Data Democratisation and its Revolutionising Potential

23 Sep 2021

Technological advancements in recent history have imposed on us various moral dilemmas. Do we sacrifice face-to-face human interaction for the sake of mass connectivity? Do we replace human roles with automated devices? Do we allow Artificial Intelligence (AI) to replicate sensitive human decision-making processes such as medical diagnoses?

Regardless of the side, you land on in the ethical debate, it is hard to deny the countless ways that technology has enhanced our lives, economies and even enabled our survival through crises.

A new and slightly contentious technological trend is ‘data democratisation’. This is the idea that the ever-growing stores of customer, employee, and operational data collected by a company should be accessible by the average employee, not just senior management and data specialists. Employees are given basic training to work comfortably with data and are aided by self-service Business Intelligence (BI) tools.

55% of data across an organisation is unused (Splunk). By expanding the level at which this data is leveraged you can produce more actionable business insights. This idea of being data-driven is championed by experts and is believed to deliver a significant competitive advantage. According to Forrester, ‘by 2021, insights-driven business will steal $1.8 trillion a year in revenue from competitors that are not insights-driven.’

We have recently seen the acceleration of data democratisation and its benefits, especially in underpinning the survival of businesses during the global pandemic. Although, historically it has been the cause for much speculation as society navigates the implications of leveraging something as sensitive as data.

Picture of laptop with data shown

Democratisation- a Company’s Best Friend

‘An important focus when nurturing this type of data culture is explainability.’ Explains Bernard Marr, technology thought leader. ‘Simply, people are more likely to believe in the potential of a new piece of technology if they understand how it works, as well as what it does.’

We do possess the technology to implement data democratisation. Deep data insights can be obtained using AI and Machine Learning (ML) and adapted to the specific needs of industry to launch sophisticated analytical plans. Data-driven business culture has been limited only by our cultural sensitivity to handling data.

When our hand was forced during the pandemic, however, and we had no choice but to deal with remote working and depleted productivity, many of our eyes were opened to data democratisation techniques. The innovation achieved ‘explainability’.

Cloud-based, mobile data analytics were essential for driving collaborative professional experiences during the pandemic. Accenture reports that 64% of companies saw improved efficiency after leveraging data analytics and improving data literacy.

Data plays a particularly fundamental and versatile role in financial services. AI/ML is often used for on-demand anti-fraud algorithms, replacing the retrospective investigative services of the past. This technique significantly averts crises by blocking fraudulent payments. It also has an indirect effect of streamlining compliance processes, allowing new products and services to be created accordingly.

To give an example, telecommunications and fintech company MTN group, launched a programme during the pandemic to make data as accessible to its workforce as possible. Senior Vice President of Product Development at Oracle, T.K Anand describes this project and the significance of applying a data-driven culture to a large company. ‘They were able to start to converse with each other during a really challenging time, with data that was current.’ He continues, ‘Finance was kind of in the driving seat in those tough times, but they could communicate with the business teams in real-time and be all speaking the same language based on data. It wasn’t an easy situation, but they were able to weather the storm.’

By giving business users access to relevant data, you’re essentially allowing them more autonomy to perform their role and generate actionable insights. This increased accountability allows your employees to be more invested in the outcome and more creative in their approach.

It has become blindingly apparent that businesses are losing money by not democratising their data. Immense stores of data with the potential to provide unique insights are left untouched and essentially act as bottlenecks for end-users within an organisation.

Data Governance- Avoiding Anarchy

The mere idea of sharing data is by itself not enough to effectively boost your business’ potential. With great power comes great responsibility and the topic of privacy is not to be taken lightly. Allowing more employees access to Personally Identifiable Data (PII) opens the door to a multitude of risks.

The quality of the average user’s training and technical knowledge is the main issue. By making data more widely accessible throughout a company, people that are less technologically inclined or unable to work with a large volume of material may misinterpret data and make the wrong call.

Data Literacy for All

To maintain a high level of data literacy throughout the company it’s necessary to provide intensive, regular, training through seminars, self-help guides and (to use the problem as the solution) cloud-based learning.

It is also important to limit access to information based on the department or business objective. There should be a hierarchy of access reinforced by adequate security measures ensuring data is encrypted, its sharing tracked at all levels.

A data catalogue is an excellent tool that facilitates ideas through collaboration between ordinary team members and data analysts. It also moderates the use of data through input from different teams.

Machine Learning — The Data Wingman

Neural network, machine learning concept image.

How do we achieve data-driven decision making without data scientists? We facilitate the end-user’s ability to understand and stratify data using AI.

The job of the data scientist is to analyse and clean data, conduct parameters, and design algorithms to make data usable. Automated machine learning makes data palatable for the average end-user and gives them the option to generate their own machine learning models to fulfil their specific needs.

Visualisation

Visualisation is any process by which data is presented visually, i.e., via diagram, chart, or image. Machine learning allows us to group patterns from raw data, essentially accelerating the rate of data discovery. The algorithm will ‘learn’ continually and improve the quality of its analysis as it progresses which makes it ideal for organisations that have a constant stream of data.

The stratified data can be condensed into a user-friendly, graphic dashboard using visualisation tools. Specific parameters can be applied according to whatever data is needed. Ultimately, the average ‘citizen data scientist’ can search for, read, and make decisions based on vast amounts of data.

Predictive Analytics

Machine learning can analyse past statistical data to support predictive modelling. The algorithms respond to new data, determining trends that inform future business objectives. It can be used, for instance, in financial services to detect and reduce fraud and measure market risk.

Keeping your company objectives in mind, to ensure these processes work efficiently you must have appropriate analytics software. You must also have an effective governance program that feeds your algorithm high-quality, coherently formatted data.

The Limitations of Self-Service

The goal is essentially to allow the end-user to analyse data and determine insights without having to learn code or sophisticated data processes. AI facilitates their understanding, allowing the user to create actions applicable to their goal.

However, the sheer extent to which the data is simplified can cause serious problems, leaving room for much human error.

· Multiple data-consuming groups with differing interests can cause several versions of the truth and conflicts of interest when developing actions.

· It is hard to determine the quality of data, conclusions may be drawn from unreliable sources.

· Minimally trained analysts may deploy poorly designed searches causing biases in results.

Although, there are certainly many obstacles to overcome in the pursuit of a big data democracy, its benefits, and more importantly its achievability is explicit. Of course, there will be an adjustment period but the utilisation of our ever-expanding hordes of data by the citizen data scientist seems an inevitable, and even, an essential aspect of the new digital world.

A data-driven business culture, free from gatekeeping and data silos can and will produce a mass increase in efficiency and security. From discerning faults in existing organisational practices, indicating fruitful areas of brand enhancement to detecting and preventing fraud. Putting data in the hands of the people will only improve the agility efficacy of each business unit.

Further reading:

https://www.linkedin.com/pulse/democratization-data-science-financial-services-beyond-bernard-marr/?trk=eml-email_series_follow_newsletter_01-hero-1-title_link&midToken=AQGPZZ48l6faJQ&fromEmail=fromEmail&ut=18VHTRCf0e6pY

https://hevodata.com/learn/a-guide-to-understanding-data-democratization/#tipdd

https://www.forrester.com/blogs/category/big-data/

https://www2.deloitte.com/content/dam/Deloitte/it/Documents/financial-services/Automated-machine-learning-essay_Deloitte.pdf

https://www.ibm.com/blogs/journey-to-ai/2020/10/todays-best-practices-for-embedded-and-self-service-analytics/

Data Analysis Ai Machine learning Sharing