Why so many analytics projects fail when moving out of the lab and into the field


It’s been a few years now since companies began dabbling with big data and analytics applications, and a clear shift is now happening. Enterprises are beginning to move from “experimental analytics” to “industrialized analytics” as they get a better feel for the kinds of business outcomes they can realize.

But it’s not the kind of shift en masse we might expect. Companies are struggling to translate the results of skunkworks projects into viable solutions that deliver measurable business outcomes. As s result, a significant number of analytics applications fail to move from the lab to the field. How does that happen?

Often companies jump into analytics without establishing outcomes or use cases. Instead, many projects are chartered as experimental exercises driven by internal teams to see what might be possible. Without a grounding in real business outcomes or viable uses cases, many projects (that seem to go on and on) deliver underwhelming results in the end. Without demonstrated results, sponsorship and excitement wanes and those projects get put on the shelf, along with the other failed experiments of the past.

Even when the discovery phase of an analytics project is successful and sponsors are on board, a lot needs to happen. Rollouts often require work to be done on the data pipeline, infrastructure and environment to make sure the data that’s needed for analytics models will be available in real or near-real time. Once those models are deployed, there is an ongoing need to tune and maintain them.

Integration is another big task. You need to be able to integrate analytical apps into existing business apps and processes like your CRM or MRP or solutions like Salesforce. Integration is the critical step that turns analytics output into actual business benefits.

The field of analytics continues to evolve and mature. More options are becoming available to help companies experiment and deploy faster and more effectively. End-to-end analytics platforms combine a scalable infrastructure and all of the latest analytics software with pre-built configurations and templates. The services needed to identify good use cases, find the right data sources, ask the right questions and integrate the insights are included as well. That not only reduces time-to-value, it also reduces risk. And because platform solutions are built on scalable cloud infrastructure, analytics workloads are always operating on “right-sized” resources that can be added or subtracted as needed.

Every company can benefit from analytics. And when you consider what a platform-based approach offers compared to the prevailing DIY approach, it becomes easier to see how your company can finally get its ideas out of the lab and into action.

In future blog posts, we’ll highlight more of the challenges that companies encounter as they seek to raise their “analytics IQ,” and offer strategies to overcome them. We’ll discuss the best practices we’ve discovered that will help any company stand up an analytics solution with confidence.

Ashim Bose headshotAshim Bose is Global Leader of Analytics Product Management, DXC Analytics. He is focused on helping clients achieve business outcomes from their data by leveraging DXC Analytics offerings. He has over 20 years of industry experience in automotive, industrial, airlines, telecom and space exploration. Ashim holds a Ph.D. in artificial intelligence and a master’s degree in mechanical engineering.

Speak Your Mind


This site uses Akismet to reduce spam. Learn how your comment data is processed.