Creating value from data with AI, APIs, and a business model first approach

There is a data revolution driving the digital transformation movement. Organizations can reap plenty of business benefits in the back office by digitizing data and automating their processes. However, the companies setting the bar in the digital economy are the ones that maximize their use of data to deliver captivating customer experiences. These companies recognize the value of raw data, know how to refine it, and how to disperse it throughout their digital touchpoints. To enable this process, they combine data with AI and APIs. For enterprises looking to emulate these digital pioneers, is it enough to just define a digital strategy that encourages the adoption of data management, artificial intelligence, and API management?

Well, not completely. The thing about digital transformation is that it has become an umbrella term encompassing almost every trend in technology and every innovative approach in business. When confronted with the challenge of defining a digital strategy, IT leaders of an organization in isolation might aspire to go cloud native through microservice architecture while establishing agile methodologies and DevOps culture. Those techniques could improve delivery output, but they are not a sufficient starting point for a business strategy. Defining the digital transformation strategy should be a collaboration between IT and business leaders. However, even the collective perspective of these leaders could become enamored with the innovative possibilities offered by data, AI, and APIs and subsequently descend into dogma, losing sight of business outcomes. Just think about the many unrelated ways that AI and APIs intersect, from machine learning-enabled API security solutions to RPA-enabled APIs to API-enabled AI algorithms. If you try to define your strategy by focusing on the technology movement first, you will probably lose your way.

So where should you begin? After working with companies on their API-enabled digital strategies for a few years, I think I have an answer. I’ve been talking a lot lately about business models in the context of API business strategy. The reason for this is that I’ve become convinced that an envisioned business model — one that has a defined “steady state” but also provides for extensibility — is the most important navigational marker in an early stage business strategy. If you know the business model you are aiming for, you can map out how to target the customers and other stakeholders needed to make it function at scale, and also what capabilities you need to establish to support it. A well-articulated target business model creates a center of gravity for all the dimensions of an effective business strategy.

A new set of values

Now, I’ve been coming at business models from an API perspective. The reality is that almost all business models involving APIs are part of a broader digital ecosystem. In a recent blog post, I talked about how you can view that ecosystem as a value network, a concept defined by business innovation expert Clayton Christensen. Christensen talks about the need to study business models within the context of their containing value networks. Another business strategy expert who emphasizes the notion of value in business models is Alex Osterwalder, creator of the business model canvas. In his words, “A business model is how an organization creates, delivers, and captures value.” Osterwalder’s inclusion of value delivery and capture influenced the value exchange mapping approach to business models I defined in that blog post. Let’s put value exchange mapping aside for the moment, and instead look at how all three value actions — creation, delivery, and capture — work together in a high-functioning digital organization. Specifically, we will look at the role AI and APIs play in creating, delivering, and capturing a unique type of value: data.

There are many types of value that can be exchanged in a value network. There are tangible assets like money, products, and services. There are also intangible assets such as time savings, reach, and risk reduction. Data has special properties that make it hard to classify in either category. Data can be a product/service, data can be a vehicle for an intangible property like trust, or data can even be derived from a value exchange interaction. Regardless, data is certainly the newest and most untapped asset in the digital economy, which is why those organizations that maximize the value of data are the ones finding the most success. So how do they do it?

Data-as-value: capture, creation, and delivery

We can use the Osterwalder business model definition to illustrate data-as-value, and also where AI and APIs fit in. Value capture — listed last in his definition — is a common economic term that usually refers to  the revenue-generating activities in the business model. But if we view data as a currency itself the way digital pioneers do, then value capture — or in this case data capture — is actually the first step in the data-as-value cycle. Data can be captured through customer interactions, event streams, unstructured media, batch files, or any other data source. The key to optimizing the business model is to capture the data as cheaply as possible. And the cheapest data is generally data that is either captured in raw, atomic form, or perceived by the data provider to have little value in the exchange context. For example, an Instagram user may have no inhibition about sharing a photo with their followers on Instagram, not thinking of it as providing image data to Facebook. In a mature digital organization, the data arrival rate is so high that inbound processing is kept to a minimum. APIs could be a point of data capture, especially in user interactions, but there will generally be many non-API sources of data as well. Once the data-as-value is captured, data-enabled value creation can begin.

In a previous era of data processing, much work was done to structure and normalize data into data warehouses. In the current era of big data, the scale of structured data makes normalization impractical if not impossible, and much of the captured data is unstructured. The good news is that innovation in analytical tools means all data can be processed, regardless of whether or not it is structured. It is the field of artificial intelligence — especially machine learning — that makes this possible. In fact, machine learning depends on copious amounts of data in order to train its models and refine their accuracy. Training these models is incredibly resource intensive from a storage and compute perspective. They require all data inputs to be centralized in large datasets (stored in data lakes or cloud storage) and powerful compute (CPU clusters or GPU units) to process efficiently. Once the models have been trained, they can be deployed in a more federated way. The inferences of these distributed models can be applied in multiple value-creating ways, from time series-based predictions to image recognition to customer targeting and recommendations. The more data can be correlated and contextualized, the more potential value it holds. But that potential is not released until it can be delivered to consumers.

Companies that differentiate themselves in the digital economy do so through customer experience. That means reaching their customers through all possible digital channels, and providing seamless experiences that solve their problems and meet their needs. The insights that inform these experiences are produced through AI-processed data, but they won’t reach customers on the myriad channels until they deliver these insights through APIs. APIs are not the most efficient mechanism for ingesting bulk data, but they are the de facto means of articulating data as business capabilities and reaching customers through systems of engagement. It is not an accident that the companies synonymous with customer experience-based digital disruption — from Amazon to Netflix to Uber — are prolific API providers and consumers. In the context of business models, APIs are their conduit for value delivery. 

Fig. 1: A virtuous cycle of value involving data, AI, and APIs.

The rule of three: data, AI, and APIs

This digital rule of three — combining data with AI and APIs — is definitely a strength of digital giants like Amazon, Facebook, and Google. The good news is that established companies are reaping benefits as well. Consider BMW, a 100-year-old company that has become a world leader in producing connected cars. The World Economic Forum estimates that a single connected car produces four terabytes of data each day. BMW captures this data, then analyzes it to produce actionable insights around fuel consumption and maintenance needs. This information is packaged and delivered through APIs to power driver UIs and provided consent-based integration with third parties, like auto repair shops and insurance companies. Data is revolutionizing every industry, and the digital rule of three is applicable in all.

Three being the magic number, here is a trio of tips on how to harness this approach in your own digital strategy:

Consider your data, AI, and API strategies together

My colleague, Andrew Dent, wrote a blog recently about not confusing your data strategy with your API strategy. Another colleague, Hugo van den Berg, talks about the potential of AI and where APIs can help. Both posts underline the need for data, AI, and API strategies to be harmonious. To know what data you need to collect, you need to know what you’re going to do with it. To know what you’re going to do with it, you need to understand how it will benefit your customers. Make sure none of these strategies are defined in isolation, and — as mentioned earlier — make sure they are grounded in business need.

Ground your strategies in an envisioned business model

By aiming your strategy at a target business model, you will naturally need to address your data, AI and API strategies together. More importantly, you will ground your strategy in intended business outcomes, and create a common ground for business and IT leaders to strategize together. Mapping out the business model by showing how value gets exchanged in your digital ecosystem, as well as how value gets created within your own organization, will help you determine how to get to your goal, and even give insight into how you can innovate and grow once you get there.

Find your “data-as-value” opportunities

As mentioned, data is the biggest untapped resource in the digital economy. The good news is that your organization probably has lots of data. It may be tempting to just slap an interface on that data and wait for paying consumers to arrive. The reality is that refining that data into exchangeable value requires an iterative process. Think about what data you have, and how it might be contextualized and correlated to augment your direct or indirect customers’ experiences. But don’t stop there. Step into the customer’s shoes and think about their overall jobs-to-be-done, apply design thinking, engage in rapid prototyping. These techniques should help identify opportunities to synthesize more data on hand, or capture new data that you can weave into the experience. Consolidated customer information is especially valuable. I recommend checking out David Rogers’ book, The Digital Transformation Playbook. He includes a chapter on how to turn data into assets.

The digital era of business has arrived, and with it come many digital distractions. The best way to navigate through this storm is to create a business model beacon to illuminate your forward path. Expansive amounts of data, mind-bending ML algorithms, and web-enabled APIs can distract you or be a tool you use to disrupt your digital ecosystem. These techniques can drive your innovation engine by using them to capture, create, and deliver the value that defines the essence of your business model, and therefore your business.

Thanks to Radu Miclaus and Gib Bassett for their help in shaping this post!



We'd love to hear your opinion on this post