The month of May was punctuated by events and books that will have a profound effect on the world of Data, AI and Analytics. From the way decisions should (and should NOT) be made, to how leaders can shape a sustainable data driven culture, here is a quick recap of what you might have missed this past month.
Analytics is Key To Digital Transformation
If you’re involved in technology, Gartner summits are events you do not want to miss. They are often full of new research and surprises. This year’s Data & Analytics Summits will not disappoint the ambitious Chief Data Officer (CDO).
According to the research firm, CDOs with business-facing KPIs are:
- 1.7 times more likely to be effective at consistently producing clear business value
- 2.3 times more likely to be effective at reducing time to market
- 3.5 times more likely to be effective at data monetization
This is great news for those who believe that data should be at the center of their organizations’ transformation. Over the past year and a half, CDOs have had the opportunity to shape and drive their company’s agenda: 72% of them say that they are either leading or heavily involved in digital transformation initiatives.
Yet, organizations that know how to manage data as an asset and use it to drive innovation are the exception, not the rule.
According to a survey highlighted in the Harvard Business Review, close to 7 out 10 organizations are still struggling:
Perhaps more alarming, “data-driven abilities” are reported to be on the decline: the same survey indicates a 13+ point drop, from 37.8% to just 24% of respondents saying that they thought their organization was data-driven.
So Now What?
Perhaps the strategies of the past have stopped working. Perhaps, rapid technology disruption has been challenging for many to incorporate. The database management serves as a great example of such an incredible shift. At this year’s event, Gartner indicated that:
If the way your company stores and uses data is drastically going to change over the next 24 months, what other abilities do you need to build to take advantage of this new development?
I’ve had the opportunity to work with and study the success of enterprise leaders and digital natives who have experienced breakthroughs through data. I don’t pretend to have a magic formula but I can share at least 3 traits I’ve observed “data leaders” demonstrate.
- Data as a competency
- Data as an ecosystem
- Data as a fabric that runs a company’s decision making system
Data as a Competency
You’ll find a lot of resources around “data culture” and the science of sound decision making. My most recent favorite books on this topic were in fact published this past month: Framers by Francis de Véricourt, Kenneth Cukier, and Viktor Mayer-Schönberger and Noise by Daniel Kahneman, Olivier Sibony, Cass R. Sunstein. Both are must-reads for the ambitious data executive.
Combining learnings from these books with best practices like analytics training and “data mentors” can help organizations build the foundation of their “data culture”. A good way to assess your chances of sustaining that culture though is to calculate the percentage of “data people” your organization employs.
Consider this: leaders often say that the source of their competitive advantage comes from their people, not just their products or services. If that’s true, then the best way to tell what your company’s core competency is to look at your employees and their functions.
Piyush Gupta, the CEO of DBS Bank, was just quoted this past week on this very topic and here is what he said: “we are now increasingly thinking of ourselves as a technology company offering financial services, rather than a traditional bank”. The fact that we have twice as many engineers than bankers is perhaps a testimony to the shift in the nature of the company that we are.”
With this in mind, I released a LinkedIN survey at the beginning of the month and asked “how many data employees should you employ as a percentage of your total employee base“. The survey gathered wide attention with close to 80,000 views and 625 votes.
One of the respondents, Kirk Borne, an esteemed Data Scientist and PhD from Caltech, provided useful guidance around the term “data employee”. He explained that:
- Approximately 100% of employees should be data literate (ie. “recognize, understand, and “talk” data”)
- A third should be “data fluent” (ie. “analyze, create arguments, and present results with data”) and,
- Less than 10% should be data professionals or “paid to create value from an organization’s data assets (data scientists, data analysts, BI specialists, data/AI/ML engineers, database/data warehouse engineers)”
What is your current percentage? Building a “data culture” requires the books, the training, the constant reinforcement and the posters on the wall. But it is sustained by your company’s org. chart. So, make sure that your numbers match the level of your ambition.
Data as an Ecosystem and the power of interconnected “data networks”
Another important consideration when building a “data culture” is the type of data that your employees get exposed to. For the longest time, leaders have complained about the fact that employees struggled to gain access to the data their company produced. That is, unfortunately, only a small portion of the problem.
As Nick Blewden, the Head of Data Products at Lloyds of London shared online this past week, the problem with “data myopia” might be more acute than you think. The picture he posted below tells the full story: the red dot is your company’s data (which, again your team might see very little of today) and the grey area is the data you could see.
All this points to a trend worth watching: the rise of data exchanges. I had the opportunity to share my thoughts on this topic this past week on the DM Radio podcast, hosted by Eric Kavanagh, CEO at the Bloor Group, which you can listen to here. But, at a macro level, you should know that the most progressive companies look at data as an ‘ecosystem’ opportunity, where insights arise from the combination of data emerging from interconnected “data networks”. Consider this:
Yet….only 5% of data-sharing programs correctly identify trusted data and locate trusted data sources today. Which brings me to my third topic: Data Fabric.
Data as a Fabric
The term “Data Fabric” is not new. According to Wikipedia, it’s been in existence for at least 2 decades and its meaning has been transformed multiple times since then. The term garnered tremendous attention this year when Gartner published their Top 10 Data and Analytics Trends for 2021. The research firm predicted that data fabrics could:
- Reduce time for integration design by 30%
- Deployment by 30% and,
- Maintenance by 70%
It is also worth noting that the term “Data Fabric” currently sits at the tippy top of the research firm’s Hype Cycle for Data Management, the phase that indicates a maximum gap between customers expectations and solutions reality (for a quick explainer on “how to decrypt hype cycles”, watch this). A lot goes into these predictions and you should refer to Gartner’s research to understand what’s a Data Fabric and what isn’t.
The fundamental reason for why companies look to deploy a data fabric is well understood: as data is growing increasingly distributed (across data lakes, data marts and data warehouses), highly diverse in type (structured, unstructured) and location (on-premises and across clouds), data and analytics leaders are challenged to manage data access and governance without major trade-offs.
When IT can’t govern data effectively and users can’t access information they need in a timely manner, innovation suffers. According to a recent Dimensional Data research survey, 68% of data analysts have ideas to drive profits, but don’t have the time to implement.
The increased amount of analyst guidance in the space should help you form your own opinion: a great resource to use is IDC’s paper on what the firm calls the “Data Control Plane” concept.
Chandana Gopal, Stewart Bond, and Dan Vesset do a great job explaining this new architecture across integration, access, governance, and protection as well as explain the role of its 3 layers, namely:
- Intelligence layer — for transparency of data profiles, classification, quality, location, lineage, and context
- Governance layer — a policy engine to control data accessibility, movement, usability, and protection
- Data engineering layer — for data integration, ingestion and transformation, in-memory data virtualization, federation, replication and streaming
As usual, I hope you find these best practices helpful. For suggestions and comments, please don’t hesitate to contact me directly here.