Every two days we create as much data as we did from the beginning of time until 2003, according to Bernard Marr in his 2017 book Data Strategy. Nearly a decade later, this statistic has grown exponentially. Mobile devices, in particular smartphones and the Internet of Things (IoT), have conspired to make creation of data simple, spurred by supposedly free applications enjoyed on account of the benevolence of social media businesses: Instagram, Facebook WhatsApp and so on. Users do not pay money, yet they can chat, upload and view images, comment, and make video or voice calls – all too easy to accomplish with super-fast internet connections.
With every adventure onto the IoT, users leave digital traces of their identities. Companies then harvest user preferences, behaviors, routines and more. Such developments make it remarkably easier to define patterns and calculate probabilities on an industrial scale. Similarly, enterprise data can be leveraged, albeit in more modest terms and with less diversity.
How useful is well-used data? The most valuable companies globally, measured by market capitalization, comprise entities with superior data capabilities. Twenty-five years ago they were all minnows. Big data is now a mechanism of natural selection. Decisions must be anchored more on empirical evidence. Organizations failing to master this competence risk forfeiting market share to more adept competitors.
Forging a viable data strategy becomes essential and demands a long-term view. Below, I address four key aspects to set it in motion.
1. Data collection
Data centers have become lucrative businesses to support cloud computing services. Spectacular increases in computing power and rapid data creation have stimulated storage demand on the cloud. Artificial Intelligence (AI) models thrive on huge data sets. Their accuracy bears a strong correlation with the volume of data they are trained on. Because data storage is expensive, enterprise leaders must have a lucid tactical plan, at a granular level, for which data must be accumulated to deliver the business strategy. Use cases must be defined ahead of collection. Data without defined use cases needlessly increases costs and can expose the enterprise to data breaches and related complexities. Additionally, collecting data does not transfer ownership. Utilization of personal data, especially, is a thorny subject. Obtaining consent to use data as intended is crucial, as evidenced by the Facebook Cambridge Analytica debacle.
2. Data governance
Defining how data is handled and stored is imperative. Data governance triggers critical questions: How do we ensure the accuracy of data? Are there regulatory obligations; what is the scope? Who owns different enterprise data and is accountable for it? What controls exist for access to personal information? Cyber threats elevate reputational risks relating to customer and third-party data. Criminal elements opportunistically breach information technology security systems and use data for ransomware. The impact is multifaceted: disrupted customer service, damaged corporate reputation driven by relentless adverse media coverage, and a distraction to management and the board. With data breaches, expensive lawsuits are often in tow. Information security perils cannot be eliminated. However, management can reduce risk by continuously uplifting threat detection systems and educating staff on basics like phishing emails.
It is irrefutable that data is an asset and clearly must be accompanied by robust data governance to maximize its value.
3. Data architecture
Big data strategies must be predicated on how data is collected, processed, stored and availed for consumption. Appropriate technology infrastructure should be deployed to support the ambitions of management and optimize analytics effort in return for high-quality insights and better decision-making. Can you imagine using a BMW sedan vehicle in a Formula One race? Abysmal performance and disappointment would be inevitable. Poor data architecture easily manifests in inferior data quality, integration challenges between data sources, slow processing and security vulnerabilities. Additionally, every management team harbors ambitions to grow the business. Where solutions are not scalable, costly redesign is unavoidable.
4. Defining success metrics for big data
Data programs entail heavy expenditure on data assets, including infrastructure. Analysis and visualization of insights has led to the emergence of new specialist skills such as data science. Inevitably, additional financial investment is required to secure this talent. Given the extent of upfront cash outlays and effort, business executives are expected to estimate return on investment for such projects. However, like marketing costs, quantifying returns on technology initiatives is often challenging. Businesses are hardly static. Tactics, personnel and the operating environment are always in flux, which increases the complexity of measuring impact in monetary terms. Nevertheless, defining key performance indicators provides a reasonable basis to estimate the impact over time.
While big data is no longer as novel, it still presents a significant opportunity for improvement using empirical evidence. The rapid adoption of artificial intelligence also positions businesses to make better decisions, faster. It is difficult to envision a business devoid of a long-term data strategy as an ongoing concern. However, without posing quality questions and considering unique scenarios faced by each enterprise, big data analytics will not pay a monetary dividend.
