Identifying the Obstacles to Analytics Success
Identifying the Obstacles to Analytics Success
Why do so many companies fail in their attempts to adopt big data analytics, and what can they do to improve their chances for success?
Join the DZone community and get the full member experience.Join For Free
Learn how to operationalize machine learning and data science projects to monetize your AI initiatives. Download the Gartner report now.
For those of us in IT management, analytics is a familiar topic with a familiar value proposition. But for many organizations, success with IT analytics is less familiar. Why do so many fail in their adoption attempts? And how can you avoid failure and embrace success? As we’ll see below, the problems usually boil down to products, people, and processes.
An Overabundance of Analytics Tools
Simply put, there are too many tools to choose from. The overabundance of tools complicates our decision about which tool will adequately and reliably meet our needs. Going through the implementation cycle with each tool being considered usually isn’t feasible, yet the only way to know if a tool is going to work for you is to put it to work.
An Abundance of Analytics Tool Categories
Related to the sheer number of tools is the number of tool categories in the analytics tool market, such as:
- Dashboard tools that focus on presenting information at a high level for end users.
- ETL tools that extract data in various formats and transform it in a meaningful format for analysis and presentation.
- Streaming data tools that assume data is coming in at the high volumes and high rates of speed associated with big data.
- Statistical modeling tools that perform predictive analysis and slice-and-dice data.
- Self-service analytics tools, an emerging class of software that lets regular, non-expert end users explore data, look for patterns, run what-if scenarios and create dashboards.
Of course, IT pros might not be familiar with any of the tool categories above — unless they are data experts. But if you don’t know the categories, it can take a long time to figure out what your organization, department or end users need. And those needs can vary by department, team and end user.
Silos Stifle Analytics
Organizational departments tend to work in silos. Each team has its own isolated data set that is not shared with other internal teams or departments. Consider a customer who interacts with a company’s pre-sales, marketing, sales, and support teams over the course of a year. If the teams don’t see each other’s customer data, nobody has a holistic view of the customer. It’s very difficult to correlate the teams’ disparate data sets and present an account manager or other manager with an end-to-end view of the customer. As a result, the company is missing a big chance to improve its internal processes as well as its customer service.
Executive Sponsorship Error
For most internal analytics projects, you need an executive to back the project and secure the budget, people, and resources to make it happen. Without an executive sponsor, the project will fall through. If the team tasked with setting up the analytics solution fails to hit key milestones on time, the sponsor may lose interest. Whoever is championing the analytics strategy must keep the sponsor updated and provide visibility into the project and its progress. Similarly, the champion must convince the sponsor that the analytics project is a must-have — rather than a nice-to-have — technology, otherwise, it will get cut the next time the company is looking for ways to reduce costs.
Some metrics look good on paper but have little value or significance. Take gross sales, for instance. While the number itself may be impressive, it ignores sales-related expenses and gives little indication of a company’s fiscal health or profitability. Such metrics are known as “vanity metrics,” and you don’t want to measure them. At the executive level, picking the “right metrics” usually boils down to industry. As an example, the industry-specific metrics associated with a retailer might include categories such as apparel, shoes, and accessories as well as store location. In addition to that data, retail executives might want a simple dashboard that shows sales distribution and the percentage of sales online vs. in store.
Not every user is looking for the same kind of analysis, dashboards or reports. And not every user is a data expert. That means the analytics tool has to be easy to customize, implement and use. You want a non-expert user to be able to quickly and easily customize it themselves rather than relying on third parties, either in-house analytics pros or external consultants. With such a flexible tool, analytics teams can be agile and turn around new reports, analyze data in new and different ways, and they can do it fast.
To accommodate the wide range of user types, an analytics tool should support different data types or formats and ingest and correlate multiple data sets. And regardless of the user’s skill level, it should let users create dashboards and reports as well as visualize and present data in their preferred view such as a pie chart or bar chart.
The challenges above constitute common roadblocks to the analytics success. By recognizing the challenges and avoiding the mistakes others have already made, you improve your odds of an analytics implementation that gives your organization what it needs and your users what they want.
Published at DZone with permission of Sridhar Iyengar . See the original article here.
Opinions expressed by DZone contributors are their own.