Climbing the data value curve
When investing in data science and big data capabilities, business leaders expect to gain a greater level of agility to find answers to increasingly complex questions. High hopes for insight discovery are driven in part by the overstated claims of tech vendors, and a continuous need to optimise business performance.
Every leader, every worker, and every organisation is looking to move up the data value curve in some way, as shown in Figure 1. The amount of business value that can be generated from data is directly tied to an individual, team or organisation-wide capacity to ask insightful questions of data, and to learn and discover new sources of business value.
When individuals, teams and organisations reach the limits of their capacity to learn from data, then they’ve hit what we call a “value ceiling”. Beyond this point, the potential to discover new insights and opportunities to fuel innovation and growth becomes much more limited without access to new technologies, skills and methods.
So why the value ceiling?
In the era of big data, we have identified five major impediments that limit the ability of an organisation to move up the value curve, which have gone largely unaddressed by existing vendors.
- There is no easy way for people to unearth insights buried deep within complex data.
- Data science models ignore human perspectives, which limits the user’s ability to learn according to their own objectives.
- There is no common visual interface for models made with third party data science tools, such as R, SAS and Python.
- Knowledge and data are generally scattered throughout the organisation or trapped in silos.
- Data science skills can’t easily be learned without going to universities.
It is important to understand these impediments in order to objectively assess the suitability of competing technology options.
Figure 2 is an adaptation of our value curve concept, showing the organisation-wide breadth and innovation outcomes supported by current data intelligence technologies.
Value Curve #1
Many people can ask simple questions of data using self-service Business Intelligence (BI) and Data Preparation (ETL) tools (searching for Known Knowns).
Value Curve #2
Fewer people (often limited to data scientists) can ask harder questions of data using Predictive Analytics technologies and methods (searching for Known Unknowns). Due to the high level of technical expertise required and limited supply of data scientists, climbing this curve is often costly and difficult to scale horizontally.
Value Curve #3
Many people, regardless of their skill level in data science, can ask even more complex questions of data using self-service AI-powered insight discovery tools (searching for Unknown Unknowns). Solutions at the top end of the curve are the holy grail from an operating model standpoint, because they deliver unmatched levels of insight, and promote organisation-wide learning and discovery without dependencies on deep skills in data science.
Taking a closer look at Curve 3, solutions like Salesforce Einstein, HyperAnna, and Stories have each addressed the human learning challenge by making the user interface more predictive, human-like and intuitive. Although features like a “virtual data assistant” aim to make the user experience more “humanly”, in our experience the machine-generated outputs don’t deliver any real insights above and beyond what current-generation BI technologies can already provide. The results remain query-driven, and like with any tool, users can feel underwhelmed if they fail to learn anything substantively new in their data.
So how is DeepConnect® different?
In recognition that most organisations already have investments in data science and/or BI technologies, we designed DeepConnect® in a way that extends an organisation’s existing capabilities, transforming the way business people and data scientists interact with data to solve problems.
DeepConnect® provides a common, self-service, visual analytics interface that allows data scientists and business users to explore and share the outputs of any predictive analytics model made with tools such as R, python and SAS. In other words, black-box models now become full transparent, and open to further investigation by data scientists and business users alike.
Moreover, DeepConnect® accurately captures the way humans learn in real-life. We saw the need to develop AI and machine learning algorithms that can be actively guided by a user’s own learning goals – to generate a diversity of insights and perspectives above and beyond what they already know. DeepConnect® is the only platform of its kind that lets humans and machines actively learn together in this way.
In our view, DeepConnect® is a “game-changing” technology for climbing the data value curve and becoming a truly insight driven organisation. To learn more about our DeepConnect® solution, please contact us or check out our Platform and Academy pages.