high-speed rollout, but for whatever reason speed is not achieved, for instance because of lack of capital, personnel, C-suite attention, or other resources. The result is a missed opportunity. If competitors exist that master smart scale-up, they will typically outcompete the entity that fumbled their scale-up.

The joint venture of UK supermarket chain Sainsbury’s and Danish low-cost retailer Netto to roll out discount stores in the UK in the mid 2010s is an example of a fumbled scale-up. Discount stores are modular and replicable by design and therefore easy to build at speed. Nevertheless, after three years Sainsbury’s and Netto had established only 16 stores in the UK, while the competition — Germany-based Aldi and Lidl — had set up hundreds with hundreds more planned.

What happened? Sainsbury’s got sidetracked by a major acquisition and lost focus. Moreover, availability and cost of appropriate sites for discount stores caused difficulties, because of increased competition and demand. As a consequence, Sainsbury’s and Netto were slowed down, which is the one thing you cannot afford in a competitive market where smart scale-up decides who succeeds. In the end, Sainsbury’s and Netto were forced to close down their joint venture, lay off several hundred people, and write off the investment. Scale-up takes no prisoners when competition is fierce, as it is in discount retailing. If you fumble you die.

Ask yourself, how can we do things more modular and faster, now?

In sum, you can make yourself and your organization — whether a business, a government agency, or an NGO — hugely more valuable if you, first, become clear about which of the four quadrants in Figure 1 you currently operate in and, second, move your focus and your activities systematically and effectively towards the upper right-hand side, to smart scale-up. This is not an either-or proposition but a matter of degree. Your organization and most of its ventures — existing or planned — will be neither fully dumb nor fully smart, but will have elements of each. Your task is to increasingly and tenaciously tip the balance towards smart by beginning to introduce smart-scaled ventures, and elements of smart scale-up into existing ventures, to make your organization less dumb and ever smarter.

Specifically, you need to ask, “How can we make each thing we do more modular and less bespoke, more replicable and less one-off, more speedy and less slow?” or, in short, as expressed in the following fast-and-frugal heuristic:

Ask yourself, how can we do things more modular and faster, now?

If you do this — and if you get your team and wider organization to adopt the approach — you will unleash potential and achieve results you never thought possible. That’s smart!

— — — -

*) For a longer version of this text with full references, please see Flyvbjerg, Bent, 2021, “Four Ways to Scale Up: Smart, Dumb, Forced, and Fumbled,”​ Saïd Business School Working Papers (Oxford: University of Oxford), free pdf here:



Bent Flyvbjerg

Professor Emeritus, University of Oxford; Professor, IT University of Copenhagen. Writes about project management.