Space cover image

Data Science

Data science is easy when three elements meet:

  • clear objectives that yield high business value

  • curious minds with a can-do mindset

  • reasonable data quality & infrastructure

Storytelling for impactful Data Science

Let’s face it. There’s a huge gap between data science culture and other business units, especially sales or marketing. How to make an impact when we don’t speak the same language?

Overcoming communication gaps

If you seek understanding, offer it first.

  1. What is the current business strategy and what goals does your organization tries to achieve?

  2. How would you best support revenue-generating teams? What do they think you shall do for them? Listen and write it down as you hear it.

  3. Do you agree with them or what are the main differences?

Knowing is power. Use it well.

Structure your work

Data science work can be chaotic with a lot of uncertainty. We never know if our ad-hoc analyses provide valuable insights or if our models train well and perform on unseen datasets. Maybe we again hit data quality issues and spend weeks duck-taping our pipelines.

  1. Keep track-record of your hypotheses and experiments. Update them when before you jump to another project.

  2. Document your data quality and write a roadmap for improvements - prioritize them to maximize business impact with the lowest effort

  3. The document you models - nothing is worse than data-science work that has close to no documentation and is misused for the application it has never been created for or not used well enough just because people do not understand it

The context for your Charts & Tables

Data glossary

Maybe you know what all those attributes and metrics mean and how are derived metrics calculated. But do your users know it too? Try to get what a scatter plot with “recency” on the x-axis and “frequency” on the y-axis truly means. What definition of “retention” do you actually use?

Building a glossary for metrics in your organization improves clarity and reduces misunderstanding. Maybe business people do use your analyses because just falsely assume different definitions. And very often also you team members.