What are the most common data literacy skills gaps in your team or industry?

Lack of data governance can result in real financial costs, including data storage and processing expenses. Duplicate data increases storage demands, and even with affordable storage options, costs accumulate in big data and cloud storage contexts. Inconsistent data types, caused by the absence or disregard of a business's data dictionary guidelines, also raise processing costs and time. Although converting a unique ID column's data type (e.g., from a Customer Table to a Sales Table) may seem trivial for small datasets, it becomes a significant challenge for larger datasets.

I would say that data governance falls under the “understand” category of data literacy for this article. Strong data quality and governance creates the foundation for common understanding of information within your organization. As data proliferates, data governance is a way to explain what decisions are being made with the data and why. It will also help ensure you are meeting requirements for data handling in regulated environments. In terms of data quality, especially for technical users, I’ve seen transparent data transformations increase user understanding. Imagine searchable or visualized data pipelines. This ensures errors are minimized, increases transparency and trust, and creates a mechanism for feedback from users.

1. Lack of Data Quality and Governance: This is the most common skill gap in our team and industry. Ensuring data accuracy, completeness, consistency, and timeliness is crucial for reliable analysis. 2. Limited Understanding of Data Analysis Techniques: Many team members struggle with applying appropriate statistical and analytical methods to extract meaningful insights from data. 3. Difficulty in Interpreting and Communicating Findings: Effectively communicating data-driven insights to stakeholders in a clear and understandable manner is a challenge. 4. Insufficient Knowledge of Data Visualization Tools: Proficiency in using data visualization tools to create compelling and informative visuals is often lacking.

Depending on the data, it is critical to understand the importance of Personally Identifiable Information (PII) is critical. Users must recognize and correctly handle PII, understanding its implications like customer trust for instance. Effective data governance must include PII identification, classification, and procedures to protect sensitive info. Teams should be trained about PII and how to encrypt this information.

I have seen the lack of data governance in multiple firms. Data governance framework need to be created by data and analytics SMEs and mandated by senior leadership. I also feel that analytics managers need to improve in communicating the criticality of data governance in dollar/business terms to top leadership. There are firms where there is more focus on immediate needs and less on future growth/roadmap. This results in making investments to improve data quality and building data governance frameworks and mandating it to the background.

One of the most common gaps in data analysis and visualization is dimensional modeling. Dimensional modeling is core to being able to properly visualize and present data across multiple Facts and Dimensions. A strong dimensional model serves as the foundation for data analysis and visualization. Often this is a gap in skillset.

I agree. I feel the culture of data analysis to find patterns and relationships in the data is yet to be established. Also patience is needed for doing data analysis which is also missing, professionals want to quickly move to building algorithms. However data analysis is a fundamental component of ML. Patterns and relationships in the data is what gets captured in ML algorithms. Before building these ML algorithms we need to see by ourselves through data analysis these patterns/relationships. Data visualization also includes making intuitive visualizations and telling a story. Technical professionals know to create charts/plots, process & extract data. However, ability to create intuitive visualizations which can tell a story is missing.

Improving data analysis and visualization is key in addressing the data literacy skills gap. I use Tableau to simplify intricate data, offering easy comprehension and effective communication. It aids in error detection, storytelling, and decision-making. To bridge the data literacy skills gap, data users must acquire skills in data analysis techniques (descriptive, inferential, and predictive statistics) and master data visualization principles (selecting the right chart types, colors, layouts, and annotations).

Data analysis and visualization skill gaps often include a lack of proficiency in advanced tools and an understanding of storytelling with data. These gaps can lead to poor decision-making and missed opportunities. When hiring for these roles, look for: Proficiency in Tools: Experience with tools like Tableau, Power BI, and Python. Data Interpretation: Ability to derive actionable insights from complex data. Storytelling Skills: Communicating data insights clearly and compellingly. Statistical Knowledge: Understanding statistical methods and their applications. Investing in these skills ensures your team can effectively analyze and visualize data, driving better business decisions.

In my industry, key data literacy gaps include limited knowledge of advanced analysis techniques, insufficient focus on data cleaning and transformation, and challenges in selecting appropriate visualizations. Additionally, there is a lack of proficiency in storytelling with data and advanced features of tools like Power BI, impacting decision-making and data communication.

Its important to know your audience. Specifically if you're seeking to convince them of a point by leveraging data, or you're seeking to simply explain that same point and let them come to a decision. This will differentiate in how you present the same set of data to your audience.

Storytelling, articulations, ability to land messages, identify top messages etc. are extremely critical for a successful data scientist or analytics professional. Its a skill which is often found missing. The data scientists are more skilled towards hard technology skills. Often data scientists also don't enjoy telling the story and communicating the insights. We need to bring in the culture were story telling and communication are equally valued. It will be the difference between a data scientist who becomes a successful leader and other data scientists.

To create a compelling data story, I keep the following principles in mind: 1. Connect to the organization's goals and objectives. 2. Tailor your data story to your audience's level of familiarity with the subject matter. 3. Visualize data effectively. Use interactivity to provide instant context and meaning. 4. Your narrative should inspire action or decision-making. Offer a clear action plan based on the insights you've gained. 5. Revisit and revise.

Other experts discuss high-level problems, but I would say that many people just have gaps in their school math knowledge. It's quite often when a person doesn't understand the difference between median and average and what this difference could tell you. Also, quite a lot of people confuse the terms percentage and percentage points. Therefore, if you see that your interlocutor is perplexed by some part of your speech, I urge you to explain the difference between these concepts unobtrusively. This will help both you and the person in the long run.

Most people don't know how to speak about data. They don't fully understand the impact of limitations and assumptions. And they don't know how describe the technical aspects of technical analyses. When preparing presentations, work with your team to ensure they are equipped to answer the hard questions if asked.

For an analytics team collecting or working with customer data, it is essential to implement Personal Identifiable Information (PII) protection practices as this is sensitive information that can be traced back to the customer. Data governance and ethics is critical in setting up a foundation of a successful analytics team.

When data falls into the wrong hands or is used to further individual interests, rather than those of the organization (this is especially true of HR data like compensation data), it can lead to deleterious consequences. Further, when data is too freely accessible, it can also be used in ways that are not intentionally malevolent, but when consumed by untrained audiences, it can be wrongly generalized (i.e. making generalizations on small populations of data) and business action can be taken on invalid or misinterpreted data. It is the responsibility of data owners and stewards to protect key data categories and ensure that adequate training on ethical use of data as well as ensuring technical competency to understand it is firmly in place.

Often there are disparities and shortcomings in how organizations and individuals handle and manage data ethically and responsibly. One of the primary areas of concern is privacy, where organizations collect large amounts of personal data from individuals, sometimes without their explicit consent or awareness. This data may be used for purposes beyond what was initially disclosed, leading to privacy violations and breaches of trust. Another concern is of data bias and fairness, where a department or business function (mis)uses data to show a biased view that may be half truth or even misleading. Other issues arise with lack of data transparency and accountability, data security risks from improper data governance, and unethical use of AI.

One thing I have found helpful is to have Data Ethics and responsibility standards defined and documented on Day 1 for the team to be used as a gold standard for any model built within the team. If every model has set of KPIs, folder structures, accesses, allowed data types which are already defined and aligned with stakeholders, this saves significant time in cutting back and forth of multiple use cases. Ofcourse, you need to keep reviewing the standards periodically to keep it updated.

In the field of AI and large language models, a significant data literacy skills gap lies in understanding and applying data ethics and responsibility. While many possess technical expertise, there's often a lack of awareness regarding the potential societal impact of AI systems. This includes issues like bias in training data, privacy concerns, and the potential for misuse. Many lack the critical thinking skills to evaluate the ethical implications of their work and ensure fairness, accountability, and transparency. Bridging this gap is crucial to develop and deploy AI responsibly, mitigating potential harm and fostering trust in these technologies.

In order to foster a culture of data curiosity, you need to take into account whats relevant and impactful for your end users. Data solutions built towards your own interests, as opposed to the insights and data points your users are seeking, inhibit data curiosity. Be sure to account for the high impact items users are seeking when you're building data analytics solutions.

A large part of managing an analytics organization is seeking to build in time for innovation. Analysts should be encouraged to engage with business/customer stakeholders both qualitatively (via consultations) and quantitatively (via trend analysis of current and retroactive data requests) to begin understanding the demand temporally and categorically for different types of data and building solutions that cater to the desired business outcome. Moving toward a model whereby robust self-solve tools (dashboards, etc.) can empower users to answer 80% of the questions they want answered and leave 20% to deep-dive analysis and follow-up can help to reduce the ad hoc workload and free up space for innovation, research, and product development.

Lack of curiosity and creativity often leads to loss of data sanity and wrong decisions. Any change is data trends, or outliers need to be questioned. Also, any change in product or service needs to be measured and the data generated should be combined with other factors to produce new use cases and experiment. This is possible when everyone involved is familiar with data processes, definitions and usage. Once the culture of data curiosity and creativity is inculcated, it leads to many new initiatives that otherwise would not have been possible.

One very important factor is the lack of application to understand how the measurement framework works. At times, analysts overlook this factor. For instance, Google Analytics or any other web analytics tool, many times, would give you numbers against any dimension/metric combination even if they aren't related ! Unless the analysts understands how the measurement framework works, they would end up reporting on numbers which were never meant to be there ! The thinking that measurement framework is important for the implementation experts to understand and not for a data analyst to have grasp of, is the reason business stakeholders lose confidence on reported data. Because it just doesn't add up...

Identifying and critically assessing skill gaps in data literacy and interpretation should be the foundation for any effort to help build in data L&D. By identifying these gaps, a cross-functional work group should be able to put together a compelling business case that can garner executive support and move forward with a development plan. Without executive support and a desire and communicated incentive (i.e. career mobility) to move toward a data-driven culture, these efforts often fail or fall short. Stating clear, measurable targets for learning pathways or single initiatives and developing sound materials that can be consumed in multiple channels (video, written, in-person, on-demand, etc.) can help reach different styles of learners.

Another data literacy gap exists in identifying project prioritization, and this gap persists across industries. As an analytics practitioner for two decades, I've consistently observed that the most critical factor is aligning analytics resources with the organization's overall strategy. Unfortunately, some of the most skilled analytics talent spends their time doing work that is necessary but not necessarily the work that will drive competitive differentiation and breakthrough insights. A Harvard Business Review study reveals that 80% of analysts' time is spent simply discovering and preparing data. Ideally, a greater portion of their time, closer to or exceeding 50%, should be dedicated to analysis itself. Thoughts?

Users must also understand how data is collected, and what kind of limitations this causes. For example, it still comes as a surprise to many web analytics users that the data lacks information about visitors who did not give their consent to being tracked.

Learning the data language of a business is vital. This means having a deep understanding of the business function that you are addressing with analytics/AI. This is not a generalizable skill, but a specific one that requires you to visualize the business operations through data. This could help you unlock additional value from data and find new insights.

I have been assisting with Data Literacy training our data team has been conducting for the product and business teams. Many teams have no idea what is possible with their data. They don't realize how many options they have. The teams also need tools that are accessible to their level. Friendly, self-service options for accessing data that get them to dip their toes into the data lake and see what is there.