Data has permeated business operations so rapidly that companies often presume they’ve become a data-driven organization when, on principle, they are not. In fact, some of our most forward-thinking clients arrived at Headstorm with the realization that data-driven transformation — just like culture-driven transformation — required a level of active planning & supervision that they’d simply taken for granted at earlier stages of growth.
Here are some of the imposter syndromes we’ve seen companies perpetuating while they mistakenly paint themselves as “data-driven”:
- Data-as-a-Shoehorn: the business still operates on the whims and gut feelings of the loudest executives, and data is merely used as a lever to achieve those agendas
- Data-as-a-Skunkworks: there are a handful of mysterious folks off in the corner who make magic happen with data, but no one else in the org knows what to do with them
- Data-as-a-Swamp: the business generates a ton of data, but management isn’t sure how to navigate it or build something stable on top of it
It’s important to recognize that there’s no fault to be laid here, necessarily. First, you had some robust Excel reporting on the traditional business. Then came the analytics platforms, SaaS providers, and automated operations. Eventually the business reached a state where it was ingesting big data, sensitive data, unstructured data, you name it. This process happened at light speed for many orgs, and what needs to be understood is that it’s never too late to take the reins on it if you intend to have control over where data is going to take your business.
This is a clean slate moment: the time to conceptualize and structure your organization’s data-driven transformation. Below is a simple, introductory path.
1. Data Governance
Data governance ought to be the first phase of transformation, and it proves just how important top-down buy-in is. Your data governance strategy will define how you ultimately gather, contain, process and apply your data — internally or externally. This strategy, when fully composed, ensures that every employee has the necessary access and familiarity to the data they need — and only what they need — to optimize their business function and identify opportunities. Here are a few initial questions to ask your org regarding data governance:
- Data Quality. How is your data cleaned, and what gets scrubbed off? What’s the life cycle of your data, and who can observe it along any part of that cycle? Is there automation in your data mining & data cleansing process, and how does that automation alter its state? What standards and ground truths do you hold new data sources to?
- Data Security. Do you encrypt your data in transit and at rest? Who utilizes your data externally, and who do those parties interact with? Are employee roles or security levels set to ensure? Does a new hire understand how they can apply for these roles? Does your org implement a zero trust or risk-based approach to security?
- Metadata Management. Where does a particular data set come from, and how reliably can the organization recall the source as the data gets applied reworked? Does a particular data set contain personally identifiable information (PII) or require GDPR concessions? Can you think of ways to utilize connections in metadata to stitch together useful correlations & combinations?
Clearly, these are not questions one person — or even one department — can or should answer. And again, it’s the reason buy-in from the top is imperative, and also why it may never have happened in your organization while data grew unchecked.
Having a data governance strategy means documenting and democratizing the responsibility; once that framework is laid out, upskilling and integrating your entire org to be data-driven becomes significantly easier. As an example, consider how many of your employees can (or should) use macros in Excel for day-to-day work:
- If an employee can build an Excel macro, they can learn SQL
- If they can learn SQL, they can sit at the table with engineers and data scientists in order to exchange ideas across product, tech, and business needs
- If they can do all of the above, they can optimize their own business function to automate their work, standardize KPI reporting, integrate other operations, and unlock interdepartmental opportunities for data-driven efficiencies
2. Data Structure
Once your organization is structured to manage data, the data itself can be managed and maximized. To break it down into two camps, you’ve got structured and unstructured data.
Structured data, at this point, is probably well-managed by your organization and its platforms. Structured data is pre-formatted and constrained in a way that both people and machines can typically understand — e.g., addresses, phone numbers, credit card numbers, etc. Given how readily it’s found in the wild, and how formulaic it is, the challenge of managing structured data is more around its security than its utility.
What’s skyrocketing in availability and potential is the other side of the coin: unstructured data. More likely to be proprietary, unstructured data has no ubiquitous format — this might include sensor data from your products, image data produced by your facilities, or text/voice data generated by your customers.
The rub with unstructured data is that you decide, for better or worse, what to do with it. A phone number is a phone number — its utility is inherent. But consider that your organization might receive monthly satellite images of all its facilities… what do you do with that? How unstructured data is utilized says a lot about whether a business has truly undergone a data-driven transformation, because a well-governed data-driven org will have employees from all departments chiming in with theories and proposals for the practical use of satellite imagery:
- Finance might want to gauge real estate trends based on the pace of adjacent lot development in order to consider investment strategies
- Marketing could see the images as an opportunity for competitive intelligence research
- Operations may learn that weather patterns drastically impact use of outdoor space and parking facilities, which could lead to recommendations in telecommuting policies or indoor space for congregation & co-working
As may seem obvious at this point, the missing puzzle piece to activating unstructured data is technology — a machine that can be put to task in order to report on and learn from the data, so that departments can implement the wisdom. Enter: machine learning.
3. Machine Learning (And A.I.)
While AI is the blanket buzzword familiar to most CxOs, most of the unstructured data you have access to will be operationally maximized under some practical subset of AI; namely, machine learning (or ML).
Simply put, ML is data-driven programming that improves over time, through reinforcement and supervision. Terabytes of satellite images are too cumbersome for humans to manage (hence the moniker “big data”), but also too unstructured for engineers to simply apply an if:then process to, as traditional programming would dictate. If humans can teach a machine to improve a process on its own, that process is ripe for machine learning.
Those satellite images, and all the great ideas your employees conceived around them, require machine learning to come to fruition. The investments made in such ML programs are often substantial, which is all the more reason why an organization must first know why and how it can make use of the data to drive business goals. For companies struggling in their own data swamps, it is often these efforts in highly advanced (and expensive) programs misleading them into believing they are data-driven, when in reality, their inability to navigate and implement the data is bogging them down.
Pitching Data-Driven Transformation
Data-driven transformation is never achieved by a sole actor. The organizational or departmental majority needs to buy into the benefits of this strategy, both short and long term. However, the ball always gets rolling with one small push, and anyone with a plan can provide that push. Using the guidance above to assess responsibility, security, workflows, costs and benefits can expose just how much further your organization can go when it truly adopts data-driven transformation.
Of course, organizations often prefer to have experienced Sherpas guiding them on this path to transformation — that’s where Headstorm comes in, and we’re happy to talk with your team about how we can help.