By: Rafael Fanchini on July 15th, 2019

Print/Save as PDF

Why You Need a Hybrid Data Analytics Model for Your Business

Data Analytics | Augmented Analytics | Data Science

Cowritten by Rafael Franchini and Betsy Romeri, 

The goal of automating data-driven solutions for your business is far from impossible. Businesses need to look at data analytics from a fresh perspective and embrace the next generation of analytics, which are essential for all companies—big and small. Given the vast differences in the level of analytics maturity in the current market, there is no one-size-fits-all roadmap to data-driven solutions.

Data Analytics is a must-have. As impressive use-cases for advanced analytics in business continue to roll out, the adoption of this disruptive technology is quickly becoming the number-one strategic business initiative across industry verticals. From a strategic viewpoint, there are two fundamental reasons for a company to adopt data analytics: 

  • To generate tangible value through specialized applications that leverage and/or automate recurring decisions
  • To augment and improve how specific business decisions are made through leveraging data analytics 

Although businesses and executives struggle to keep up with the fast pace of changing technology, the need to establish a more data-driven, decision-making environment is well understood and quickly becoming a "must have." When developing analytic capabilities for your company,  you are faced with many choices. First, should you hire a team of data scientists and build your own analytic applications and systems?  Second, should you buy dozens of software solutions to relieve your pain points?  Finally, should you opt to outsource all of your analytics needs? 

3 Considerations for Developing Analytic Capabilities for Your Company



Formula One Car 2


Building an internal data-science team is often assumed to be the best option for businesses. Business executives believe this is a way to build customized analytic capabilities and retain all IP. If the endeavor is successful, these beliefs can become true facts. Based on our experience, however, this approach is most often not the best option.  Without a deep understanding of the specific skills and experience levels required, companies often build teams that do not have the data-science experience or technical capabilities to build the solutions they need.

A typical rollout to establish an internal data-science team starts with the IT department hiring a small group of data scientists with different areas of expertise to be overseen by an executive responsible for the strategic initiative. The result we have most often observed, six to twelve months down the road, is that IT is left overseeing an expensive, inexperienced, and ill-equipped team who will succumb to pressure coming from executives to address business issues, which are more like discretionary wishes than analytically solvable challenges. These attempts almost always result in a long time-to-value, poor adoption, and low ROI. In reality, building a data-science team for your company has to be done with the precision of building a Formula One Race Car Team. 

In the Formula One analogy, each team member has specialized skills which must fit seamlessly together—from engine design to race car design, the pit crew, and a highly skilled driver to bring the trophy home. Similarly, a data-science team is made up of members with very specific skillsets that need to fit seamlessly from the Machine Learning model design to analytic application production, data engineering, and the executive to guide the project toward realizing its value within the business. Building a Formula One Data Science Team is an extremely difficult and expensive endeavor that is more likely to fail than to succeed. 


Purchasing preconfigured software solutions to address business pain points is the quickest path toward solutions. However, with this approach, companies can give up customized solutions and an advantage over competitors since competitors have access to the same solutions. These are the obvious shortcomings of this approach. 

While packaged software solutions are built to address a set of problems, that set of problems is predetermined by the software provider. There is rarely a "perfect fit" between a company’s needs and a software package. There is no such thing as an all-in-one analytics solution that is optimized for every business. In addition, patching together new software with existing software in order to build a data-driven business has numerous layers of complication that require the IT department to patch and maintain the system. 

We have never found a company with just one business problem, which means many software solutions will be required by a company. The risks of integration and adoption delays are multiplied with each purchase. Integrating all solution platforms to make them work together will require full-time IT efforts, and this is extremely difficult to do well. Poor implementation and adoption of new software is a risk for any company. The level of data-analytics maturity or knowledge within a company is typically not sophisticated enough to understand exactly which software is a good fit that will:

  • seamlessly integrate with legacy systems and
  • be adopted by employees


Fully outsourcing all data analytics is an option similar to having a completely internal data-science team— except that the team would be external to the company. Like the internal data-science team, it would require that each member of the external team have the skills, experience, and business knowledge specific to your business. An external team could be composed of an ecosystem of vendors or a single entity.

One basic, absolute requirement for any data-analytics effort is the involvement of business-domain experts to help identify a company’s pain points, which can be addressed with data-analytics solutions. The domain experts must work closely with the business stakeholders, data scientists, and engineers to formulate a strategy and identify the data needed to solve a problem. The best business-domain experts are those that work for the business looking to solve a problem. They are at ground zero; bringing in outside domain experts who do not know your business leaves you with a consultancy model, which will be a long-term engagement at a very high price with very uncertain results.

Typically, companies that decide to outsource data analytics are the very companies that do not have a business- domain expert with enough knowledge to drive the effort and work with an outside group of engineers and scientists. With no internal domain expert, whether an IT professional or a business stakeholder, completely outsourcing data analytics will fail. Therefore, it is just as difficult to build a Formula One-quality external data science team that is customized to your business as it is to build an internal team.

Planning and executing a data-analytics strategy is a very complex process. We believe that spending valuable time and budget focused exclusively on building an internal team, buying multiple software solutions, or completely outsourcing your data-analytics capabilities will result in solutions that are not repeatable, scalable, or profitable.

If You Can’t Have a Formula One Team, Go Hybrid


Hybrid 2


In terms of the best customized data-analytics model for a given business, there is no simple answer. Your best option is certain to be a hybrid data-analytics model—comprised of some building, some buying, and some outsourcing. How much emphasis should be given to each endeavor will be the subject of a future blog post.


Connect with Rafael Fanchini

Email Rafael at:

Connect with Rafael on LinkedIn here.