Over the last few posts, I’ve discussed the need to embed intelligence into the core of your organization—to make it pervasive. Some of the feedback I’ve gotten concerns the difficulty in making intelligence pervasive when many companies struggle with simply implementing analytics to begin with.
To be sure, most mid-to-large companies have analytics capabilities in place, but many also certainly struggle with those capabilities. However, it is precisely because you struggle that you must invest in making intelligence pervasive.
Even if you’re struggling with your analytics program, there are steps you can take to embed analytics into the fabric of the organization. Taking those steps can help you do two things. One, it can ease your analytics struggles, and two, it can give you a foundation on which to build your analytics future and provide improved ROI.
Step One: Start Fast, Fail Fast, and Learn
My mom likes to use the word dither. It means to vacillate, to be ambivalent, to go in circles without getting much accomplished. If you’re shaking your head ruefully right now, it’s probably this best describes the state of your analytics program. It doesn’t have to.
The first step in making intelligence pervasive is simply to start. Pick a high-value proof-of-concept (PoC) project, get the right data and the right tools, and solve the problem. Don’t be afraid to fail; you will. Learn from it and go on. Modify your model, succeed, and spread the success from the PoC to other areas.
The key to starting fast, and learning from inevitable failure, is to use an approach that combines processes and tools to address the gap from insight to pervasive intelligence and realized value. This systemized, or ‘AnalyticOps’ approach productionalizes analytics, making it an almost industrial process. AnalyticOps stresses automation, standardization, and governance. It brings analytics from theory and numbers to production and value, and it helps unlock the value inherent in your organization—and it does this at scale.
Step Two: Design for the People who use the System
People use systems. That’s on the first page of the book of duh. But still, it bears keeping in mind because, when I talk to a lot of clients, their number one complaint is that, despite investing significant resources in expensive tools, users still engage in spreadsheet anarchy—they don’t use the tool and instead create their own information silos.
They don’t do it maliciously; they do it so they can do their job. So why not invest in giving them what they really need. How do you know what they need? Ask. Seriously. Ask them. That’s where the theory of cognitive design comes in. Cognitive design stresses empathy and understanding what end users need to meet their goals. And it’s a simple equation. If users meet their goals, the systems you develop will meet yours.
It really is that simple. Cognitive design hits that sweet spot between technical feasibility, business viability, and end user needs. Starting from a foundation that presupposes empathy, cognitive design provides clear definition on the business outcomes you’re delivering, enables understanding of your users’ end-to-end needs–emotional, functional, and technical—and delivers a clear vision of the end-to-end solution.
What this gives you is analytics capabilities that meet the needs of the users they were designed for, thus enabling you to get the answers you seek—no matter who asks the question. Cognitive design goes a long way in eliminating data silos, because it reduces spreadsheet anarchy via encouraging users to trust—and therefore to use—the systems you design. And, if these systems integrate data that spans functions and divisions, that intelligence is, de-facto, pervasive.
Nothing I’ve said here is new. It just bears repeating because I hear over and over again that companies are struggling with their analytics—often needlessly. Sure, there’ll be difficulties in your analytics journey. Nothing worthwhile is easy. But, you don’t have to unnecessarily complicate things by, as my mother would say, dithering.
Pick a place to start. Start, learn from rough spots you encounter, and productionalize the results so that they’re repeatable and provide demonstrable ROI. Do that by designing systems that truly meet user needs so that people actually use them, and the information you strive to collect and analyze can serve its purpose.
This approach isn’t easy. It requires you to question everything about the way you implement your intelligence capabilities and to make a commitment to change. But you can do it. And, if you’re struggling now, isn’t the risk worth the potential payoff?