What to Read Next
Companies that embrace remote work — which is here to stay — can also drastically reduce their coordination costs through modular organization. Spotify announced earlier this year that it would completely shift to remote work. Salesforce expects its workers to be in the office just one to three days a week. Last fall, Apple CEO Tim Cook said he believed the company would not return to the way people worked pre-pandemic, noting, “There are some things that actually work really well virtually.” Ping An — China’s largest insurer — took its remote work one step further: It developed a suite of tools that lets its 1.4 million employees and agents work remotely. It’s now offering those same tools to other financial service companies.
These companies have one thing in common: They are all moving toward or have already achieved a modular setup. Whereas modular organizations were challenging to implement in the past, they are becoming commonplace today thanks to microservices and application programming interfaces (APIs).
Email Updates on Digital Culture & Strategy
Get monthly email updates on platforms, digital leadership, digital transformation, and ethics.
Please enter a valid email address
Thank you for signing up
The concept of modular organization is not new. What’s new today is the ability to achieve such architectures with relative ease. APIs help codify interactions among departments, which in turn reduces ad hoc communication and minimizes coordination complexity. As a result, traditional business processes — whether financial, legal, or HR-related — can turn into microservices. That’s how a monolithic, highly interdependent organization turns into a modular one.
Note that the converse is also true: A lack of microservices can prevent a company from achieving a modular architecture and keep it stuck in widespread complexity. These types of organizations face no choice but to order their headquarters staff to return to their physical offices.
Our findings are based on a research program at IMD Business School that aims to understand how organizations become future-ready. We have conducted extensive interviews with dozens of forward-looking organizations and have built a database that tracks over 40 ecosystems. We’ve also ranked companies’ readiness based on publicly available information. Our executive education programs essentially serve as a laboratory, allowing us to test out hypotheses with senior executives.
Organizational Complexity Demands Constant Coordination
Consider a task that depends on the input of others. An executive at an established fashion manufacturer explained that every time his company prepared to launch a new product, he would have to dial into a briefing from headquarters, after which he would call his team to parcel out the responsibilities. Each team member, in turn, would communicate with their external counterpart — the freight forwarders, the retail stores, the advertising agents — all the while keeping headquarters updated on every milestone. With Excel spreadsheets flying, everything was done manually — and this was early in 2020, right before the onset of the global pandemic.
This is complexity — a sprawling operation where anything can affect everything else. As a result, managers working in such a company need a lot of face-to-face meetings to coordinate and strategize.
Relying on a slew of meetings or exchanges across communication platforms — whether Zoom, Microsoft Teams, Slack, or email — to connect the dots among your team members quickly becomes overwhelming. Connecting the dots across the organization as a whole becomes nearly impossible. It’s this sort of organizational complexity that prevents companies from optimizing remote work.
The solution: A modular and distributed organizational design.
How Microservices Power Modular Organization
A business that is separated into independent operations can be described as having a modular organizational structure. In the late 1970s, Tom Peters of McKinsey pronounced that we should move beyond the matrix organization, a structure he viewed as too slow-moving to cope with the accelerating pace of the business environment. In the 1990s, James Moore discussed the death of competition due to emerging ecosystems. More recently, Gary Hamel declared that bureaucracy needs to be eliminated to attain a more agile way of organizing. These influential thinkers all sensed the importance of reducing internal coordination.
Speed is an advantage. The pandemic rewarded companies that minimized the high cost of coordination. For instance, Qingdao-based Haier Group, one of the world’s biggest home appliance manufacturers, had returned to full-capacity production within a month after the COVID-19 outbreak, mainly due to the company’s widespread use of intelligent manufacturing and remote-working solutions.
But every fast-moving giant you see today has had a near-death experience at some point. They thrive only because they have simplified. They took pains to rebuild themselves from the ground up to avoid collapsing under the weight of their own complexity.
Amazon started in 1996 as a single-application entity. It ran a web server that talked to a database on the back end. Over time, the single-application entity sprawled into a monolithic architecture of a million lines of code. Observing that Amazon’s application developers were in a constant struggle with the hardware server team, CEO Jeff Bezos instructed the developers to create some standard APIs for accessing and allocating computing resources.
Bezos was adamant about reducing coordination complexity. The new directive allowed for vastly fewer face-to-face meetings and emails. Instead, the big functional system — in this case, IT — was broken down into smaller service modules called microservices. Each microservice communicated with others through the APIs. Because the format of the requests sent between app developers and IT operations was made consistent, routine coordination became automated.
The end result is that a small team of software programmers can now launch an additional feature on Amazon.com independently, in a way that’s similar to how a small merchant runs a Facebook campaign. These sellers don’t call up Facebook customer service to schedule a campaign by phone. It’s all self-service, and the execution is automatic.
For the Amazon engineers, launching a new product offering is not so much like coordinating across departments. Rather, it’s like assembling Lego bricks, mixing and matching capabilities for new offerings that fit the changing marketplace. That’s the complex made simple.
Expanding Microservices Beyond IT
Amazon is turning its entire IT infrastructure into tens of thousands of microservices by standardizing the input and output of each function. These microservices can then exist as independent, decentralized modules that interact with other modules.
But the idea of microservices doesn’t need to stay within IT. For years, Haier has organized itself as a swarm of self-managing business units.
Haier has some 4,000 microenterprises (MEs). Each one comprises just 10 to 15 employees and has full autonomy to deliver offerings to other MEs and, oftentimes, the final products to end consumers. Instead of being centrally orchestrated, these MEs independently interact with one another. Certain MEs provide services like HR or product design; others manufacture specific component parts. Coordination is managed through internal platforms in the cloud, as in an app store. Because each team is small and self-contained, an isolated failure won’t cause global disruption.
Haier’s Access Control Solution
Since the interfaces among these MEs are highly standardized, it’s possible to automate information exchanges. One concrete example is the solution Haier developed for access control.
At a large business with many employees, people are constantly joining, leaving, and moving, which makes it difficult to control who has access to which enterprise applications. Under the framework of microservices, Haier designed its adaptive identity governance solution, which frees IT staff members from the repetitive manual tasks of authorizing and de-authorizing business users.
Instead, HR personnel can track all of an employee’s access rights from a single platform and modify access with the push of a button. Because identity management is also centralized in one database, Haier enjoys substantial savings on maintenance costs. Individual departments no longer need to run their own management platforms. This centralization is an important feature. For microservices to be effective, data needs to reside in the cloud rather than on local servers. That way, there will be only one “true” version of the data.
Using Externally Created APIs
Smart companies also choose not to write up their own APIs every time and instead deploy some externally developed open source applications that can be found in the cloud. Examples of open repositories include Google Cloud APIs and OpenStack. These APIs are essentially software components that provide common services for cloud infrastructures.
With all of its internal microservices established, Haier found it easier to leverage the open APIs developed by others, including those initially designed by the financial sector. Today, an application developed by Singapore bank DBS is providing digital financing for distributors of Haier’s products. Haier’s sales and distribution channels are given extra support, and DBS gets the additional business.
Ping An was also an early adopter of financial API technology and has been using it since 2013. At first the company focused on automated retrieval of real-time balance information. Gradually but surely, however, Ping An evolved from being an API receiver into a producer. By 2018, Ping An’s OneConnect fintech subsidiary had developed hundreds of interfaces, similar to what DBS had done.
From Remote Work to Business Growth
The urgency of remote work has underscored the advantage of modular organizations. Our research shows that companies that leverage microservices and APIs not only find themselves able to reduce complexity and manual coordination but are better suited to a geographically dispersed workforce that interacts asynchronously.
From a process perspective, executives must embrace these key principles in designing a remote-first organization.
- Build a cloud-first operation. Executives may hesitate about bringing their company’s valuable data into the cloud, and some view it merely as a tool to store more data efficiently. But a cloud-first operation serves as the basic infrastructure for becoming a modular organization.
- Digitize critical business processes. Almost every company these days is in the process of digital transformation. No doubt this has been accelerated by the COVID-19 crisis, but digitizing business processes is necessary anyway. Only with widespread digitization can a company break up interdependent operations into small pieces, thus creating microservices, in an efficient manner.
- Develop digital interfaces among microservices. While much focus goes into hiring top talent in areas such as data science and machine learning, it’s also critical to build up API interfaces. The value of the API exchanges is massive, and no company can afford to ignore it. As the Haier example shows, once you acquire API development capability, you can further leverage other existing, open source APIs.
- Embrace the strategic value of microservices beyond IT. The real power of microservices is in reducing coordination and communication costs throughout the organization. Look beyond the IT functions through the lens of broader business operations.
Complexity is independent of the size and scope of a company. Amazon, Apple, and Google are enormous organizations but are elegantly simple in their setups. Haier is an incumbent manufacturer with massive operations around the world, but its organizational setup is remarkably simple. Ping An is organized as a full stack of interdependent businesses with high autonomy.
Each of these modular businesses leverages a wide variety of microservices to create advantages. What was once an aspirational concept of organizational design is today an achievable reality, accelerated by a global crisis.
Editor’s note: An adapted version of this article appears in the Fall 2021 print edition.