From Buzzword to Business Strategy: Four Tips for Achieving Digital Transformation Success

Remember the good old days when every organization thrived by building in Total Quality Management (TQM) processes and training operations staff in all the glories of less-than-four-incidents-per-million Six Sigma methodology? Or what about that "Net Promoter Score" -- the?"One Number You Need To Grow" -- all other consumer metrics be damned? Right, me either...

Well, while undoubtedly some businesses did, in fact, enact foundational improvements utilizing frameworks and metrics like these in their heyday, many of them have been relegated to the annuls of "organizational buzz trends" where we should have known better than a one-size-fits-all approach rarely works is easy as the Harvard Business Review would lead you to believe.

Today, there seem to be a lot of organizations going through a "digital transformation" as they attempt to:

• Migrate local, on-prem compute resources to cloud VMs for cost, scaling, and elasticity optimization
• Develop data analytics skills within business-side employees and promote data-driven decision-making over gut instinct
• Achieve additional bottom-line earnings and/or savings by enabling machine learning (ML) processes and decision automation
• Ceaselessly toss around the term "digital transformation" to placate shareholders and investors while having only a cursory plan in place for what they actually will do to achieve this

Ouch -- what about that last one? That bullet point aside, these are all great (and achievable) goals for what a digital transformation should result in, but often these initiatives lose their way and turn into financial money pits. Let me preface the rest of this article by saying I don't think the term digital transformation is destined for the dusty file cabinet of business trends of yore, but I do think without careful planning and execution they could seem like that's where they belong when we look back from 30 years in the future.

Here are a few tips for achieving digital transformation success:

1. Start With the Data

Although maybe it's fallen out of favor to say that "Data is the New Oil," the analogy is still relevant. Just like you aren't going to get an airplane to fly or manufacture your polyester disco pants without first refining raw crude oil, the same can be said about data and its downstream use cases and applications: if you don't start with the data you're setting your organization up for a rough ride in its digital transformation.

While this may not elicit an epiphany to hear, it's surprisingly common how many organizations, large and small, suffer from not starting with the data. We can rationalize this problem as one structured by the "Hype Cycle" developed by consulting firm Gartner (although, personally, I think they should have called it the "Hype Rollercoaster" given its non-cyclical construct).

According to a quick Google search, the term "digital transformation" became commonplace around the year 2011 -- around the same time Google Trends indicates an upward slope around the search terms "data science" and "ML." While I'm not saying these were the drivers of digital transformations for every organization around this time, I feel it's safe to say that many of them were driven, in part, by the new benefits and business value of robotic process and decision-making automation promised by ML. So, what happened? Millions of dollars were funneled into ML projects in order to 'Keep Up With the Joneses' (i.e., competitors doing the same under the same promised benefits of achieving a competitive advantage in their industry). And you know what? A lot of organizations did achieve some value from these early initiatives -- the only problem is that they were unsustainable and not reproducible because they skipped several steps early on.

So, Tip One is to Start with the Data. While it may be boring compared to all the glitz and glamor ML receives every downstream application will fail if your data assets are not well understood, properly maintained, and governed tightly. The framework I adopted at Expected X for putting a stop to this "cart-before-the-horse" practice is CMMI's Data Maturity Model -- a graduated approach for developing a data-centric organization from the ground up. Without going into too much detail, the framework covers six major categories "that help organizations benchmark their capabilities, identify strengths and gaps, and leverage their data assets to improve business performance" including Data Strategy, Data Governance, Data Quality, Platform & Architecture, Data Operations, and Supporting Processes.

2. Not Just Another IT Initiative

When they hear the term "digital transformation," many executives believe that the onus of responsibility for enabling this lies mostly (or worse, solely) within the IT department. While, indeed, the majority of operational sub-initiatives may fall under their banner, leave no doubt that successful digital transformations are exercises in organization-wide change management involving every employee from CEO Alice to Mailroom Bob.

There is a strong "If You Build It, They Will Come" mentality in organizations that set themselves up for failed digital transformations. Not everyone is an analytics expert, not everyone trusts making data-driven decisions, and not everyone is capable of learning a new software system quickly. On top of simply getting people to use the new tools of a digital transformation rather than stick with their bread-and-butter methods, business users (internal) and customers/clients (external) need ongoing training and support.

Tip Two is that a digital transformation is Not Just Another IT Initiative. A common misstep in this area is the belief that "shiny-new-tech-tools" are the be-all/end-all solution to their success. The latest and greatest can quickly lose their luster, especially if you're an early adopter of bug-riddled software. More importantly, it's the requirements gathering and up-front consulting with your end users regarding their feature expectations, their needs, and answering questions like:

• "How will this make my job easier?"
• "How much will this save/earn the organization compared to the 'old' way of doing things?"
• "Where can I turn for support when I have a question about a new process or tool?"
• "Can I trust and explain the predictions this ML model is making compared to my own?"

These are not simple questions to answer and finding people who can navigate the gray area between linking business KPIs and other monetary objectives to the technical performance requirements of these new tools and systems is often difficult (although these people do exist -- wink, wink). However, giving it only a perfunctory effort or avoiding it entirely will spur costlier rework if not a total failure of the digital transformation.

3. Don't Scale Too Fast!

When promises of competitive advantage are blended with millions of dollars invested to achieve this level of supremacy, organizations tend to get a little carried away -- especially when those promises are passed on to the shareholders who expect timely results and dividend payouts. Due to this, many digital transformations tend to "leap before they look" and initiate several projects simultaneously without a real plan for how these projects will all fit together as part of a greater, cohesive whole (more on that later).

While quick wins are obviously important to show Proof of Concept to both external and internal stakeholders, the importance of methodically developing the platform cannot be understated. The term "platform" means a few different things in the context of a digital transformation:

1. "Platform" is a reference to the actual tools used to develop the systems of the digital transformation -- a standardized approach to each component: data engineering, data governance, etc. (see CMMI's Data Maturity Model mentioned earlier). Are all of your disparate teams united under a single platform? This isn't as easy as it sounds but thankfully most cloud solution providers today offer standard managed services and approaches for platform implementation regardless of industry or application. Before scaling out and investing in several initiatives for your digital transformation, be sure you have a standardized approach across your teams.
2. "Platform" also means the ways of working within the digital transformation -- the human element. How are project teams assembled? Do they always have representation from each contributing group (i.e., data engineering, data science, ML engineering, business SMEs, etc.)? How do they gather requirements from stakeholders and end users? How do they plan their product releases? How do they document their solution designs?

I often tell colleagues "my job is to make your job boring!" This usually raises a few eyebrows and might not sound appealing, but what could make a job boring? Well, how about eliminating manual, repetitive tasks that require human intervention? This should be the whole impetus for a digital transformation in the first place the last time I checked! No, ChatGPT is not going to take all of our jobs -- it's going to reduce or eliminate the repetitive and sometimes redundant, menial chores that make many of us dread punching the proverbial clock every day. This gives us all more time to focus on things humans excel at like creative and abstract thinking (well, at least until we write some code for that, too).

Let me address this from another perspective: Why are companies like Apple and McDonald's successful? Because they've engineered a platform -- Standard Operating Procedures -- before they engineered a product. If you buy an iPhone in Brazil and shoot on over to, say, Morocco and buy another, both iPhones should be essentially identical internally. Same with McDonald's -- buy a Big Mac in Indiana or Hawai'i and you're getting the same thing even though it's created by teams of people nearly 5,000 miles separated (although a Spam Musubi may be the preferred lunch choice in the latter).

Tip Three, Don't Scale Too Fast! Standardize your platform as it relates to the two definitions above and create a team culture that embraces process before product. Only then should you consider scaling your digital transformation initiatives across multiple use cases.

4. Create a Centralized Data Product Hub

Why are 'Big Box' stores and mega e-tailers like Target, Walmart, and Amazon so successful? Well, aside from the cost savings economies-of-scale typically provides these organizations, it's also the convenience factor -- everything you might need all in one place. In keeping with the concept that digital transformations should be thought of holistically and implemented (as best as possible) under a single, standardized platform on the back end, it makes sense to centralize user access management on the front end as well.

The goal here is twofold:

1. The benefit end-users receive from having a "Single Pane of Glass" or "Single Source of Truth." Everything is available in one location and brought together by one overarching piece of software. Productivity (as well as motivation) declines the longer it takes someone to find what they are looking for to complete the task at hand.
2. Having all of your new tools together in one place also provides an intangible, psychological benefit related to what I mentioned previously: a digital transformation is organizational-wide. It is not centralized into one or a few teams or departments. Everyone is part of the change, everyone is experiencing it together, and this change is designed to make us all work together better and more efficiently (yes, that means CEO Alice and Mailroom Bob are all playing together in the same sandbox)!

Most major tech hubs like 微软 and IBM provide tools for IAM (Identity and Access Management) as well as dedicated providers like Okta and Ping Identity . Cloud providers like Amazon Web Services (AWS) , Google Cloud , and Microsoft Azure Cloud all have managed services for IAM as well. These allow us to customize access to fit users' needs so they are not totally inundated with dashboards, tables, and other data products.

Why is it important to have a "Centralized Data Product Hub?" While the business-centric idiom "Too Many Cooks in the Kitchen Spoil the Soup" is one of those eye-rollers we've all probably heard more times than we can count, it's nevertheless meaningful when describing why not doing this can negatively impact an organization currently going through this change:

• If you have too many data and/or product managers working in a siloed environment you start to create redundant data products that address the same business problems.
• You start to duplicate data, (cloud) managed services, documentation, employees, and other types of tech debt which all drive up costs and begin to negate the benefits of the digital transformation.
• Teams will stray off and begin to develop their own platforms which will fracture the foundation needed in Rule 3 and slowly dissolve the belief that a digital transformation truly is a holistic undertaking.

Tip Four, is to Create a Centralized Data Product Hub. This doesn't really differ much from what we've already talked about up to this point, it's just focused on the technology delivery aspect -- the User Experience.

Remember, if a tree falls in the woods and no one is around to hear it, does it make a sound? Similarly, if $100M is invested in a digital transformation and no one knows how to use the tools you gave them, was it a waste of money?

I hope some of these tips will provide you with pause and reflection, especially if you're just starting to think about a digital transformation in your organization. Good luck on your journey!

DISCLOSURE: The ideas expressed above are mine alone and do not reflect any affiliation with any business entity I am formerly or currently employed by.

John Sukup ?is AI/ML Release Engineer at Levi Strauss and Founder/Principal Consultant at Expected X, a data strategy and MLOps solutions consultancy providing clients with detailed, graduated approaches to realizing the full potential of working with data. His experience working with data spans 16 years from consumer market research, to data science, to machine learning engineering. He has acted as the lead professional trainer for machine learning and related topics at Cisco Systems and has been featured in Forbes, Oracle, and Data Science Central.