There’s been much said about the promise of big data and what enterprises can achieve by harnessing it. So brands have made significant investments in analytics tools and process.
There are many analytics options to choose from—“analytics” continues to be a truly big tent without a clear definition. Yet despite analytics proliferation, there are still many kinks to be worked out.
There’s not a single problem, but several related technology problems that we hear over and over from brands trying to leverage analytics to drive continuous, data-driven improvement across the enterprise. In no particular order, here are the top 10 most frequently cited analytics pain points.
1) One-Shot
Frequently, analysis is undertaken in an ad hoc way—a one-shot process used to find value. But enterprises wanting to improve the business continuously need analytics to be systematic and repeating.
2) Insight—A Means, Not an End
Many products promise to convert data to “insight.” But what is insight? All too frequently, “insight” is a static report or an interactive dashboard with beautiful graphs that let you slice-and-dice a ton of data any way you want.
Neither add immediate value to the business. For them to matter, someone needs to expend effort to make sense of it all, and figure out what actions should be taken. Businesses aren’t making investments in analytics because they need insight. It’s a means to an end, and the end is “answers.” Business want answers-specific, practical actions to take to improve the metrics they care about.
3) No Scale
Some enterprises are collecting high volumes (terabytes) of data every from machines, transactions, and beyond, yet many tools and methods can’t keep pace with the volume and speed brands are collecting.
4) Post-Mortem—Half a Solution
Much analysis on offer today is a post-mortem look at old data to determine what happened and why (descriptive analytics), in order to make beneficial changes in the future. It is helpful to know that customers who are male, and have a particular product, and have called the contact center 3 times churn within the first month unless offered a promotion. If you offer a promotion to customer meeting this “profile,” you will reduce churn – that’s the valuable takeaway.
But to capture this value, you have to know when you’re talking to a customer with this profile, so you can make the offer. And you need to know all the thousands of other profiles that lead to churn. That’s where predictive analytics becomes critical. It gives you the ability to recognize what events, transactions, interactions are likely to lead to a particular outcome—such as churn—and identify them as they’re happening so you know you need to act and can do so at the right moment.
5) Biased & Incomplete
Because most analysis requires humans to query data, the results of the analysis illustrate only the questions the analyst or data scientist thought to ask, ensuring that answers are biased and incomplete.
6) Answers Stale—High Opportunity Cost
Most analytics still requires experts to spend months and even years cleansing, querying, coding, and modeling before real answers are produced and changes to tactics are deployed. And many enterprises collect a high-volume of data—terabytes daily—exacerbating the problem.
By the time these answers are produced and new tactics are deployed, they’re stale, even obsolete because of the long cycle times. This is because competitors, customers, and environmental pressures change the facts on the ground every second, minute, day, or week, depending on your business. When businesses can’t analyze incoming data quickly enough to respond to changes in the market, the opportunity cost is huge.
7) High Effort
The installation and set-up of most analytical tools can be manually intensive. What’s more, the actual analysis, frequently takes humans hours, days, weeks, or months of querying, coding, modeling, experimentation, and deployment.
8) Untapped Data
Some analysis methods and tools only analyze numerical data, and not categorical values. Most organizations aren’t joining related data across siloes to try to understand how variables captured in one department’s system combine with variables in another department’s system to drive a KPI up or down. And free-form text and unstructured data from sources like email, social media, and calls is a treasure trove of intel, but is rarely mined.
Brands have invested significant resources in wringing value from data, but many are only tapping a small percentage of data available to them, leaving enormous value on the table.
9) Accessibility
Most tools available, from coding-heavy data science toolkits like R to drag-and-drop studios, require users to have significant expertise in data science, statistics, coding and software to transform data, choose and develop models, etc. But the people who need to leverage data are department managers—users without this expertise.
10) Time-to-Value
Installing tools and software packages is complex and takes time, and the set-up required to get started creates a long lead time to value. But enterprises want to get started right away, and many can’t afford to wait.
What are your biggest pain points?
