In This Article

A common question we receive from customers is: "How can I achieve a faster return-on-investment when developing new data capabilities?" Customers will often start their journey growing data maturity with a data strategy assessment effort and then follow it with a data science proof-of-concept (POC). The former results in a strategic framework and roadmap for building a high-value data organization; the latter provides immediate benefit derived from mining your data. 

Some organizations switch the order and start with the data science POC efforts first -- but why choose? We recommend doing both at the same time.

When building data maturity -- something we formally define via the six stages of the data maturity curve  -- there are always aspects that are more foundational in nature and take longer to develop before delivering tangible cost savings and/or revenue growth (e.g. building a data platform, standing up a data governance program). Often it is assumed these foundational efforts must occur first because a messy data environment will be too difficult to work in effectively and will distract from the strategic aims of building the proper data foundation.

The accelerated approach directly challenges this assumption. It consists of a two parallel streams to advance data maturity. The first stream is focused on strategic development and the second stream is focused on a data science POC. In essence, it combines a top-down approach with a bottom-up approach. 

From firsthand experience, we have determined the hands-on data science work conducted in parallel to strategy work provides noticeable advantages:

  1. Working in the existing data environment with targeted goals provides direct experience and findings that are used to supplement and reinforce the strategic recommendations.
  2. It generates excitement within the organization by showcasing the potential for more sophisticated data capabilities by addressing real problems with real data.
  3. It provides immediate, tangible business value to compliment the strategic outcomes of the overall initiative.
data strategy assessment and strategic roadmap

But can you start data science work that fast?

Often, we hear concerns from customers about getting started with data science right away. Comments usually sound something like: 

While it's true a data science team will be much more efficient and drive higher value with better data quality and a modern data platform integrated with key data sources, we firmly believe it is a mistake to wait until these milestones have been achieved before getting started with data science. In fact, we believe actively conducting AI and ML work helps to accelerate and prioritize data maturation efforts (e.g. by identifying where and how to focus data quality improvement efforts). 

More importantly, it provides immediate value to the business and helps to justify investment into data capabilities. We believe a much larger risk is for a company to invest heavily into building a modern data platform, standing up a data governance program and creating a new data organization, but then not yield business value quick enough or to the magnitude expected. 

What does this look like in practice?

WWT has always leveraged a use-case driven approach to assessing and growing data maturity -- this serves as the "top-down" portion of the approach as previously described. A key output from this exercise is a strategic roadmap (usually spanning multiple years) that outlines a plan to develop a holistic set of data capabilities that supports the needs identified from the assessment efforts.

The combination with the second stream, or "bottom-up" portion of the approach, is the data science POC. We recommend using a small team (typically a few data scientists and business analysts) to pursue 1-2 well defined data science use cases. The team's goal should be to move quickly and to prioritize progress -- this is especially true for customers who are lower on the data maturity curve where it can be easy to get stuck in the details dealing with challenges of messy data. 

Figuring out which data science use cases to even pursue in the first place can be very time consuming. Here we offer a couple of recommendations:

  1. Find-a-Friendly – Work with business leaders who are excited about the prospect of data science and have a high data IQ. You want to work with groups who are motivated to support the data science efforts and have a good understanding of the potential value from data, especially at a leadership level. This is important to keep the POC moving at a quick pace.
  2. Progress Over Perfection – It can be tempting to find the "perfect" use case, which is usually up for debate, but typically is based on some combination of business value vs. complexity. Often the business value portion is highly subjective and depends on a leader's priorities, while complexity is usually poorly understood. What is most important is to get started on something that seems reasonable from a value and complexity perspective. The goal shouldn't be to pick the "perfect" use case, but rather to avoid analysis paralysis so you can get started on the actual data science work as quickly as possible.
  3. Pivot Quickly – Often data science use cases are closely intertwined, especially if they are within the same domain. If you start down a path that doesn't yield the results expected or run into an unforeseen challenge, don't be afraid to move on to a use case that looks more promising (especially as you learn more information). When you let the data do the speaking, the results are not guaranteed, and it is important to be flexible in approach.

Don't wait -- get started now! 

While advancing data maturity is not an easy task -- especially getting started -- we believe combining a strategic approach with hands-on data science work is the best and fastest way to increase the value your organization is getting from data. This approach combines the need to think hard about strategically investing into foundational data capabilities while simultaneously starting data-driven decision making. 

Whether your organization is just getting started on its data journey or well on its way, we hope this approach can help accelerate your journey!

Questions? Reach out to us to discuss your unique objectives.