Big Data Analytics: Build or Buy…or Both?

  • Posted by Zafar Daud
  • June 11, 2018 11:56 PM BST
  • s

by Zafar Daud, Solutions Architect, Guavus

When does it make sense to buy a big data analytics solution vs. building one in-house? This is a question that many departments within service provider organizations find themselves faced with time and time again. These departments are filled with highly skilled individuals who have the know-how and capability to build custom software with advanced capabilities.  In many cases, they are the technical owners of company-wide solutions that serve multiple functional areas such as operations, marketing, finance, purchasing, or customer service and are charged with finding solutions to meet the particular requirements of each department.

As Gartner points out in their Machine-Learning and Data Science, Build, Buy or Outsource report, making the decision to build or buy an analytics solution involves multiple factors such as the urgency of the business requirement, the time required to build it and the availability of skilled staff, among others.   It may make sense to build a custom software solution when there is no solution in the marketplace to meet your needs, the solutions available are too rigid, or are not compatible with your existing environment.  But this path is often expensive and time consuming. While there’s no one-size-fits-all decision, let’s review some of the most common factors to consider, based on our experience working with some of the top service providers worldwide.

Buy vs. Build
1.     Cost: When purchasing, you know the cost.  It is written in the quote.  Building a custom analytics solution can be an expensive endeavor and the cost can go up exponentially, after you’ve committed to the project.

2.     Time: Developing custom big data software takes a significant amount of time to design, build, and implement. Consider the cost in lost opportunities while you go through all these phases. Why spend time and cycles reinventing the wheel when you can deploy a solution that is available now?

3.     Support: This is one of the most overlooked considerations when going through the decision process of buy vs. build.  Consider the following:

• Other software, processes, and data structures change over time. Do you have support infrastructure in place to keep up with these changes? The people who build a custom software in-house may or may not be around to support later it due to the fluidity of the labor market today.
• Is your organization ready and staffed to keep up with new IT trends and changing regulatory environment, such as: cloud services, network, and data security, etc.?
• Do you have a process to manage upgrades and bug fixes?  This can get expensive and hard to manage over time.

4.     New Features: Software vendors are compelled to continuously improve their product by adding new features, implementing new technology, updating security patches, and keeping up with the latest industry trends.  Do you have teams who can do that?  Partnering with a reputable software solutions provider can give you these benefits as vendors need to evolve with the market or become obsolete.

5.     Best practices: By purchasing a solution you automatically benefit from the learning of others in your industry.  Cross-pollination of best practices is one of the most understated benefits of partnering with reputable software vendor that has a footprint and history in your industry.

Many service providers have found it makes better business-sense to buy than build for the reasons above.

A Hybrid Approach
However, what if the analytics solutions available on the market today don’t meet the specific needs of your company or a particular problem you need to solve?  It feels like your hand is forced to make the significant investment in building your own.  

There’s a third alternative many service providers are exploring: a hybrid solution - buying an analytics software framework, a fabric of sorts, and then building off of that. Complex data processing models and analytics algorithms are built-in and are ready to use, without needing years of in-house development time.  You can leverage the software to accelerate development of your own specific analytics applications using the analytics fabric to get the best of both worlds.