To Modularize or not to Modularize - that is the question
This is about deciding what activities a company should keep in house and what can be outsourced or contracted. Thinking of an end to end solution which could either be a product with various components or a solution with product and service components, one needs to decide if a company is best placed to deliver the complete solution or be just the assembler who assembles the solution using the various components from vendors or provide just one or two components. Now the answer is it depends in which stage of the innovation cycle the technology associated with the solution is. For an early stage innovation which is not meeting all performance requirements a clients need, integration works better as companies have to continuously tweak every component to arrive at the best performance. One company managing all the components can have much more agility and faster time to market. On the other hand, a more mature innovation is better served by breaking solution into solution components some of which which may form part of other solutions as well and acquiring economies of scale, and giving customers greater flexibility to have a solution assembled with cheap standard components that meets their specific needs.
To take an example, consider solutions in areas such as Web3 and Metaverse say an enterprise learning solution, a distributed payment solution or an immersive command center. Today the technologies for these solutions are not sufficiently mature to meet the needs of potential customers. So it makes sense for one company to build the end to end solution that would include the core technology platform, the custom build, and the strategy to develop the business models around these technologies so that they can fine tune and rapidly meet the needs of the clients. But what we find today is that there are multiple different organizations offering the different components such as the distributed platform, the XR platforms, the headsets, the system integration and custom features for participant organizations, the change management and the business processes development. That could be a reason why we have not yet seen highly successful solutions in these space yet. We could potentially see a strong solution when one big company emerges to develop an end to end solution optimizing all the components of solutions in this space.
On the other extreme , consider something like an ERP solution implementation. It is highly mature and probably meets most of the customers' performance needs. Customers today probably feel each new version is so expensive as are the customizations for their specific needs. They may probably be better served if they have an ERP system consisting of multiple standard components which they can buy off the shelf and assemble together at low cost and quickly unplug outdated components and plug in new components as required instead of waiting for the next mega roll out.
The key aspects for a modular architecture are specificity, verifiability and predictability of output at the interfaces of one component to the next. Where this is not possible an integrated architecture is better.
In the innovation cycles, solutions move from modular to integrated and back. As performance of a solution improves and exceeds current needs of market, things move towards modularization. As new performance requirements emerge and existing solutions fall short, trend will again shift towards integration to enable companies to develop the solution that will meet the performance needs by playing with the various solution components. Computers are one area that has seen rapid successions of these cycles. Performance improves to exceed current business needs. And then new business needs emerge. Then again performance improves and yet again new needs emerge. That could be the reason why one has seen plenty of vertical integration and horizontal stratification happening in succession happening so fast that it almost looks chaotic and directionless. But it could actually be making sense with this cycle happening in different phases in different segments of the marker and different companies leading and lagging the market. It is possible a slow starter company with inertia may be still in the middle of stratification when industry has already moved towards reintegration.