3 Essential Ingredients to Achieving Data Dominance in 2021


Data teams that are more agile and adaptable will be better equipped to stay competitive and realize the benefits of AI across the enterprise.

The ever-changing nature of business over the past year convinced enterprises across industries of the necessity of data and AI in decision making. However, not everyone has mastered how to use these tools effectively yet. Many enterprises still struggle with formulating the strategic business advantage they can unlock by applying AI, let alone developing and deploying AI solutions. Venturing an educated estimate, I’d say we are probably less than 5% down the path of achieving the full potential of AI across enterprises.

Recently, I’ve thought more broadly about what industry changes we’ll see this coming year — what opportunities are available for the data and AI industries, and what challenges teams will need to overcome. Looking ahead, I believe there are three key steps enterprises will need to take to stay competitive and face what the next year will bring.

You need the data to be readily available and high-quality

An obvious first step to working data, and AI algorithms trained on that data, is to have data to use. It’s rare to find an enterprise today not actively gathering data on most aspects of their business, so this shouldn’t be a major concern anymore.

Less obvious is how to optimize what data you keep and where you keep it so it can be quickly accessed for accurate and helpful processes. In the past few years, we’ve seen a boom in demand for data scientists to assist with this effort and an associated rise of platforms targeted to best meet those data scientists’ needs. Industries with these data assets already in place and ready to roll will quickly realize the potential of AI in 2021.

We can learn a lot from industries that already have data assets in place and are poised to advance AI faster, as well as the steps later bloomers are currently taking to get there. For example, enterprises in the media and entertainment space are poised to benefit greatly from deploying AI as their industry has recently made a strong push into digital, equipping them with the necessary data assets. With the rise of online streaming platforms, companies now have a wealth of data about their viewers and can harness that data into AI algorithms that customize programming to the individual.

In other industries, there is still room for data leaders to emerge as enterprises slowly realize the potential but, in many cases, first, have to get their data assets in place. Retail is a prime example of an industry where there are currently a few data-first leaders (read: Amazon and, by extension, Whole Foods) that are making clear for the rest of the industry that data assets, like purchasing behavior tied to individuals, will be essential for competitive advantage and survival.

You need the expertise and technical skills

Even with the right data assets in place, you’ll struggle to get any value out of them without the right talent. The speed with which AI is deployed across industries heavily depends on the expertise and technical resources to deploy ML. These skills are in high demand, with Data Scientist occupying the #2 spot on Glassdoor’s Top 50 Jobs in America for 2021. This skill set takes time to develop, which is why the supply of highly skilled data scientists still hasn’t caught up to demand.

Proficiency in data and AI is essential in data scientists, but if they don’t know anything about your business, they won’t be able to be as impactful. The ideal combination of expertise in the specific business with data and AI skills should be a priority for data teams. Neither are sufficient by themselves. This can happen either by training up people already familiar with the business to be data scientists or hiring data experts and embedding them deeply in the business.

The biggest opportunity for AI in 2021 is still to broaden its accessibility, which also necessitates increased training efforts. If you are really serious about tapping into the power of your data at every level, you should expand your training efforts across your organization. With more departments trained with at least a basic knowledge of data and AI to make them self-sufficient in smaller tasks, your data scientists and engineers can focus more sharply on AI efforts that will drive your enterprise farther into the future.

You need to be agile and adaptable

2020 highlighted the need for more agile and adaptable data teams. In a world of COVID and economic recession, a company with ten years of historical data has no advantage over a company that just started out. With ever-changing data and patterns, data teams need to be able to react quickly and not waste months on deploying simple solutions. Data teams that are more agile and adaptable will be better equipped to meet the challenge of technical questions such as, “How can AI be applied when the amount of historical data is small?”

To create agile teams, you need three key elements:

  1. The team members themselves need to stay up to date on techniques and training. Management can help create self-sufficiency in learning by encouraging, incentivizing, and subsidizing ongoing education and mentorship.
  2. The technology you use needs to be able to deal with small data. Examples of this include transfer learning, where you can take advantage of models that have been pre-trained on similar but unrelated data and then fine-tuned to your specific problem with your own data.
  3. When the data team comes up with creative ways to make an impact on the business, realizing that impact is often constrained by the speed at which they can deliver on that idea. This makes the capability to quickly deliver innovations critical. This is often summarized under MLOps and requires well-defined processes and reducing manual steps in the deployment lifecycle.

Regardless of where you stand in this process currently, there is no doubt that harnessing data and deploying sophisticated AI is no longer a luxury for today’s enterprises – it’s a necessity. The longer you wait to take full advantage of your data, the further behind you will fall from the competition. Make sure you are devoting the time and resources now to invest in your data and AI strategy by ensuring your organization has the right data in the right place so that the right teams can utilize the right skills. Implementing these ingredients now will serve as a foundation that will help your business survive the current moment and thrive as you build upon that foundation to lead out in AI implementation in the future.

Clemens Mewald

About Clemens Mewald

Clemens Mewald is Director of Product Management, Machine Learning and Data Science at Databricks. He leads the product team for Machine Learning and Data Science at Databricks. Previously, he spent four years on the Google Brain team building AI infrastructure for Alphabet, where his product portfolio included TensorFlow and TensorFlow Extended (TFX). Clemens holds an MSc in computer science from UAS Wiener Neustadt, Austria, and an MBA from MIT Sloan.

Leave a Reply

Your email address will not be published. Required fields are marked *