Data Forward: How to Unleash the Power of Your Organization's Data
Learn why a changing data landscape and the need to stay competitive in this digital age are the reasons to transform your organization to become "data forward."
“Data forward,” “data first” and “data centric” are among the phrases used by organizations today to define their strategy for interacting with data. Such organizations invest significant resources to unlock business value from data, and so the question is — how?
We believe that while successful companies can build data around them, exceptional companies need to be built around their data from the inside-out. The spawn of these digitally native companies built around their data was partially due to chance in the early days of the Internet boom. However, transforming your organization today into one isn’t a matter of luck anymore, but rather clear and direct strategy.
The simplest definition to a somewhat vague term — data forward — is the strategic philosophy of centering products to service all end users of data, rather than a specific application or technology. This article highlights tangible steps organizations can take to move in a data forward direction with examples of organizations that have successfully embarked on this journey.
Data as the basis of all business operation
Walmart: Commitment to transform
The retail industry has recently witnessed trends that include reduced visits to physical stores, the rise of online shopping and changing customer behavior focused on convenience, all of which have resulted in cost pressures only exacerbated by the pandemic. However, Walmart has stood apart during these changing times through its continuous investment in technology coupled with the adoption of a data forward strategy, enabling the organization to truly harness the power of data to drive its business success. 
Walmart recognized the need to shift to an omni-channel strategy and deliver a seamless customer experience across its physical and digital locations. The company also recognized that data would play a crucial role in successfully delivering that experience and began building a comprehensive view of its customers’ individual needs.  It also established “Data Café,” a dedicated analytics hub connecting to over 200 datasets both internal (e.g., transaction data, customer profiles) and external (e.g., social media data, weather data, location data) to successfully deliver a personalized and seamless shopping experience to its customers. 
The organization has continued to generate value across the entire customer journey by digitizing its physical stores and enabling its operations team to both restock shelves in real-time as well as forecast future customer demand. As a result, Walmart continues its journey of being successful by servicing customers faster, improving customer satisfaction and reducing costs. 
What is the process?
As was true for Walmart, becoming a data forward organization is a process rather than a decision. The following section outlines a framework and key elements to focus on along that path.
As shown in Figure 1, three aspects frame the behavior and infrastructure of a data forward organization:
- An inclusive definition of the end users for data with corresponding interaction points.
- Consistent underlying data layers for the interaction points.
- Multiple environments, including one specifically for data science.
1. An inclusive definition of the end-users for data with corresponding interaction points.
Traditional end users of data access and interact with that data primarily through applications, and organizations traditionally optimize the experience of accessing and working with data for that use case. But a data forward organization broadens the definition of who is considered a data user and what these users’ experiences with data are. These users should include everyone from a salesperson pulling quarterly performance numbers to a C-suite individual downloading a spreadsheet to check a statistic to a data scientist developing a model on petabytes of information.
An organization should even consider external partners or customers to be data users since they also leverage the organization’s data assets. The common feature across all these users is that they can generate business value from interactions with data, so Data Forward organizations should support them in that effort.
This often requires investing in additional or improved interaction points. These users likely have a wide range of data literacy and comfort with analytical techniques, so lowering the barriers to working with data will both help more fully capture the largest possible population of data users and generate value.
2. Consistent underlying data layers for the interaction points.
In addition, data forward organizations should maintain a single, consistent data layer undergirding each type of interaction point for end users. No large organization is able to keep all its data physically collocated; the data is instead dispersed geographically as appropriate. But data forward organizations maintain logically consistent data across those geographic boundaries via portals, metadata management and role-based access control.
This method allows for a more comprehensive view of an enterprise’s data assets, enabling different business units to have a holistic range of insight into customers and processes. And although it requires more planning and organization, this approach is also more efficient. By treating data as a primary shared asset throughout an organization, different business units no longer need to store duplicate tables or hoard information in silos. Removing those silos allows data users in one part of the organization to easily pull in data from another group if the need arises (and they have proper access controls) without having to go through a lengthy service request.
One contrasting example is that of an “application-centric” organization, which has a series of bespoke solutions for reporting and analytics with apps that are often connected to just a small number of databases (Figure 2). These point solutions may work well individually, but they often make connecting data across an enterprise more difficult, create the potential for redundancy and result in a diminished customer experience. In a data forward paradigm, all applications push and pull data to and from one location.
3. Multiple environments, including one specifically for data science.
That is, all applications within a given environment use one data location. The common data layer underneath each interaction point with data users corresponds to just one environment, ensuring data for different purposes is properly segregated. For instance, the data supplying production applications and models should be completely sanitized and inaccessible for any modeling or manipulation.
And similar to the common test environment where developers experiment with code before deployment, data forward organizations set up a “discovery environment” for data scientists and other highly skilled data users. The data available in this environment should be supplied from the production environment and likely comprises a periodically refreshed copy of data feeding production apps as well as a supply of the data generated by the interactions of users with those apps. This discovery data should be fully accessible (with the proper access controls in place) to data users to use when applying analytics techniques, and the separate environment allows them room to experiment and refine their work before publishing it in the form of models, dashboards or similar.
Setting up a discovery environment is also more effective, as data users can accelerate the training, testing and deployment pipeline of different data products while working in an experimental sandbox. This environment creates the ability for data users to “fail fast” and iterate quickly. And while the exact implementation may vary between organizations, a discovery environment will usually incorporate a raw zone, a discovery zone and on-demand zone. The raw zone acts as an initial landing zone for newly ingested, read-only data. The discovery zone is the primary working space for data scientists, allowing them the space to experiment and iterate with different modeling approaches. And the on-demand zone is supported by additional compute resources in the form of GPUs and CPUs so that models can be rapidly trained and tested before deployment.
None of these elements can be achieved overnight but starting with specific use cases and developing the necessary infrastructure for exposing high quality data to end-users is generally the first step to moving toward a data forward organization. Then, as user interest in accessing data grows, organizations are able to build further momentum for developing a consistent data layer and separating out a new environment for data science.
The use cases are essential to building organizational support for a data forward strategy by demonstrating value and will clarify which aspects of the data forward framework are most relevant for each organization.
Culture?! It’s all we have.
The New York Times: Transformation backed by cultural change
As The New York Times (NYT) set out in its ambition to become the ultimate destination for readers, it recognized that the shift in innovation within the media industry would leave the organization vulnerable if it did not accelerate its digital transformation process. It began by making a major shift to stop employing workarounds to accomplish its digital processes, instituting a clear, organization-wide strategy to tear apart those traditional technological barriers.
The organization began evaluating its processes by differentiating between ways of working that underpinned core NYT values vs. what they were doing because of the way processes had always been done. This shift played a crucial role in NYT setting a clear direction for its corporate culture, which would ultimately support the success of its new strategy. 
As a result, NYT now sets itself apart from its traditional competitors by being a “subscription-first business” instead of trying to maximize clicks, selling low-margin advertising against them and trying to win the pageviews race. This transformational success has been supported by the strength of NYT’s data culture, resulting in the implementation of new technological solutions which include real-time feeds, specialized newsletters and customizable parts to its news app, all powered by recommendation algorithms. 
The New York Times has added more digital subscribers during the pandemic period than it had during any quarter since implementing its new subscription service in 2011. Now with more than 5.7 million digital-only subscriptions, NYT feels that it is just getting started. It aims to continue this data-driven change even more rapidly to build a digital business large enough to support the company’s future ambitions on its own. 
What is the process?
NYT has proven that organizations who differentiate themselves from their peers are those that propagate an organizational-wide culture of data literacy and acceptance. This difference between tolerating data and accepting data is oceans wide, since the former instills an attitude of dealing with the surge of data while the latter spurs on employees to leverage data analytics and insights for every business process taken on. Transforming an organization into a data forward one is as much an OCM effort as it is a technical one.
We believe that it is the industry-wide trend toward building a data culture that has spawned the rise of the “Chief Data Officer (CDO)," a dedicated C-suite role to push all the initiatives described above. This is a good sign of organizations realizing that the push for truly being data forward has to be coupled with strong leadership support.
Here are some steps to start creating a culture of data-forwardness.
1. Promote the idea that information and data are shared and important assets.
Information is the foundation of our decision-making. Most corporate assets are carefully managed, and information should be no exception. Encourage the cross pollination of data initiatives to leverage the intellectual capital across business units. Stand up training opportunities for employees to become more data literate. Coaching employees to understand data and have the ability to manipulate and play with it is a precursor to viewing data as an invaluable asset necessary for innovation and progress.
2. Stop viewing everything data-related as exclusive to your IT department, but rather as the foundation for the entire enterprise’s success.
Gone are the days of viewing data and analytics as an IT-specific task. More than ever, businesses should realize the need to influence decision-making and strategy based on data and information. As long as data is seen as a siloed entity that belongs to the IT side of the enterprise, gaining buy-in from key business stakeholders will be more challenging.
Start by educating the business about the value and purpose of data collected. If the business starts to have an appreciation for what data are being stored and the insights that can be unlocked, discussions that were once viewed as IT-exclusive, such as data governance, management and security, will now start to have much larger business representation and buy-in. This reduces the perception that IT operates within a ‘black-box’ and simultaneously builds an appreciation for the significant data landscaping and engineering required to achieve the business outcomes expected.
3. Create a space for R&D.
Creating a “safe space” for internal exploration is truly showing the proof in the pudding. It is easy for an organization to say that it fosters an environment of data discovery. Actually having a channel for employees to pitch innovative ideas and making the resources available for them to pursue merit-worthy fronts also allows the company to leverage its in-house talent and gain from the benefits of research and development.
Google takes this to the next level — the company encourages its employees to pitch business ideas, and once an idea is deemed valuable, the product owner has unlimited access to any data within the company. Imagine that, a company’s worth of data available to anyone who kickstarts an idea or product. This approach has trickle-down benefits, since the data collected by a certain product eventually get reused by another, multiplying the value and knowledge across an organization. This is why it is hard to find a product of Google’s or any other tech giant’s without analytics and customer insight baked in.
A parting thought
In order to create long-term value in an ever-changing business environment, organizations have to be flexible and efficient. Being data forward gives an organization the opportunity to do just that by unlocking the ability to make business decisions based on meaningful insights, optimize business operations and create new revenue streams through innovation backed by data. Walmart has shown tremendous success by employing such a mindset and setting an example for other organizations to view their data fabric as entities with the ability to serve data users, rather than only specific applications or services.
In addition to addressing the technological pre-requisites, such fundamental change also needs to have a strong focus on the cultural shifts necessary to ensuring success. As much as the benefits of being data forward are clear, the journey to being a successful data forward organization is a marathon and not a sprint. There are a number of notable challenges that must be overcome to execute a winning transformation as clearly identified the New York Times’s emphasis on cultural adoption and adherence to process change.
Transforming an organization’s culture along with its existing processes is no easy task, because a data forward strategy affects the ways in which employees, customers and other data users interact with the organization’s data, behaviors which take time and intentional effort to undo. It is vital that all organizations today understand the importance of being data forward and why its success is dependent on the combination of technology, people, processes and most importantly, culture.
- Bryan Pearson, Walmart Is Investing in Shopper Data: How That Will Change the Grocery Aisle, www.forbes.com, Dec 21, 2017
- Jeff Loafman, Choice and Convenience: Omnichannel Around the World, Today and into the Future, www.corporate.walmart.com, Dec 9, 2020
- Bernard Marr, Really Big Data At Walmart: Real-Time Insights From Their 40+ Petabyte Data Cloud, www.forbes.com, Jan 23, 2017
- The New York Times, Report of the Group, Journalism That Stands Apart, www.nytimes.com, Jan, 2017
- Howard Tiersky, Navigating Digital Transformation, www.cio.com, June 8, 2017
- Anna Nicolaou, New York Times digital revenue passes print for the first time, www.ft.com, Aug, 2020