It is a capital mistake to theorize before one has data.

Sherlock Holmes

Data is often referred to as the new oil, and for good reason. It is the critical fuel for your business’s new economic engine and your digital strategy.

Most pre-digital incumbents have a data value problem. That is, they struggle to gain business value from all the data, or any of the data, they are collecting. This mass of data has become known as big data, with many companies investing heavily in the infrastructure and skills required to support it. According to Dresner Advisory Services’ 2017 Big Data Analytics Market Study, 53 per cent of global companies reported that they ‘use big data today’, compared with 17 per cent in 2015. Notably, there is still regional variation in the use of big data: North America (55 per cent) narrowly leads the Europe, Middle East and Africa region (53 per cent) in its current levels of big-data analytics adoption. Asia–Pacific respondents report 44 per cent current adoption and are most likely to say they ‘may use big data in the future’. However, according to a 2016 survey by technology research company Gartner, of the companies that had invested in big-data capabilities, only 15 per cent reported a live big-data project. This reflects the lack of capability required to drive value from data. Known as the data value gap (see figure 4.1), this is either your biggest threat or your greatest opportunity.

This chapter will help you focus on data value and opening up opportunities to drive business growth. You’ll discover three key things you need to know about data, and what you can learn from car manufacturer Tesla. We’ll also discuss the application of thick data, present a five-step framework to harness and apply data, and reveal three key areas where data can drive growth.

Three things you need to know about data

The rise of big data, and the associated realisation of its importance, has been accompanied by a major misconception: the idea that data volume alone is the most important consideration. Having more data is great, but it isn’t everything. The value and application of data to your business strategy is what really counts. There are three factors you need to consider when thinking about data within your business.

GRANULARITY

The granularity of your data is instrumental in helping make insights not just directional but specific. Granular data helps you bridge the gap between your strategy and your execution. For example, when you know your customers’ buying habits and preferences in detail, you can focus on specific customers in specific circumstances within your marketing campaign. Data granularity directly impacts the accuracy of each prediction and your return on investment.

SOURCE AND THE CONNECTIONS WITHIN

Having customer, transactional, supplier or usage data sitting in silos makes it difficult to use. Connections between related types of data help you answer questions like: which customer will buy if I run a promotional offer on this product? To do this, you need to know your customers’ shopping history, as well as their profile and preferences, and the correlation in style between the product you want to sell and the products you’ve already sold. The answer may be hard to conceptualise and difficult to intuit, which is where data science and data modelling come into play. Both data science and data modelling are terms used to describe the connections between data and the process of gaining insights from it. That said, if you are going to build these models, your data needs to be centralised and accessible, which for many pre-digital incumbents is not the case. This process will be explained further in this chapter.

QUALITY AND COMPLETENESS

There’s a saying, ‘Garbage in, garbage out’, that captures it. If the data you use is incomplete or inaccurate, the resulting predictions will be useless. There are a number of techniques for machines to help improve data quality or statistical methods and mitigate the effect of outliers. As a general rule, however, you should focus on more than just data quantity — quality is also very important.

High-quality data is data that meets the needs of the analysis required, and accurately represents real-world information or constructs. Low-quality data can have a significant impact on a company’s strategy, resulting in higher operating costs, lower customer satisfaction, poorer decision making and diminished internal trust around decisions.

Over the course of the rest of the chapter we will explain how these factors influence the way in which you approach data and the strategy you employ.

Data on wheels

What happens when a company builds its entire business model around data? Well, if you take a look at Tesla, one of multi-billionaire Elon Musk’s most reputable companies and a considerable data aggregator, you’ll see that data has played an integral role in defining the company’s success.

Tesla, which, as of April 2017, is the United States’ most valuable car manufacturer, heavily employs a data-driven mindset. Since the introduction of the electric Model S in 2012, Tesla has in fact become one of the most valuable companies in the world, and has forced a rethink of strategy for Volkswagen, Renault-Nissan, Hyundai-Kia and General Motors, the automotive industry’s traditional powerhouses over the past few decades.

When you think of car manufacturers, you most likely think of mechanical and automotive engineers. Tesla is a far cry from that, with a huge team of software developers and testers who work across a range of projects, including the development of in-vehicle systems, driverless technology and solar panels.

Tesla cars self-report data that is recorded and transmitted to central servers. This data has a number of uses. If, for example, vehicles are detecting faulty pumps, this information can be processed and a model built that can predict a specific pump failure before it occurs. The car will then notify the driver, who will be prompted to visit a Tesla mechanic before any problem arises. If an issue is critical, Tesla will save time by sending a mechanic to the owner’s home. In the future, driverless cars will automatically drive themselves to repair shops for diagnostics and troubleshooting while not in use.

A Tesla vehicle is a data magnet. The company’s ability to redeploy data through operations to create a better, safer driver experience is an extremely powerful competitive advantage. No other car manufacturer has such a data- and software-driven system in place.

Data not only improves the driver experience, but allows Tesla to continuously optimise its supply chain by identifying opportunities to improve its vehicles, which could involve replacing a particular supplier, using a different model of a certain part or even re-engineering. Further, the data is useful not only in identifying what to do, but also for understanding the underlying reasons for failures. In some cases, it may be due to poor-quality materials, while in others it may be a human error or a shortcoming in the process itself. This is essential, because if you understand the root cause of failure, it’s much easier to find a solution to the issue.

The key lesson from Tesla is that a data-driven mindset has not just changed the product, but has begun to reshape an entire industry. Studying this new business model, you will get the sense that everything Tesla does is fact-based and designed to optimise an overarching set of objectives — sales, product quality, maintenance costs and supply chain efficiency. Uncovering and satisfying customer needs becomes more than an imperative — it is the norm. With a data-driven approach there is also nowhere to hide, so decisions are made with greater levels of transparency.

The application of thick data

In 2015, the big-data industry was worth $122 billion. The International Data Corporation expects this to grow to more than $187 billion in 2019, an increase of more than 50 per cent over the five-year forecast period. These figures clearly demonstrate that businesses are spending big on big data. And yet, according to ethnographer and big-data expert Tricia Wang, 73 per cent of big-data initiatives in 2015 were not profitable.

Examples of where big data is applied range from optimising delivery logistics to helping manage genetic coding issues. Not all systems can be neatly contained, however, and when you are dealing with variables that change, especially when humans are involved, outcomes can be unpredictable. Take the stock market, for example. For decades, people have searched for patterns in the stock market that could lead to riches, but the key has proved elusive both because the system itself continues to change and because humans, who are highly erratic and unpredictable, are involved. There is an important distinction between quantitative and contained systems versus dynamic systems. Quantitative and contained systems do not change, and typically they do not involve humans. A dynamic system, on the other hand, does change and is influenced by human behaviour — an infinitive variable. The whole, evolving field of behavioural economics exists to explain and predict human irrationality; for our purposes here, it is enough to flag that people — your customers — introduce a mercurial element to any data model. As astrophysicist Neil deGrasse Tyson said in 2016, ‘In science, when human behaviour enters the equation, things go nonlinear. That’s why physics is easy and sociology is hard.’

When humans and big data clash, a paradox emerges. Big data answers questions that humans would otherwise find hard to resolve. However, predictions from big data raise more questions when humans are involved. In addition, a bias can occur where humans unconsciously value the measurable over the immeasurable. The healthcare industry offers a good example of this. Patient wellbeing, especially when psychological in nature, can be difficult to assess. Nonetheless reviews of hospitals and care providers are conducted, and practitioners are punished or rewarded on quantified performance.

When dealing with data, it’s important to recognise that not everything that’s valuable is directly measurable. To ensure valuable insights don’t slip through the gaps, your big-data systems need people who can gather thick data — that is, precious information from humans, stories, emotions and interactions that cannot be quantified. Consider enlisting the help of an ethnographer — someone who can take a scientific approach to observing a culture or group — to help find this information. Thick data can come in small sizes, such as a brief statement from a user about their experience with a new device, but it can also be deeply insightful, covering many perspectives related to the experience. Most importantly, this qualitative material helps provide the context around your big-data models. Big data is able to provide insights at scale through standardising and normalising aggregated data, whereas thick data can help rescue the context and ‘human’ element that is lost from making big data usable. When you integrate the two, you get to ask the million-dollar question: Why?

For example, Netflix was able to leverage both big data and thick data to provide a better experience to its user base. Netflix was more successful when it hired a thick-data expert — someone who knows how to ask questions of customers related to what they need and to test what they want. In doing so, Netflix discovered its customers love to binge-watch TV and they don’t feel guilty doing it. This led Netflix to redesign the viewer experience to encourage binge-watching by releasing the entire first series of House of Cards in 2013. By leveraging this thick-data insight, Netflix not only improved its business, but also changed the way its customers watch and consume content.

A framework to harness and apply data

Data should not merely support your business — it should play a strategic role and provide value as a driver of growth. We’ve identified five steps to achieving this outcome, known as the Data Value Framework (see figure 4.2).

1. START WITH STRATEGY

In part I of this book, we discussed the strategies for success in our new economy, the organisational design you need to execute and the culture required to sustain your approach. Now think again about strategies for success. What insights from data will help differentiate your product or service or provide you with a competitive advantage? Consider the data you need to make money and drive your economic engine today, then the data you want given an alternative value chain. Thinking like this will help you prioritise the data you need.

2. IDENTIFY AND VALIDATE SUPPORTING SCENARIOS

Step 2 involves taking the ideas captured in step 1 and grouping them into specific actions, or use cases, your organisation can realistically undertake. For each of these scenarios, you should identify not only the sources of business value, but also the implementation challenges that are obvious at this stage. Map out each use case using a consistent template; the widely circulated ‘business model canvas’ (see figure 4.3, overleaf) is one type of quick synopsis that can help you compare the relative strengths, weaknesses and potential impact of each scenario to address your strategic priorities.

Throughout this exercise, you need to consider factors such as risk, privacy, levels of access, availability of data, ease of accessing accurate information, and how it should be organised for your end purposes. To move forward from this point, you need to be keenly aware of what it is you want to do, and what you need to know to do it.

3. PRIORITISE BASED ON RETURN ON INVESTMENT

You may find this step the most difficult because it requires you to do two things that many large companies struggle to do: prioritise and focus. Prioritising and focusing are not popular concepts for many organisations because of political and organisational pressures. As discussed in chapter 2, consensus is hard to achieve when there are many agendas. Articulating alternative scenarios in step 2 will help you lay out a logical progression of steps to build your big-data capabilities through a series of initiatives through which data and analytics are increasingly woven into the fabric of your business operations. Trying to do it all at once will only end in wasted resources and lacklustre results.

If you can convince your organisation to build out a big-data business strategy one use case at a time, it will enable you to become an expert at harvesting data, building analytics tools relevant to your organisation and applying those tools to subsequent use cases. Seek opportunities to demonstrate ROI in each step of the process, both to manage your stakeholders and as a way to gauge the impact of your initiatives.

4. GET THE DATA

In step 4, you need to work with your team to brainstorm different data sources that can support your top-priority use cases. Step 2’s focus on implementation and potential challenges should have laid the groundwork for these conversations. In most cases, data gathering will be an iterative process, because data science naturally focuses on the variables that better predict business or operational performance. Your team will learn as it goes what data proves most valuable for your end purposes. Ensure business stakeholders can collaborate with the data science team to identify and test different data sources that might yield the best predictive models.

5. DETERMINE THE ECONOMIC VALUE OF YOUR DATA

The final step involves linking the financial value of your new strategy with the data sources and the predictive capabilities necessary to successfully execute each use case. The financial value of each of your new data-driven projects becomes the basis for appraising the supporting data sources and their worth to the business. This valuation will ultimately drive metrics and targets associated with capturing and using this data, allowing you to track the overall success of each data-value capture initiative.

Three areas where data can drive growth

Now that you have a framework, this section is designed to help you identify the areas of your business and the associated data you should be prioritising.

Your business’s worth in financial terms is the discounted value of its forecasted cash flows. As Jeff Bezos rightly points out, ‘It’s the absolute dollar free cash flow per share that you want to maximise.’ So, building on this, what data do you have, and what can you do with this data to generate additional future cash flows? Our experience shows there are three primary areas of focus that, if seen to, will ensure future free cash flow:

• improving decision making
• improving operations
• monetising data as a valuable asset.

Figure 4.4 illustrates these areas and their impact on your growth agenda.

Before we dive into these three areas, here are two case studies to help contextualise them.

Companies are now being bought and sold based on the value and nature of the data they have. In 2015, IBM announced it was acquiring most of The Weather Company, which owns weather.com and Weather Underground, for a reported US$2 billion. Why? For the company’s data. Its weather-related data sets are vast, including data from three billion weather forecast reference points, 50 000 flights and more than 40 million smartphones per day. It’s no wonder nearly three-quarters of The Weather Company’s scientists are computer and data scientists, while only one-quarter are atmospheric scientists and meteorologists. Now IBM has access to and owns the data, meaning it can monetise it by selling it to companies whose day-to-day activities are impacted by the weather. Weather data has uses well beyond the obvious ones of agriculture and transportation. The weather is known to affect consumer shopping behaviour, employee wellbeing, auction prices and general productivity. We don’t yet know what kind of value can be squeezed out of many of these large data sets. For those who know how to interrogate the data, future applications for findings are limited only by the human imagination.

In 2016, Microsoft purchased LinkedIn for US$26.2 billion, giving Microsoft access to LinkedIn’s networks of more than 400 million users, as well as the associated data. Let’s consider the characteristics of the data Microsoft has acquired and can use. First, there’s the fact that the data is time-series based. An individual’s activity on LinkedIn (posts, shares, likes, connections, profile updates) changes over time. This can help any business gain insights of a directional nature — an individual’s propensity or likelihood to change jobs, for example. This data is also useful in order to target advertising or identify thought leaders who can influence buying patterns. With a small degree of creativity, it’s not hard to see where and how the data can be valuable. Finally, there’s the value of LinkedIn data in improving the features and value proposition of Microsoft’s existing products, such as Azure, Skype, Office and Outlook.

Both IBM and Microsoft are making significant investments in data, and the underlying infrastructure and skills to leverage this data, allowing these advanced technology companies to take on bigger, more ambitious data projects. If you are only just starting out, from a prioritisation standpoint, you are going to be challenged (and maybe a little overwhelmed). There is an explosion of data available and, through applying some of the ideas in this chapter, you’ll have a mass of opportunities to pursue. As tempting as it is to do everything, you’re better off picking a few key combinations to ensure you are competitive. Think quality over quantity. With that in mind, here’s an outline of three ways you can use data to drive growth.

USING DATA TO IMPROVE DECISION MAKING

You already know that data can help inform better decisions about your business. In this section, we will examine the process of humans interpreting data in order to make better decisions.

There are four key areas of your business where data can improve your decision making: finance, internal operations, people and customers. Determine which of these seems to be underperforming, and define what could be done to improve performance in that area. This may involve a comparison metrics in these areas to an industry benchmark to determine where to focus. Then ask questions that relate to this objective to guide you on where to source the relevant data. If you are looking for a place to start, here are some key areas to consider, along with questions to ask:

• Predictive maintenance. What machines are we using? What failures typically occur? What does the failure process look like? What failure indicators exist?
• Call centre routing. What do the most frequent calls relate to? Who has the knowledge base to deal with these calls, and where are they located? What information do they need to know to deal with these calls? What is the root cause of these calls?
• Demand forecasting. When are our busiest months? What are our most popular items? Where are our bottlenecks in delivering our most popular items? Where is the highest demand for our most popular items?
• Supply chain optimisation. Where are our biggest supply chain bottlenecks? Where is our biggest demand? Where are our biggest costs in the supply chain?
• Fraud or theft identification. Where in the business are we most susceptible to fraud or theft? What measures do we currently have in place, and are they effective? What are the most common fraud or theft issues in our market sector, and have we accounted for them appropriately?
• Location or area planning. Where is the biggest demand for our items? Which areas will drive the biggest growth of our key target market over the next year, five years and ten years? Do we have enough warehouse space to support growth?
• Employee engagement. Which division is the most productive? Are our employees happy at work? Is our decision-making process slow and ineffective? Do employees enjoy being at work? Do teams collaborate with one another?

Once you have the necessary data related to your objective, you need a way to identify and then communicate the insights you garner from this data. The good news is that there are now excellent and inexpensive tools that allow you to communicate insights in a more effective way, including Tableau, Microsoft Power BI, Qlik and many others. A spreadsheet is unlikely to be an effective tool, especially if you need to ascertain buy-in from stakeholders, as its visualisation capacity goes only so far.

Don’t underestimate the effectiveness of thick data to counterbalance the insights you have gleaned from big data. For example, an organisational transformation may be needed, but if not undertaken properly, it can undermine a company’s culture and leave you in a worse off position than when you started. This type of information is more qualitative than quantitative, and is therefore not something your big data would warn you about. You can mitigate this risk by engaging thick-data experts to work alongside you and your leadership team to understand the most effective path forward with respect to your transformation and, where issues may arise, to help ensure success.

As an effective leader, you know the power of storytelling in helping spread lessons quickly throughout your organisation. Balancing thick-data and big-data insights provides a narrative you can use to help share the fact that your insights are grounded and substantiated, taking into account strong data signals, as well as what people feel and perceive.

IMPROVING OPERATIONS

In the previous section, our focus was on how data can help humans make better decisions. Machines, however, can use data to help your business run more efficiently — from the warehouse to recruitment, service delivery, customer services and everything in between. Machine-to-machine communication is a key element of this, and something we will explore further in chapter 6.

As with using data to improve decision making, you need to look at your operations to identify which areas to focus on. Examples include:

• assortment optimisation. Software solutions can ensure you have the right goods in the right place through leveraging customer purchasing data.
• cross- or up-selling. Software can make tailored recommendations based on the data gathered as customers interact with you.
• automated stock control. Warehouse and inventory management solutions track your warehouse inventory in real time, and can reorder when stocks are running low and order more when demand is expected to be higher.
• fraud detection. Fraud detection tools can identify malicious anomalies in your data and recommend or automatically implement appropriate responses.
• predicting customer churn. Data in customer relationship management (CRM) software can identify patterns in customer behaviours, including historical purchases and last points of engagement, and based on analysis across your whole customer base, correlate certain events with future actions, like defection or negative reviews. With such insight you can set up response systems, for example automatically offering a special deal to recognise loyalty and reduce churn to competitors.
• dynamic B2C pricing. Pricing algorithms dynamically change the price of a product or service through leveraging data models based on overall demand.
• value-based B2B pricing. Algorithms can change the price of a product or service based on the perceived value of the product or service to the customer.

The best opportunities for data-driven decision making in your business will be a function of your strategic objectives, industry, competition and budget. The key is to think first about the greatest potential ROI, and to couple this with a clear-headed appraisal of your available high-quality data.

MONETISING DATA AS A VALUABLE ASSET

Across an increasing number of industries, there’s a growing opportunity to monetise data in a structured and curated form. Quality data sources are hard to come by but increasingly valuable to help enhance artificial intelligence applications or big-data analysis initiatives. Many companies are willing to pay for data sets they deem valuable, and you too have the opportunity to purchase valuable data where it will enhance your organisational objectives. Parameters such as data volume, time series, frequency of access (download) or data granularity can be used to vary subscription characteristics. For example, your company might provide access to a particular data source, including all historical data and with detailed levels of granularity, for a price of $1000 per month, or only current and future data for $500 per month.

Data-as-a-service opportunities can be developed in partnership between public and private organisations. For example, the cloud-based analytics software business Tableau facilitates access to hundreds of third-party data sources that can be examined by licensed users in combination with their own data sets. This service, combined with proprietary data visualisation tools that allow users to commercialise and extend access to the data, have the effect of increasing Tableau’s user base and the volume of data available for its subscribers.