Data Science Value Chain

Creating insight form data requires a mindset. Silos are hard to escape, whether they are technical or conceptual. To holistically exploit the opportunity presented by Artificial Intelligence (AI) and Big Data tools and architectures; a new way of thinking is needed that frames data as a raw material of business.

The answer is to focus not on the functional components—what organisations do to the data—but on business outcomes and how they can be achieved—what they do with the data. This novel approach can be cultivated by looking at the data science value chain. From discovery and ingest through explore and analysis, generate insight and disseminate, automating insight generation and augmenting human intelligence with insight.

The following diagram depicts the data science value chain that an organisation can adopt in order to reach the pinnacle of AI revolution. The steps in the value chain are not linear in nature and not always dependent on the previous steps. The best way to harness value from data science projects often depends on the length of the projects and where possible by parallelising multiple steps. The entire value chain is overseen by the data governance process but is not discussed in this article.

Build a data foundation. To become a data-driven organisation, the first step is to either build new data foundations or fix the existing data foundations. There are, typically, three areas to focus:

• Understanding the data. Organisations need to understand the data they have or the data they need to develop or acquire – and the requirements to build a solid foundation and drive data-based insights.
• Organising that data. After they understand and procure the right data, they need to streamline, organise and integrate it – which typically involves data in the cloud, data lakes, and other emerging data storage and integration technologies.
• Building trust in the data. It needs to be transparent and secure, with good data governance, data quality and data cleansing processes in place. According to a recent survey by Forrester Consulting on behalf of KPMG on 2,165 data and analytics decision-makers in 10 countries, only 60 per cent of organisations say they are not very confident in their analytics insights.

Primary roles involved building data foundations include data architect, data engineer, data modeller, data developer etc. Though this setup is best run either by chief data office, chief information office or chief technology office, regular inputs from business, especially around validating the model and data quality improvement, is pivotal.

Executive insight. When building data science products, first set of insights should be built for the senior executives of the organisations.

The same study also reported that only 51% of the respondents believe their C-suit executives fully support their organisations' data and analytics strategy.

Organisations must think about trusted analytics as a strategic way to bridge the gap between decision-makers, data scientists and customers, and deliver sustainable business results.

Develop trusted insights. One practical step to address the trust gap is to build the data foundation and generate insight in the form of dashboard and visualisations.

KPMG recommends organisations address seven key areas to close the trust gaps:

1. Assessing trust gaps.
2. Creating purpose by clarifying goals.
3. Raising awareness to increase internal engagement.
4. Developing internal data and analytics culture.
5. Encouraging greater transparency.
6. Providing a 360-degree view by building ecosystems.
7. Stimulating innovation and analytics R&D to incubate new ideas and maintain a competitive stance.

By providing trusted and actionable insights, organisations can gain support for data science from senior executives. Organisations looking to make the most of analytics must recognise that it takes time to build trust, and that success requires the involvement of the entire company towards a data-driven culture.

Primary roles involved building insight analytics products include data analyst, business analyst, business intelligence analyst, data storyteller, UX consultant etc. These roles can sit as a dedicated data science team or as part of individual business units.

Management insight. The second set of data science products should be built for the management staff of the organisations.

Analytics are becoming increasingly integral to business decisions. And by empowering the management to take informed decision, organisations can reduce decision biases at a greater rate than just empowering the executives with analytics insight.

Explore & analyse data insights. Once management staff are getting used to analytics insight, organisations should invest in more than just descriptive analytics. By conducting exploratory and diagnostic analysis on historical data, organisations can discover deeper insights.

Exploratory and diagnostic analysis provides utmost value to any business by helping analysts understand if the results they’ve produced are correctly interpreted and if they apply to the required business contexts. Other than just ensuring technically sound results, exploratory analysis also benefits stakeholders by confirming if the questions they’re asking are right or not.

Exploratory analysis often turns up with unpredictable insights – ones that the stakeholders or data scientists wouldn’t even care to investigate in general, but which can still prove to be highly informative about the business.

Semi-operational insight. At this point, organisations should often use data to inform action and make decision. Not only operations as a business unit, but also every business unit should start to use, trust and getting reliant on data science products.

Decisions can be taken at a much faster rate. Requests to create more data science products are being made from different business units. This is when organisations expand their data science and analytics capability. If not already established a data or analytics office, this is also an ideal time to do so.

Predict what will happen next. Organisations should consider having an advanced analytics engine to run predictive modelling. As predictive analysis help organisations to prepare for the future demand based on historical data. As a result, business can deliver superior customer experience by analysing what they would be needing in the near future.

Another advantage of predictive analytics is that businesses will be able to maintain operational efficiency by dispelling uncertainties. It would help them avoid cost consuming activities like lead times and unused inventory.

New roles will be needed at this point to build predictive models such as data miner, predictive modeller, predictive analyst etc.

Operational Insight. Operational insight is all about unlocking the power of information and providing real-time analysis for business benefits. Data is utilised properly for all decision making and action across the organisation. Dashboards and reports will be available for every business unit helping with their daily activities.

Insight will be leveraged to drive internal business operations and encourage differentiation with customers and partners. Organisations no longer suffer from not being able to distribute operational information in a timely manner. ROI will be measured by usage and availability of data within a data-driven system.

Data is a game changer in driving positive business performance. Usage is the metric that every CFO should review. Not just the usage of information and reports but the extent to which data is accessed, discussed and debated. A reasonable sample of customers, products or channel data that is relevant and available for a business team to use should be the threshold for “accurate” analysis.

This is the time we see organisations move towards a decentralised data team (if they already haven’t done so) by embedding data analysts and scientists into every business unit so that they can take ownership of their own insights.

Prescribe next best action. Not just predicting what will happen next, data science model also advise what action should be taken by the business.

Prescriptive analytics will facilitate further analytical development for automated analytics where it replaces the need of human decision-making with automated decision-making for businesses. For example, automated analytics will be able to use applications to choose the best marketing email to send to customers instead of needing a marketing director to make this decision.

The prescriptive analytics market is also growing exponentially and is expected to increase by 22% between 2014 and 2019 to $1.1 billion. Moreover, it is projected to be built into business analytics software by 2020.

Automated insight with Cognitive and AI. Automated analytics is the next stage in analytic maturity. Automation will emerge to address the shortage of data scientists, and perhaps completely change the way in which analytics is implemented. Systems and algorithms are rapidly evolving that enable dynamic learning from streaming content, thus automating the learning from such data sources.

When dealing with large volume of data spread across an organisation, manual analysis is slow and prone to errors. By contrast, automated systems keep the data analysis process running without a hitch, so organisations can get straight to the data to reveal usable insights. Automated analytics can break down old barriers by automatically compiling data from different sources and converting them to the same format, preventing new silos from forming and making the most up-to-date data accessible to everyone who needs it.

Automated analytics is essential for keeping track of the many sources of data modern organisations use today, ensuring data scientists don’t waste time working with bad, out-of-date, or incomplete data. With a more streamlined data analysis process, important opportunities can become apparent, introducing agility to big data analysis and, ultimately, increasing the organisation’s business intelligence and competitive edge.

Machine learning engineers and machine learning programmers are the new roles needed by the organisations to achieve such automation.

Augmenting human intelligence with Cognitive and AI. Automation, however, falls short for solving most important business issues, which tend to have unique and idiosyncratic goals, processes, and contexts.

Driving business strategy—what do we need to do to please customers or to grow or expand the business—is not a decision that relies exclusively on a predictive model. Business strategies rarely fit into neat, binary predictions, and they almost never have easy-to-implement solutions like an on/off switch.

That's why, fundamentally, analytics is a managerial and leadership problem – not a data science and IT problem. Managerial processes are difficult to automate because the connection from analytics to action can mean very different things across different contexts. Context is key: managerial actions typically require context-specific judgment and content expertise, which are lacking in analytics alone.

Analytics automation can do an exceptional job of finding trends and building models, but it can't solve complex business problems. That still requires deep knowledge of the business context and enough knowledge of data science to understand the role of trends and models in making business decisions.

Therefore, the top of the line automated analytics products will have a way of augmenting human knowledge before a decision can be made. Successful businesses will be the one to build clear bridges between data science and business strategy, so the former can inform the latter.

To succeed with analytics, business leaders must be conversant (or even fluent) in data analytics and they must ensure that their data scientists are immersed in the business context.