I had an opportunity to participate in the 2023 Gartner Data & Analytics Summit in Orlando. It was a humbling experience networking with & learning from various analysts, speakers, exhibitors, and industry peers. Here are 10 learnings that stuck with me from the conference mixed with my own reflections:

1. The struggle to demonstrate tangible ROI from D&A initiatives continues

   Gartner’s 2023 CDO survey shows 69% of D&A leaders are struggling to deliver measurable ROI. To change this, leaders must master the language of business, stop chasing the “shiny new objects”, and use both financial & financial influencing factors to measure & articulate value.

2. The divide between business & IT organizations in large enterprises is real & growing

   Far too many examples or conversations where Business & IT don’t seem to be operating based on a shared understanding of what success looks like. Introducing new technologies or operating models won’t necessarily overcome this divide. What is needed is empathy, willingness at the top, and then specific interventions in your context. This topic deserves more attention in the industry.

3. “Enterprise decision networks” is an emerging capability to watch out for

   “The success of an organization is nothing but a sum of all of its decisions.” Gartner predicts that 75% of Global 500 companies will use decision networks as a competitive differentiator. Not many products exist in the market that systematically close the loop between analytics & decision making. We train & optimize AI models based on data but rely on training & then just experience for humans. Expect decision intelligence platforms to emerge that help you model, track & improve decisions.

4. LOCO/NOCO are driving the next wave of innovation in D&A

   Companies like Nexla, Aible, Cinchy, and Kyligence are reducing the effort needed to organize, explore, collaborate, and trust data & insights. Nexla is automating data engineering; Aible is automating exploratory insights; Cinchy is eliminating the need for data integration; and Kyligence is automating the metrics layer. They are reducing the need for developers in the data-to-insights lifecycle. Expect shorter time to value and higher software & niche labor costs – all while hopefully lowering the TCO!

5. Are we overestimating the short-term impact & underestimating the long-term impact of Generative AI?

   ChatGPT is likely going to be the “word of the year.” Nearly every vendor had something to say about Generative AI. The possibilities in content creation, workforce productivity, software engineering and knowledge management are eye opening. To take advantage companies should develop a position paper outlining the opportunities & risks and establish an incubation team to explore use cases. To prioritize the right ones, apply the “willingness to pay” test to determine use-case worth.

6. A consolidation wave is coming in the D&A technology landscape potentially adding to your modernization debt

   As I observed the bake offs for data management, analytics & BI, and DS/ML, it’s clear that many vendors have overlapping foundational capabilities & then a few differentiating ones. Most try to differentiate based on knowledge of their own offerings. Customers need to be very systematic in evaluating these tools to identify what will be relevant in their data ecosystems or be ready to see their modernization debts rise. While the total addressable market is very large in D&A, can the industry afford so many overlapping platforms?

7. Executives need to become AI fluent to seize opportunities for & mitigate new risks to their businesses

   AI is going to disrupt almost all businesses before the end of this decade. But are executives in different organizations comfortable with AI? Given the potential for business benefits & harm, executives must invest in AI fluency. They need to be able to ask the right questions to guide their teams towards relevant opportunities in customer experience, operational excellence, or new business. They should also take questions around harm – both malicious & commercial use – very seriously & establish governance frameworks to exit early when risks are high.

8. FinOps practices are essential to avoid surprises in cloud bills

   Gartner shares that a primary complaint from cloud buyers is the lack of predictable costs & transparency in their purchased cloud solutions. Factors such as rapid provisioning, pay for what you use, & elasticity that made cloud attractive are also leading to overspend. One way to bring equity in cloud cost monitoring is to use price performance metric (total cost/total work capacity). Organizations must apply FinOps to cloud budgeting, governance, & optimization practices. FinOps requires technical & cultural shifts to ensure CFO, CIO/CTO, and Business Leaders can connect cost to value.

9. If your data literacy programs aren’t yielding results, shift to persona-based or employee-led models

   Data fluency is an organizational capability that enables every employee to move away from “just reading the numbers” to “accessing data, understanding its context, interpreting insights, and using them” as fluently as one can speak in their language. A powerful success story from New York City Health & Hospitals confirms that to get effective results organizations should consider persona-based & adult learning models in their data literacy programs.

10. Organizational change leadership is critical to close the gap between insights & actions

    And lastly, but perhaps most importantly, to get any meaningful data-driven results in your organization, you need to be able to effect change in yourself, your teams, & your organization. I enjoyed Shankar Vedantam’s talk on how the “hidden brain” shapes our capacity for change. Barriers to motivation include organizational inertia, the IKEA effect, and the sunk costs fallacy. Techniques to persuade include adopting swift irreversible change actions, using positive reinforcement, and doing hard things first.

This article was first published on Forbes.com, read the article here.

In the digital economy, data is our most valuable resource. “Data is the new oil” is now a popular saying, and it’s an apt metaphor. Companies that learn to extract and use data efficiently have the potential for huge rewards, driving profits, innovation and customer satisfaction.

Just like with oil, quality matters in data. Quality isn’t free; it requires time, money and attention, but it’s worth the investment.

What Is Good Data? 

Good data quality means the data is fit for use based on the business context and requirements. The business rules governing quality include both quantitative (e.g., “Is the date legitimate?”) and qualitative rules (e.g., “Is the date captured in the American notation?”). In addition, expectations of users regarding availability, reliability, usability, relevance and clarity may lead to perceptions of quality issues. Data quality initiatives need to address various aspects to ensure that data is trustworthy.

Think about global data related to Covid-19 vaccinations. Reliable data must include a patient’s date of birth, the date of each dose, type of vaccine, number of required doses and location of each dose. Quality issues become complicated when you consider that some users received shots from different vaccine manufacturers, or data may have been captured in different states or countries. Poor data quality can prevent various stakeholders — including public health experts, vaccine advocates, and the general public — from making informed decisions. If people perceive vaccine data to be unreliable, they may become more hesitant to get vaccinated and ultimately damage public health outcomes.

The Cost Of Bad Data 

In 2016, an IBM* study estimated that bad data costs the U.S. economy $3.1 trillion a year, and in 2020, a Gartner** survey found that organizations calculated that the average cost of poor data quality was $12.8 million a year.

In my experience leading a global data and analytics (D&A) solutions provider, I’ve seen that while everyone cares about data quality in theory, when it comes to actually making funding decisions, many customers want to cut corners.

But this is where the rubber meets the road. If you don’t finance data quality initiatives, then you won’t get the result you want. Poor quality data can lead to flawed decision making, top- and bottom-line consequences and decreased employee and customer satisfaction. Incomplete data may result in ineffective marketing campaigns, and a data breach can cause reputational damage or leave you vulnerable to litigation under laws like GDPR or CCPA.

Six Common Challenges In Data Quality Improvements

Improving data quality in your company will have significant long-term benefits, but you must be proactive. Here are six of the most common challenges to be aware of when improving data quality:

1. Lack of awareness: Because data is an intangible asset, it’s often hard to assess quality and address problems. Your stakeholders may not fully appreciate the state of data quality in your systems until a major issue impacts your business.

2. Difficulty justifying investments: To get buy-in on improving data quality, you need to be able to make a solid business case for it, showing how poor-quality data has had negative consequences in the past. But frontline staff may not be willing to document quality issues to build a future business case for something like automation, preferring instead to resolve issues manually.

3. Confusing shared responsibility with IT responsibility: Enterprise data is often used across multiple business units, moving through line-of-business systems into reporting and analysis systems. Quality ownership is delegated to IT as data flows through various pipelines, yet IT is not fully responsible for the source systems. Data quality demands shared responsibility.

4. Resistance to change: Data quality programs are heavily focused on continuous improvement, which calls for users in your organization to adopt new behaviors and perform additional checks and balances with discipline. If employees are unwilling to adapt, you will run into obstacles.

5. Fear of transparency: Data quality assessments and real-time quality dashboards may make some leaders uncomfortable. Looking into past data decisions and results may cause some in your organization to feel embarrassed or concerned, creating yet another roadblock.

6. Lack of sponsorship: Data quality initiatives often compete with new technology and product management investments. It can be tempting to throw money at a shiny new object, spending $X on a cloud computing platform instead of the same amount for a data governance consultant’s annual fee. Data quality is less glamorous and often loses out to modernization initiatives.

Five Ways To Improve Data Quality 

Once you’ve identified the challenges you’re facing, here are five actions you can take to address them:

1. Sponsor beyond budgets: To achieve successful data quality initiatives, you must be willing to invest more than dollars. Articulate the importance of data quality for the organization, inspire cross-functional collaboration, prioritize progress and hold frontline managers accountable for long-term success.

2. Invest in data quality and transparency tools: As the volume, variety and velocity of data increases, you need more specialized tools for quality management. Invest in software that automates quality management for profiling, catalogs, metadata and lineage.

3. Adopt DataOps automation practices: DataOps practices build on DevOps practices from the agile software engineering domain within data analytics. Integrate them to promote collaboration, communication and automation to bring business, applications, operations and data teams together to speed feedback cycles for quality issues and reduce cycle times for delivering D&A products.

4. Embrace a stewardship culture: To create a data-driven culture, you need to incorporate stewardship in your organizational norms and habits. This is a shared responsibility among various stakeholders and includes identifying and articulating quality issues and following governance practices to resolve issues.

5. Build a data management team: Data quality management requires dedicated skills, roles and positions depending on the scope, size and complexity of your organization’s data. Hire the people who can evaluate, design and execute the solutions you need, including Chief Data Officers, data analysts, business and technical stewards, tool experts and data management consultants.

Investing in data quality is a long but worthwhile journey. Evaluate your company’s needs and challenges, and invest in the practices, tools and leadership required to advance valuable data quality initiatives.

References:

• * Harvard Business Review – ‘Bad Data Costs the U.S. $3 Trillion Per Year‘ by Thomas C. Redman, 22 September 2016.
• ** Gartner Research – ‘Cost Optimization Is Crucial for Modern Data Management Programs’, by Ankush Jain,  Guido De Simoni,  Eric Thoo,  Adam Ronthal,  Melody Chien,  Donald Feinberg,  Ehtisham Zaidi,  Sally Parker,  Simon Walker,  Malcolm Hawker, 22 June 2020.