Thursday, April 11, 2019

Is it time to reboot your Analytics Program? Try thinking like a Product Manager


It’s been a long time between posts; but now I want to share a common theme from my recent consulting experience. I have spent much of my time working with organizations that either had no serious analytics capability at all, or were getting little or no value from what they had. These engagements, mostly with smaller firms and non-profits, reminded me of my enterprise clients from a decade ago. They felt they had some level of business intelligence and analytics, but it was not working well for some or all of the following reasons:
  • They had plenty of static reports being generated, but most largely ignored as being obsolete
  • Presentation was almost entirely spreadsheet-based, with embedded logic compensating for the weaknesses in the source reports
  • There was no effective data governance effort, let alone a formal one with an accountable person running it. This created doubt in the reliability of the reporting
  • Often, there were some dashboards being generated;  but they were there mostly for the sake of having them and failing for the same reasons the reports were
  • Dashboards and visualizations were being presented without the benefit of context or any level of storytelling
  • The reports and dashboards were undocumented, and the original authors long gone; or if they remained, they were a single point of failure
  • There would be a few scattered desktop purchases of visualization tools like Tableau and Qlik, but no real plan for fully adopting the technology
  • Those tools were being used to generate visualizations that were long on style and short on value
  • Data was not timely or updated automatically
  • Mobile delivery, if it existed, did not translate well to small screens with limited bandwidth

There was no plan in place to climb the BI capability ladder– so they decided to call for help.
The most common root cause of a failed BI program is the Big Lie the person who sold the tools often tells: that buying state of the art BI tech will, by itself, solve a business problem out of the box.

Successful BI executives know that BI implementations are software products, not just tools or implementation programs. This implies that they must think like a product manager. Product managers know that great tech is necessary but not sufficient to create a successful product. Great products need marketing, production control, quality control, user support, documentation, and an achievable rollout, maintenance and enhancement plan.

All of these traits apply to BI programs that that seek to create a great user experience defined by data quality, system reliability, accessibility, ease of use, and support when it is needed. The goal is to provide your users at all levels of the organization with a decision support capability. This requires people, and process assets to go along with the technology.

One note about people: Experienced BI pros are often scarce and overpriced.  Try to find your data analysts and system admins from within. They already understand your business and provide context. They are usually easy to train up on modern tech which has become very usable for those with little or no coding background.

Process is where outside help can be most valuable. Experienced consultants can expedite the establishment of key IT processes that are vital to success, including:
  • Data acquisition across multiple source systems
  • Data governance and quality control
  • Strategic and tactical alignment: Fitting BI products to business decisions, target audiences, and use cases
  • Self-service capability and support
  • Prioritizing maintenance and enhancement requests as part of release management
  • Putting the necessary security in place
  • This is critical - Marketing your BI products

If your organization already has a product management function, seek their advice. They know how to do this in the context of your business and can help you apply their best practices to your BI program.

No comments:

Post a Comment