In my decade + career as an Analytics worker (Title fanciness changes every few years Statistician, Business Analyst, Analytics Consultant and Data Scientist) other than the absolute fact to revamp my skill set every few months one predicament keeps shadowing me all along — “How to drive adoption of the analytics product amongst those it is intended for?”
Tell me if this situation sounds familiar to you — You have been in conversation with business stakeholders for building an all inclusive grand dashboard or a machine learning model. The outputs according to you are very insightful and should be used everyday for scientific driven decision making by the end users. But usage statistics over time reveals low or zero adoption rates.
It has taken me great courage to come out of denial and accept, that an astounding number of analytics projects I have been associated with have eventually become “shelf-ware” with minimal to zero users of the outputs.
While enterprise analytics tools are undergoing a massive shift with a focus on user experience, it is the responsibility of the Data Science teams to enhance the user adoption with a focus on “consumerization”
“It used to be enough to provide tools that improved productivity for the business. Now, every tool that is used by employees must provide a world-class user experience. Employees will not adopt tools without a memorable experience. — Forbes
As everything today from Healthcare to Banking to Meal delivery kits shift to a “user experience” and “consumerization” model, here are 4 guiding principles that have worked successfully for me in Analytics:
Start with the right questions: This article by Benjamin Cooley has coined a valuable acronym QTBA (Questions to be answered). Building an analytics product aimed to deliver insights to the users is fundamentally different than other web products. Engage the users to pen down the answers to these questions:
- What are your goals from the solution?
- What business challenges or questions are you trying to solve?
- What key performance measures are you to trying to monitor?
- What are the deficiencies or challenges of your current data and insights model?
- And all other transactional questions about — frequency, trends history, business rules, slice and dice dimensions etc.
Be mindful of User Experience (UX) — Do not underestimate the importance of consciously applied UX principles in the visualization products. Very important to keep the below principles -
- Uni-directional navigation — Simplify the usage of filters, drill downs and roll ups
- Minimize clutter in the visuals — what do you want the user to take away?
- Include tool tips explaining what the visual is communicating and how the user should interpret it
- Allow the users to customize their views (mini dashboard within a dashboard)
- Consistency in font colors and sizes and spellings
- Do not overuse colors or deviate from the color palette template
- Do not forget — multi device compatibility such as iPads, phones etc.
Engage the end consumer all along — Often times you will receive requirements on behalf of someone else. This could be a senior leader or a manager that decides what their teams want from the analytics solution or more so on how their teams should be viewing the solution. Or even worse, it is a designated analytics person or team with no understanding of the day to day operational nuances. While this is directionally useful but is eventually ineffective because the end user wants something entirely different from the requirements originally received. Do yourself and your project team a favor and insist on engaging the end user from the beginning. It is also an important facet in agile delivery as referenced in my previous blog. I am never too far from the memory of this famous illustration in these situations.
Empathetic Partnerships — Demonstrate to the end users that you are fully committed in their transformation goals and are agile in adopting to their changing needs. And of course, as they say “charity begins at home” it starts with these harmonious partnerships amongst your cross-functional project teams.
In conclusion, all of these principles assumes that the underlying data is completely impeccable and does so consistently. Data integrity and Governance are the pillars for Analytics and nothing kills user adoption and experience like the slightest apprehensions of data quality and integrity.
Thoughts? Leave a comment below or find me on Twitter