What Every Leader Should Be Thinking About Right Now in Order to Make the Critical Transition Into a Highly Competitive, Data-Driven World
According to the 2019 Big Data and AI Executives Survey, which polled C-suite executives from some of the most influential Fortune 1000 organizations, only 31% of leaders said that they believe their organization is “data-driven.”
This resource is intended to get to the crux of how to become a more data-driven team or organization; it’s a “quick-start guide” because it outlines the most important data-related activities every enterprise leader should be prioritizing right now in order gain better control of their data, make better business decisions, and ultimately, deliver tangible results.
Many tend to get caught up in the details of evaluating one solution or another, often at the expense of identifying the underlying problems correctly. Properly mapping data strategy to business outcomes makes finding the right solution not only easier, but often produces the desired result.
— Bryson Hill, Director of Client Strategy, RevUnit
Many teams and organizations make the mistake of overlooking process, especially so when dealing with data. There’s often a rush to quickly move past process (or skip it altogether) and jump straight to the specific tool, tech, or tactic that a particular team or individual believes will most quickly solve a need or problem. Often, whether knowingly or unknowingly, these types of decisions are hastily made in order to serve the interests of a particular team, business unit, or individual. A bias towards action is usually commendable, but doing so in this fashion (without first understanding the organization’s key priorities) typically leaves you feeling as though you’re chasing a moving target, as if you’re unable to “see the forest through the trees.”
Your 2020 plans may have been altered by the COVID-19 pandemic, including your organization’s top-level business objectives. Ensuring top-down alignment right now is likely more critical than at any point in recent memory. As a supplement to this guide, you’ll find more practical guidance to help you navigate through a difficult time in our Quick-Start Guide to Becoming a More Adaptable Organization
In the absence of a well-documented data strategy—one in which the data priorities of the team or business unit align closely with those of the organization—you can often make key data decisions out of feelings of self-preservation (“I need to do what I need to do in order to hit my goals”). In times of crisis, however, it’s possible that your own objectives—or those of your team or business unit—may be at odds with those of the organization. This innate conflict—especially at a time like the present—can then create manufactured inefficiencies, introduce unnecessary complexities, can increase costs, and can even expose organizations to greater risk (in the form of bloated toolsets and possibly unknown data security risks). So, your first step is to re-confirm your top-level, organizational objectives (as they may have shifted over the past few months). Doing so is your first, yet most critical step toward kick-starting your data strategy in 2020.
Not all data is created equal, and thus shouldn’t be given equal priority if your ultimate goal is speed-to-value for the business (which, ideally should be the case). It’s important to point out that speed-to-value doesn’t imply shortsightedness or sacrificing long-term goals in favor of short-term impact. Instead, speed-to-value, in this sense, means prioritizing the procedural data components that you believe are most critical toward advancing your organizational objectives. Simply, speed-to-value is about generating momentum quickly, and generating momentum in the right direction. While often overlooked, identifying and enforcing data process standards allow you to both hit your speed-to-value target(s) while also laying the groundwork for longer-term efficiency.
To start, one of the most effective ways to identify and prioritize data process needs is to work with your peers, especially operations and finance teams whose data needs are likely impacted in some way—whether upstream or downstream—by those of your own. These groups (or similar) are typically a solid starting point because they’re likely those closest and/or most responsible for delivery of critical organizational business objectives.
This isn’t the case in every organization, but can usually be an effective short-cut to quickly identify the most important objectives where obvious areas of overlap exist between teams and business units. You might then use already established outcomes and timelines to force action and closer collaboration between all involved, especially as a means to accelerate critical data conversations.
It’s critically important to keep your data governance practices current; this isn’t a set-it-and-forget-it kind of thing. These practices require constant re-evaluation to adapt to new demands, new products and services, and new data.
— Courtney Ulrich Smith, UX Designer, RevUnit
The governance practices that you need are almost assuredly not the ones you have in place right now. The list of people, systems, and tools that now need to be able to view, interpret, and analyze organizational data points continues to grow exponentially. So, too, does the rate at which all these things must create, store, ingest, analyze, and output actionable data. Left unaddressed or unmaintained, most data inconsistencies across your team, business unit, and organization may likely never be resolved. You may also run into other nightmare-ish scenarios (like regulatory compliance or data privacy issues), which no one wants.
Put simply; now is the time to either: (1) establish enterprise data management (EDM) governance practices in the first place, or (2) update and reinforce your existing governance methods on the fly. Yet, governance at scale can be difficult because many leaders struggle to truly understand which aspects of their governance practice are most deserving of additional investment and how much to invest.
Even still, many organizations—even those who are already actively investing to update their data governance systems—are struggling to maintain pace. In fact, Gartner recently reported that four out of five organizations that invest in data governance over the next three years will struggle to adapt to the new realities of a more data-driven business reality. Failure to implement sound governance practices can lead to a number of unwanted outcomes, including the continued inability to respond quickly to new opportunities when they arise.
Data silos exist in every organization; some more than others. It’s not necessarily important how those silos got there; it’s far more important that those silos be addressed and removed (ideally through a highly participatory process with representation from various stakeholders).
This process will take time—especially at an organizational level—and typically involves a variety of actors whose aim to develop, document, socialize, and enforce foundational governance standards that explicitly spell out things like data collection, storage, and use practices, management and maintenance processes, privacy rules, security protocols, and the systems, tools, and technologies that will power your governance practice.
Still, though, it’s likely that you’ll recognize a need to take a more active, immediate role on a much smaller scale — creating or implementing key data standardization practices within your own team(s). Practices or standards that you can institute on the fly so that you’re able to make better use of your existing data sets (assuming the underlying data quality is sound). If you’re just getting started, address some of the basics: for instance, are there standards in place that govern the labeling and storage of critical data? Are there methods of normalizing the data so that the right people, system, and tools can easily access the data? Answering these questions (and others like them) should help you identify common data uses, patterns, and queries.
The point here is to zero in on the most troublesome bottlenecks that directly affect your ability to positively contribute toward the business’ critical strategic priorities. This process doesn’t need to be overly exhaustive (HubSpot’s guidance on getting rid of data silos is a good place to start); instead, it’s meant to jumpstart movement toward key priorities.
Establishing a legitimate data literacy plan creates an additional opportunity to standardize your data, create processes that foster repeatability and predictability, and encourage a shared understanding among your business.
— C.J. Weatherford, Principal Designer, RevUnit
The ability to “speak data” is now a fundamental, required skill. Yet, Gartner’s most recent Chief Data Officer Survey found that ‘poor data literacy’ was rated the second biggest roadblock to creating a more data-driven culture; Gartner also reports that 50% of organizations today lack sufficient AI and data literacy skills to achieve business value, while 80% will initiate deliberate competency development in the field of data literacy to overcome “extreme deficiencies.”
These literacy deficiencies can sabotage your data transformation before it ever really gets off the ground; it’s the difference between teams and organizations who are able to regularly extract value from their data and those who become immediately vulnerable to outside competitive pressures. As a result, at a minimum, your team should possess the baseline foundations necessary to establish a common data vocabulary, specifically one that’s tailored to your unique data environment. Even a basic ability to communicate can be a force multiplier toward more effective data-driven decision making.
There are a number of tools that you can use to quickly assess data literacy among your team (think of them like any personality test you’ve ever done, but for data), but this assessment created by the Data Literacy Project is a good place to start, especially if your goal is simply to quickly assess individual and team literacy while creating conversation in rapid fashion.
We’ve created our own, custom courses made specifically for enterprise teams who need to “level-up” their data game. Feel free to check out our first data-driven course today — Data Visualization 101: Training for Enterprise Teams
Simply establishing a baseline for ongoing data literacy isn’t enough; if you plan to make a lasting impact, you should aim not just to champion data literacy for your own team (a good place to start) , but for your business unit as a whole — especially if no one else has stepped up to assume the role themselves. Assuming such a role won’t come without its challenges, but the impact on your team and organization can be felt almost immediately.
Even the slightest uptick in data proficiency allows individuals and teams to make more effective decisions, not to mention the increase in confidence that often results from increased proficiency. What’s more, data literate teams often have a more, holistic sense as to how the business operates, which typically produces more well-informed decisions. Thus, it’s up to you to do your homework here. If there isn’t a data literacy plan in place, find out why. If there is a plan in place, find out how you can support the initiative in a tactical way. If no plan is in the works, then find a co-sponsor (ideally at least one other senior leader) and begin to map a plan.
It’s important to identify and prioritize any needs and gaps in capability. The general maxim that the weakest link in the chain will break the entire chain holds true here. Not just now, but looking into the future. You need to create a process to continually review and prioritize these needs and gaps so that there are no weak links.
— Colin Shaw, Director of Machine Learning, RevUnit
Your most critical data needs are those that directly affect progress either toward or away from the strategic priorities the organization has deemed most important; these are your critical data elements (CDE). Essentially you can think of your CDE as the data that’s required in order to get the job done. This topic is deserving of much more commentary (perhaps for another time), but for now, your goal is three-fold: work with the necessary partners to define your CDE, minimize the number of CDE, and ensure the utmost accuracy for all CDE (this is where the importance of concepts like data lineage become incredibly important).
Identifying key data procedural needs (first step in this guide) will help to pinpoint your critical data elements (CDE). When paired with a strong enterprise data management (EDM) practice or well laid out governance structure, many of your most important data needs—and largest capability gaps—should float to the surface. For more tactical EDM guidance, we’ve found Tableau’s guide on EDM most practical.
Honestly assessing your ability to support your data needs is crucial; it can also be difficult to admit where legitimate weaknesses exist. Your needs will likely differ wildly from those of other teams or business units within your own organization; so, too, will your capability to support those needs. Nevertheless, it’s critical that you pinpoint exactly which gaps exist, why those gaps exist, and which options exist that will allow you to close those gaps quickly.
Start by outlining all of the data-related capabilities necessary in order to produce the desired outcome; nothing is off the table here (people, infrastructure, governance, maintenance, tools, tech, etc). This is also the time to examine less-obvious capabilities; for instance, is your team or organization armed with what it needs from a regulatory and/or compliance perspective in order to legally make effective use of said data? Is there pending legislation that may have consequences for your data collection, storage, or governance practices?
Questions like these are likely ones you’re not thinking about on a regular basis, but the honest answer to questions like this one play a significant role in determining how you go about closing data gaps.
Ideally, this type of exercise becomes a critical part of your data management practice (if it isn’t already). Your capabilities are likely to evolve over time, exposing new gaps along the way. These assessments, then, should be performed on a rolling basis, should have clear owners, and should be specific enough to drive recommended action to close said gaps.
There’s usually an appetite for small, controlled experiments — especially those you can map directly to a high-priority business objective. Doing your homework is important, sure, but at some point you just have to get your hands dirty and test.
— Michael Paladino, Co-Founder and Chief Operating Officer, RevUnit
This is the point at which the rubber meets the road; there’s really no point in identifying critical data capability gaps if you don’t intend to plug those gaps. You may be tempted to delay this part of the process, but you typically will have more to gain from targeted experimentation than you do from waiting for permission. Making quick progress requires someone who can see the bigger picture, identify the gaps, get others to see the gaps, and make progress toward identifying potential solutions to strengthen areas of weakness.
The lean toward rapid experimentation here is one that favors “stress testing” over inaction. Say, for instance, that you’ve identified business intelligence and/or data visualization as a low-risk, high-reward capability gap that, if solved, could immediately contribute to meaningful progress toward a key business objective. At a certain point, then (especially if you’ve identified tooling as a critical gap), you stand to gain much more from intentionally stress-testing tools like Tableau, Qlik, Looker, PowerBI, and others as opposed to continuing to compare the merits of each. Your job, then, is to bring a specific test case: outlining what you hope to accomplish, in what time frame, at what cost (to you, your team, and the business), and
Of course, this example is intentionally simplistic, but still “holds water” even if you’re looking to perhaps test the validity of something a bit more complex, like developing your own custom machine learning models, for instance.
Your goal is “show” not “tell.” That is, to either illustrate the potential of a certain solution or to potentially disprove it’s validity, which is just as useful an outcome in this scenario. Keep in mind that the name of the game here is speed-to-validity, not necessarily speed-to-perfection.
Your goal should be to monitor what you’re observing, compare against what you had expected, then use that information as inputs to determine efficacy (both in an experimental setting and projecting longer-term potential, if it exists).
As just a single individual (regardless of your title and position), you likely won’t be able to create a more data-driven organization without key partners and executive-level sponsorship. That’s true of most transformational change, so should be neither surprising nor discouraging.
That said, as a leader, you can control both how you and your teams make use of data, and thus, your ability to model the type of change you want to see within the rest of the organization. Doing so isn’t easy, but by taking consistent, deliberate action that’s aligned with what’s most important to the business, you’re more likely to deliver real results that create serious momentum.