APIs -- in their current, open-to-the-world-to-use form -- have been around for several years, being groomed as the swarm of mini-engines that would power the digital economy to its next phase of glory. Except, they haven't quite lived up to their reputation as of yet.
That's the latest word from some McKinsey consultants, who assert that APIs have lots of raw potential, but haven't caught on as intensely as industry watchers have predicted. Why? Their value remains hidden or uncertain.
Some legwork is still needed to make APIs the enterprise-lifting code snippets they were designed to be, say the consultants, led by Keerthi Iyengar, an enterprise architect at McKinsey. They point out "the number of firms with mature API programs remains small. Most organizations have just a dozen or so APIs instead of the hundreds needed for a robust portfolio."
The slower-than-expected uptake of APIs means missing business growth opportunities, Ivengar and his co-authors state, estimating that "$1 trillion in total economic profit globally" could be left on the table by businesses that fail to develop or participate in cross-industry ecosystems. "That makes APIs, which play a crucial role in linking organizations and technologies in ecosystems, a significant competitive battleground capability," they state.
Ivengar and his co-authors outline six ways to extract more value from APIs:
Identify and prioritize potential value. It may seem like an obvious exercise, but business and IT managers need to do more to pinpoint exactly where their APIs can deliver. The McKinsey team recommends analyzing customer journeys as a good start. "One bank pulled business and technology professionals into a joint team and tasked them with identifying where APIs could help resolve several longstanding customer pain points," they relate."Their review revealed opportunities to develop advanced calculator APIs capable of pulling from multiple sets of data, know-your-customer APIs, and product-aggregation APIs that could help customers access needed information more quickly and cut down on form-filling requests. That data-driven approach gave the bank greater mission clarity and built momentum for the API program." Good stuff.
Understand the data. Know the data environment behind the API, Ivengar and his team urge.. "Understanding what it takes to develop the APIs requires a deep knowledge of the data environment, especially back-end systems where the API work is often done. Once the best opportunities are identified, API developers can identify which and how many APIs are necessary to unlock that value."
Manage monetization actively. Actually figuring out how to capture the value is the next challenge enterprise managers face -- "a step many organizations surprisingly tend to shortchange," the McKinsey team states. "Determining what and how to charge, for example, requires quantifying how much the underlying data or service is worth (often based on how proprietary it is and its role in generating value), the revenue streams the APIs open up, and how much developers and users might be willing to pay to access them."
Enable governance. The McKinsey team advises establishing a centralized body to oversee API design and development across the organization. This body -- such as a center of excellence -- "can manage all the APIs in the catalog to avoid duplication, enable reuse, and assist with developer access." The governance team also means maintaining flexibility. Leading API organizations "continually vet and reprioritize their portfolio." The governance team also ensures that APIs align closely with business requirements.
Actively promote usage. APIs need to be treated like anything else the business is attempting to roll out and sell to employees or customers, the McKinsey team reports. Building interest encompasses a spectrum of activities, starting with pilot projects to establish credibility and gain enterprise attention.
Measure and track performance. APIs should also be treated the same as any other business initiative in that adoption and performance should be metered and measured. Ivengar and his team recommend focusing on "relevant usage and traffic metrics, such as the number of user registrations, the percentage of users by customer type, and the number of requests over time." Tracking data errors or API response times is also important.