How to increase adoption of your big data analytics platform

Although vendor-written, this contributed piece does not promote a product or service and has been edited and approved by Network World editors.

When I talk to data practitioners, something that comes up quite often is how their work with data has become an exact science. But I’ve heard repeatedly that inspiring non-technical staff to understand what to do with data insights can be a slow and painful process.

The problem is the more data that is involved, the more complex and rich the story behind the data becomes. But businesses increasingly want the “Cliffs notes” to these stories, as they don’t always have the luxury to understand all the nuances. While it can be hard for engineers and scientists to distill multiple data streams into key insights, here are five simple steps I use to help simplify things.

1. Show, Don’t Tell.  It is imperative to make it clear the platform will be consolidating data across the organization. Let’s face it: people will not be supportive until they see how the platform can benefit them by solving difficult problems they would not have been able to solve otherwise. By demonstrating how the platform quickly aggregates, cleans and analyzes data and produces impactful outputs that they were not able to accomplish before will greatly increase the chances of adoption.

2. Prepare for everything ­– Yes, even that.  The ability to be able to bring in data without preprocessing and without the need to conform to a schema is alien to many potential adopters. The idea that a platform is prepared for everything (i.e. any type of data) creates fear, uncertainty and doubt because it appears too good to be true. But if adoption is key, the platform needs to be able to ingest any type of data (in any language or format) and provide complementary tools to clean, standardize and analyze.

3. Don’t make early adopters sweat the small stuff.  One of the biggest logjams to making a platform “just work” is accepting any data type ­ and making sense of it. Take the following of data:  (800) 555­1234, 800­555­1234,  1­(800)­555­1234.  While a human can tell these refer to the same data point, data systems will typically see these as three different data points. This problem becomes especially pronounced at scale and when you consider capturing data across text, end user input, voice, etc. Having an unstructured data capture system is a good first step, but having a platform that can automatically sort and parse the data will help make your platform excel.

4. You don’t need to see the future, but you need to learn from the past.  Teams spend an inordinate amount of time in preparing for predictive analytics without first understanding the value of the data they have. In many cases, they end up creating products that are not a fit for the data they have. A platform that can provide insights into the value of their data will help a team better plan for predictive analytic products that get useful results.

5. Base your system on value, not looks.  Visualization is trending right now, but if the visuals are based on poor functionality ­non-technical staff will be making bad decisions. Visualizations are designed to simplify data, and it is easy for it be taken out of context. Ease and speed of analysis is more important than pretty visualizations as visuals can always be redesigned later, but a base foundation cannot.

Chala leads the development of exploratory data analysis, data streaming and business intelligence and has served as a key member in bringing the HPCC Systems platform to the open source community.

Copyright © 2016 IDG Communications, Inc.

The 10 most powerful companies in enterprise networking 2022