Guided analytics templates for self-service analytics
As this anecdotal evidence using search trends suggests, there is increasing interest in self-service. As advances occur in big data and analytics platforms, so do the expectations of being able to get things done by ourselves.
Self-service analytics is table stakes: Everyone expects it
Stepping back and looking at macro trends in analysis, it seems we have gone back-and-forth between centralizing and decentralizing the ability to do ad hoc analysis. As enterprises grappled with enabling users to perform analysis, they had to deal with controls and governance surrounding data quality. This would lead to large, multiyear projects to build out data warehouses, supplemented by centralized reporting capabilities. The side effect of this was to dramatically increase time to value, and the time to react to different data and analysis needs.
The next generation of analysis platforms has focused on enabling the end user, and shortening time to value. This has largely been a great experience for the user (more power to users), but not quite as much for those responsible for data stewardship. In response, these platform vendors are now working their way back to balance enterprise control and governance needs.
But the key takeaway is this: self-service is table stakes. You can’t have a meaningful analytics platform without it.
The opportunity: Users need help envisioning ways to apply analytics
As Spider-Man’s uncle reminded him, “With great power comes great responsibility.” Accordingly, there’s an opportunity in the way things stack up today. With easier-to-use analytics tools at their disposal, users now have the responsibility to figure out how to apply them to their problem. Consider, for example, a finance manager at a telco trying to understand how weather impacts revenue. Yes, she may have a bunch of tools more accessible than before, but if she’s going DIY, how does she answer her question? Which data should she use? What questions should she ask? How should she interpret the analysis results?
We believe users need a bit of help getting started from someone that knows their stuff. They want something that strikes the right balance against how prescriptive it gets, and allows users to fly but with someone trusted at their side. And, by the way, they want all this yesterday.
- What data should I use?
- What questions should I ask?
- What is this analysis suggesting I do?
- I know exactly what data to use
- Just pick from a list of great starting points, explore as I like
- Explanations help me focus on what matters, and courses of action
Experts in the self-service age
There’s a new, compelling, evolving role for experts in the age of self-service. It starts as a mix of domain knowledge and data science skills. Over time, it becomes more sophisticated as experts acquire proprietary knowledge and intellectual property. Further, as cognitive computing allows the platform to learn and tune itself over time, experts harness this power and channel it for the end user.
But back to the beginning: experts can identify the right data for a problem by taking generic tools and creating an analytic starting point around that data. They cherry-pick interesting visualizations while writing explanations to help the end user interpret results.
A community of experts are creating these guided analytic templates (or Expert Storybooks) and forming a corpus of knowledge one template at a time. End users benefit tremendously, and a positive cycle kicks in.
Expert Storybooks on IBM Watson Analytics
On 14 October, 2015, IBM announced its intent to launch Expert Storybooks that deliver on this vision. We have an impressive list of Business Partners that believe in this vision, and are teaming up with us to offer these storybooks.
I’m excited to be part of this journey and thrilled to be working with a world-class team of clients, Business Partners and IBMers that are together bringing this idea to life.