Data Warehousing, BI, Big Data & Data Science for Data Citizens

Executive Summary

This webinar covers the central role Data Literacy plays in organisations, and why effective communication and active listening matter so much in data processing. Howard Diesel explores the collaboration between business stakeholders and data professionals in Data Modelling, and works through the complexities of data storytelling and the ethical considerations involved in Data Management.

It also looks at the future of business applications and digital decision-making, along with the challenges of integrating AI into these contexts. Howard spends time on the real impact Data Governance and quality have on analytical outcomes, and the webinar closes by making the case for developing competency frameworks in Data Management and assessing the business benefits and ROI that come with AI initiatives.

Webinar Details

Title: Data Warehousing, BI, Big Data and Data Science for Data Citizens
Date: 21 November 2024
Presenter: Howard Diesel
Meetup Group: African Data Management Community
Write-up Author: Howard Diesel

Data Literacy in Organizations

Howard Diesel opened the webinar by asking whether anyone in the audience works at an organisation with Data Literacy programs already in place for new employees. One attendee shared that her organisation has set up an Enterprise Information Governance forum, made up of senior managers, data stewards, and data owners, to boost Data Literacy across every department, with a focus on improving knowledge of Data Governance and privacy in a highly regulated environment. Monthly compulsory literacy training, roadshows, and workshops on privacy and governance keep participation and engagement up, and employees are encouraged to identify and report Data Quality issues, which builds a culture of awareness and continuous improvement in how data gets handled.

Figure 1 Data & AI Competency Framework

Data Literacy and Data Management

A Data Literacy program covers various aspects of Data Management, Data Quality, privacy, and governance among them. Howard noted that while “Data Literacy” works as an umbrella term, it runs into public perception challenges, since some business professionals feel the word “literacy” implies they struggle to understand data visualisations. Examples like box plots reveal real gaps in understanding, which is exactly why Data Literacy training should not only focus on visualization skills but also encompass broader Data Management skills.

Howard shared an instance where a business professional pushed back, feeling that the emphasis on Data Literacy overlooked business and financial literacy, and argued that genuine understanding of data only comes when data professionals grasp business language too. That sentiment points to the need for a more holistic approach to data competency, sometimes called “data abilities.”

Data Literacy and Its Importance in Organizations

Data Literacy is increasingly recognised as essential for organisations. Howard’s point is that it comes down to understanding and working proficiently with data, interpreting reports from business intelligence (BI), artificial intelligence (AI), and data visualisations. What makes Data Literacy matter so much is that it bridges the gap between newer, data-savvy generations and those who didn’t grow up with advanced data tools, building a culture of understanding and skill development along the way.

Building an effective data culture means assessing current competencies and addressing areas for improvement, Data Quality and Metadata included. Proficiency tests can help organisations identify skill levels among employees, which supports strategic allocation of resources toward Data Strategy goals. Without a skilled workforce, even the best Data Strategy risks falling short.

Data Modelling with Business People

Effective communication about data and Data Management matters a great deal for engaging business stakeholders, and it helps to summarise the significance of Data Modelling in simple terms, two or three sentences ideally, so business professionals actually grasp why it matters. That means distinguishing between the various types of Data Modelling, dimensional, relational, Data Vault, and being clear about why acquiring these skills benefits them. Plenty of people, though, have struggled to convince business colleagues of that value, which points to a real gap in communication or understanding still worth addressing.

Communication and Active Listening in Data Processing

Effective communication matters a great deal for understanding complex concepts, particularly in training and data discussions, and needing extensive detail to explain something often signals a lack of real understanding of the subject. Howard’s own experiences have borne this out, and he shared another example: the challenge of clearly articulating terms like “data provenance” versus “data lineage.”

Refining communication skills to convey ideas clearly and succinctly matters a great deal here, not unlike delivering an elevator pitch in 2-3 minutes. That includes writing data in an understandable way, reading comprehensively, and practising active listening to support informed decision-making. Fostering clarity and taking feedback on board, in the end, genuinely improves both personal and group learning outcomes.

Data Management and Data Analysis

Howard moved on to the key principles of Data Management that support effective data analysis, stressing how important clear communication is for conveying concepts like structured versus unstructured data and what they mean for data engineering. He also stressed how important it is for consultants and transformation agents to be able to explain complex technologies, large language models and generative AI included, to business stakeholders. Attendees were encouraged to assess their own proficiency in these areas and give feedback through a structured assessment available via QR code.

Figure 2 Where to find the Data and AI Proficiency Assessment

Figure 3 ‘Proficiency Batch’ Questions

Figure 4 ‘Data Management’ Questions

Figure 5 ‘Machine Learning’ Questions

Data Management and Communication Skills

The Proficiency Assessment makes the case for assessing technical capabilities in delivering the Data Strategy effectively, highlighting the need for Data Literacy across every level of personnel, not just executives and functional managers, while still recognising the role experienced data professionals play.

Attendees then discussed their own communication skills around Data Management topics and how comfortable they feel with emerging technologies like machine learning and generative AI, agreeing that continuous learning matters a great deal, especially navigating complex concepts like data value realisation and data ROI. Developing the ability to interpret and communicate data insights effectively, they agreed, is what improves overall proficiency in the field.

Figure 6 Data & AI Competency Framework

Figure 7 Key Personas

The Art and Challenges of Data Storytelling

Howard asked one of the attendees to speak to the challenges of making data visualisations understandable and impactful for a diverse audience. Dr. Daan Steenkamp stressed how important industry knowledge is for interpreting data, and how much storytelling expertise matters for communicating its significance effectively, noting that dashboarding tools keep improving but still don’t eliminate the need for thoughtful storytelling. Howard drew a parallel between effective data storytelling and movie scriptwriting, pointing to elements like general relevance, engaging openings, clear structure, relatable characters, conflict resolution, and emotional connection. Both agreed that interactivity in visualisations matters for audience engagement, and that compelling narratives matter especially when proposing changes like new Data Management initiatives.

Figure 8 Communicating Effectively

Effective Communication of Data Management and Ethical Considerations

Effective communication in analytics depends on understanding key concepts like Data Management, data engineering, business intelligence, data science, machine learning, and decision science, and it’s worth clarifying the distinctions among these areas and their relevance to the business, to keep every stakeholder aligned. Responsible analytics also stresses ethics in data handling and the application of responsible AI.

Organisations need protocols in place to validate data usage and mitigate potential harm before implementation, rather than dealing with issues after the fact. That proactive approach is what builds trust and keeps data use ethical throughout the analytical process.

Figure 9 Communicating Effectively: Data Story Telling: General Relevance

Figure 10 Communicating Effectively

Future of Business Applications and Digital Decision-Making

One attendee pointed out that analytical business applications are evolving to speed up and streamline decision-making. Digital decisioning is a key development here, using machine learning models or robotic process automation (RPA) to make decisions autonomously and reduce the need for human intervention, which raises real questions about which decisions can actually be automated and what quality standards these models need to meet, particularly around false positive and false negative rates.

The Challenges of Applying AI in Business Applications

Howard pointed to the real challenges around using AI in business applications, particularly cases where chatbots have been misused, spreading inappropriate content or giving non-compliant advice, leading to legal consequences. He also stressed the potential in analytical approaches to streamline processes like loan applications, using available data to cut processing time from weeks down to minutes. Two key types of analytics came up: confirmatory analytics, which verifies expected outcomes, and exploratory analytics, which investigates customer needs for new product development.

Figure 11 Communicating Data Management

Impact of Data Governance and Data Quality on Analytical Outcomes

Data Governance and Data Quality both play a real role in improving analytical outcomes, yet plenty of people struggle to articulate why, particularly with Data Governance. Data Quality tends to be easier to explain, but communicating the benefits of Data Governance to business stakeholders is genuinely harder. It matters to explain clearly how Data Governance improves decision-making and adds value to the business, answering the inevitable “what’s in it for me?” question.

Howard noted that data scientists and engineers sometimes see Data Governance as a hindrance, given the perceived delays it adds to project timelines when implementing Data Quality rules. Without those robust governance practices, though, the risk of working with unreliable data goes up, which ultimately compromises analytical results. Building a stronger Data Governance culture means embedding these principles throughout the organisation and being clear about the benefits for every stakeholder.

Data Governance and Data Engineering

Howard pointed to how important reliable Data Governance is for strengthening the analytical models data scientists build, stressing the need for clear communication between Data Governance professionals and data scientists on model reliability, grounded in best practices from authoritative frameworks like ISO.

Building an effective analytical team takes the right skills, particularly in Data Management and engineering. Data normalisation came up as a genuinely important concept here, linking Data Modelling to data cleaning and preparation, and still relevant despite a perception in some circles that it’s fallen out of favour. The conversation, overall, stressed continually improving practices through monitoring and adaptation as data analytics keeps evolving.

Figure 12 Remaining Questions of Communicating Data Management

Data Management and Preparation Strategies

Data preparation covers preparing and cleaning data with its purpose in mind, feature analysis or something else, and comes down to selecting the right data design method, dimensional modelling, anchor modelling, or a wide dataset in zero normal form, to support effective data engineering.

Howard then touched on the challenge of communicating why Data Management functions actually matter within organisations, especially since many lack a full understanding of their own data landscape. His approach is a balanced one, bringing in essential design principles without overwhelming stakeholders, and keeping governance aligned with agile practices.

Data Engineering and Decision Science

Effective decision-making has to be grounded in data, both in data engineering and decision science. Howard covered topics like normalisation, data warehousing, business intelligence, and data science, again stressing the need for clear communication between data scientists and decision-makers on model performance, overfitting among the issues worth flagging.

Cleaning, loading, and programming data matter enormously in data preparation, and so do the right database management systems and data storage solutions. Howard also addressed the role business analysts play in decision modelling and governing decision-making processes, closing on the “Human in the Loop” concept, examining when human intervention is actually necessary in automated decisions and how to balance efficiency against oversight in data-driven decision-making.

Figure 13 Communicating Decision Science

Figure 14 Communicating Data Science

Figure 15 Writing Data & AI Clearly

Figure 16 Data Loading

Figure 17 Decision Modelling Notation Business Analysis

Figure 18 Human In The Loop (HITL)

Implementing Data-Driven Decision Making in Business

Howard outlined the key questions data scientists and business intelligence professionals should be asking around KPIs to drive data-driven decision-making: analysing past performance across various periods, understanding the reasons behind current results, forecasting future outcomes against budget and resources, and identifying actionable strategies for improvement. He closed by stressing how important it is to recognise blind spots in data analysis.

Figure 19 DDDM

Figure 20 Key Personas

Competencies Frameworks in Data Management

The competency framework works well for assessing the Data Management office’s capabilities to support organisational decision-making and efficiency. One attendee raised the challenge of communicating the framework’s importance to non-data professionals, stressing the need for continuous learning and improvement across the organisation, but noted the framework is genuinely practical and can serve as a roadmap for growth. Another attendee acknowledged the value of self-awareness within the framework and expressed his own intent to use it to push himself further.

Business Benefit and Return on Investment in AI

Howard closed the webinar by stressing how important it is to clearly communicate the business benefits and ROI tied to AI initiatives. Despite ongoing effort in Data Management and AI, articulating the value these technologies bring to the business, particularly in tangible outcomes, remains a genuine struggle. Stakeholders often see data and AI teams as simply experimenting with open-source models, without grasping the real impact on business applications and insight.