So You Think You're a Data Driven Organization?

Raise your hand if you feel you are part of a data-driven organization or company. Now keep your hand up if you can tell me what exactly that means. Is your hand still up? Are you sure? Can you show me evidence that data is at the center of the business decisions and strategic planning on a daily, monthly, quarterly, and annual basis? If the first thing you point to is a plethora of reports and dashboards, then put your hand down.

Let me pause and provide a necessary caveat - Opinions expressed are solely my own and do not express the views or opinions of my employer.

Investing in Big Data and AI does not make a data-driven organization. Switching on 1000 IoT devices to dump millions of transactions into your Data Lake does not make a data-driven organization. As with most things in business, the path to becoming data-driven is based on people and culture. In a 2019 Harvard Business Review article, Randy Bean and Thomas Davenport emphasize "the challenges to successful business adoption [of Big Data/AI initiatives] do not appear to stem from technology obstacles; only 7.5% of [executives] cite technology as the challenge. Rather, 93% of respondents identify people and process issues as the obstacle." (Bean & Davenport, 2019).

So, what exactly is a "data-driven" organization? The Data Warehousing Institute provides us with a list of five characteristics (Peregud, 2018):

1. Creative executives who run their businesses with passion and curiosity
2. Data democratization
3. Data literacy
4. Automation of data management workloads
5. A companywide, data-driven culture

Since the Data Warehousing Institute (www.tdwi.org) has been helping us identify and implement Business Intelligence standards for 25 years, this list is a great place to start. Item #1, creative executives, is somewhat subjective, and difficult to validate. I think if we take an honest look at the other four topics on this list, though, we will find that most "data-driven" organizations are, in fact, not data-driven at all.

Let's start with data democratization. To quote the TDWI article, this simply means "broad data access for all employees." Does your organization ensure all employees have access to all data? Not reports, mind you, I am talking about the underlying data. Can they run queries, perform analyses, and present the results? I'm not asking if they have the skill set, that's our next topic. I'm asking if they have access to the data?

Many organizations restrict users and allow access only to standard reports and meta layers, highly prescriptive layers that restrict access and ability to work with the data. Often the phrase "One version of the truth!" is used to justify such data vaults. (Perhaps I'll tackle the oft misused "one version of the truth" in another article.) This is, in fact, exactly the wrong approach to enabling a data-driven organization. How can the organization be data-driven if it has no access to the data? Martin De Saulles, in a 2019 online article for CIO magazine, emphasizes that "At the heart of [data analytics] needs to be the objective of making better decisions. This requires data to be made available to everyone in the organisation who needs it as well as the tools to analyse it." (De Saulles, 2019) (emphasis my own)

Assuming we've made the data available to everyone, how can we ensure they know how to use it? In the 1990s, organizations began expecting all employees to know how to use a computer. I'd argue that in the 2020s we need to expect all employees to know how to use data. Data literacy is defined in the TDWI article as "the ability to read, work, analyze, and argue with data." (Peregud, 2018)

Understanding data can be intimidating for those unfamiliar with the structure and concepts. Remember, we are not talking about how to read a PowerBI report. Data literacy requires more direct engagement with the data. To enable data-driven organizations, executives need to make an investment to ensure employees have that skill. Sarah Hippold, in a 2018 article for Gartner agrees: "Data and analytics leaders must engage and train their entire organization to become data literate." (Hippold, 2018)

The automation of data management workloads, on its surface, sounds like something most organizations have already accomplished. And, in fact, many IT shops have perfected the processes of automating extract, transform, and load (ETL). Irina Peregud, in her TDWI article, quotes James Kobielus, a Research Director at TDWI, and a Lead Analyst at SiliconANGLE at the time.?"This is not a matter of getting everybody using manual approaches such as traditional business intelligence reports, dashboards, query accelerators, etc." According to Kobielus, data-driven organizations have "taken steps to automate the distillation of data-driven insights and the incorporation of those insights into business processes." (Peregud, 2018) I may be overinterpreting Kobielus' statements, but what he is describing is a Decision Support Systems (DSS), and I've found few organizations have successfully implemented DSS solutions.

Finally let's tackle the question of a company-wide, data-driven culture. This is, perhaps, the most difficult aspect of data-driven organizations to measure and often falls into the bucket of "I know it when I see it." Sarah Hippold, in her Gartner article, identifies some key characteristics for us. "A modern data and analytics organization enables both centralized and decentralized work while creating a center of gravity for critical competencies that ensures consistency of practices and collaborative insight creation." She continues with "Being data-driven is a team sport and requires collaboration, as well as effective and ongoing training programs." (Hippold, 2018)

I would also argue that part of culture change requires acceptance that some analytical investigations, in fact most, will yield little or no insight. That said, "Analytical decisions and actions continue to be generally superior to those based on intuition and experience." (Bean & Davenport, 2019) Sometimes knowing that something did not happen, or that the data does not support a hypothesis, is just as important or more so than the discovery of something new.

Considering the extensive requirements above, would you still classify your organization as data-driven? The Harvard Business Review tells us "Leading corporations seem to be failing in their efforts to become data-driven." (Bean & Davenport, 2019) In considering the many different teams I have been part of, I offer my own list of required characteristics for data-driven organizations, obviously overlapping with the TDWI list.

1. Data Literacy - Everyone - I mean EVERYONE - should have basic knowledge of how to access and work with data in Excel, Tableau, or a similar tool. Feeling comfortable with a VLOOKUP? Great, but if you can't use a pivot table, you need to learn!
2. Data Primacy - The "Chief Analyst" should be a peer to the Chief of Staff in an organization. A data-driven executive needs a data expert that reports directly to them with the resources and tools to direct ongoing analysis of organizational performance and opportunities. We don't need data scientists in this role, though a data scientist may be a valuable team member, and a Chief Analyst should be familiar with the tools and concepts behind data science.
3. Data Exploration - This is the freedom to explore information, knowing that 90%+ of the time it will yield nothing. In other words, a significant amount of time should be spent simply investigating hypotheses without necessarily expecting immediate results. In the short term, the ROI here may seem incredibly low, but when such exploration yields new information, the ROI is incredibly high.
4. Data Democracy - Unless there is some legal or privacy constraint, no data should be "locked up." When it comes to Personally Identifiable Information (PII), I am a strong proponent of GDPR and similar privacy regulations. Further, an organization’s financial and legal data should have strong security applied, ensuring limits on exposure. Most data used within an organization, though, does not require such restrictions and should be shared throughout the organization. To encourage analytics that benefit the organization and your customers, data needs to be designed to enable access. If you control a data set that is important to understanding your organization, have you ensured that others within your organization can access and leverage that information? This also needs to be supported, of course, by the last item on my list.
5. Data Documentation - None of the above can happen without clear and complete documentation of an organization's data assets. Much of the confusion around seemingly conflicting results I see stems from a lack of clear documentation of what is being presented. Documentation needs to include:

• Standard organizational KPIs and acceptable thresholds.
• Explicit differences between standard KPIs and similar organizational metrics that are defined differently.
• Common scenarios for leveraging a particular metric.
• Attributes that are most relevant to a particular metric for filtering and comparing results.
• Which metrics are used in a specific report or analysis with links to complete definitions and use case scenarios?

The above list is likely not complete, and most importantly, part of data literacy (#1 above) requires the ability to leverage this documentation appropriately.

None of these steps happen easily within an existing organization, and they all require significant investment. Perhaps an organization's status as data-driven is a spectrum, and executives should strive to move teams further along that spectrum, investing in specific initiatives. But if you woke up one day and decided to call your organization "data-driven" because you like looking at reports on a monthly, weekly, or even daily basis, then you have drastically misunderstood the value of your data. I'd ask you to reconsider the first characteristic in the TDWI list: Creative executives who run their businesses with passion and curiosity.

Opinions expressed are solely my own and do not express the views or opinions of my employer.

References

Bean, R., & Davenport, T. H. (2019, February 5). Companies Are Failing in Thier Efforts to Become Data-Driven. Retrieved from Harvard Business Review: https://hbr.org/2019/02/companies-are-failing-in-their-efforts-to-become-data-driven

De Saulles, M. (2019, October 28). What exactly is a data-driven organization? Retrieved from CIO: https://www.cio.com/article/217767/what-exactly-is-a-data-driven-organization.html

Hippold, S. (2018, September 20). Build a Data-Driven Organization. Retrieved from Gartner: https://www.gartner.com/smarterwithgartner/build-a-data-driven-organization#:~:text=1%20Spark%20ambition.%20Transforming%20into%20a%20digital%20business,ambitions.%20...%203%20Close%20the%20competency%20gaps.%20

Peregud, I. (2018, September 26). Five Characteristics of a Data-Driven Company. Retrieved from The Data Warehousing Institute: https://tdwi.org/Articles/2018/09/26/PPM-ALL-Five-Characteristics-Data-Driven-Company.aspx