Effective communication frameworks for efficient data model development: Insights from Wellthy from Coalesce 2023

"An accurate model is useless if it's delivered six months before or after it's needed…or if it can't actually answer the question at all."

Kelly Pook, Senior Analytics Engineer at Wellthy, discusses the challenges data teams face when building data models and offers some solutions. She shares her experience at Wellthy, a tech-enabled caregiving concierge platform, and provides insights into the importance of creating mock datasets and aligning them with business priorities for successful data modeling.

The communication gap in analytics and engineering

Kelly discusses the communication gap that often exists between the analytics and engineering teams in many organizations, particularly when it comes to developing models and dashboards. This can result in misaligned priorities and the inefficient use of resources.

Kelly notes, "...you have a lengthier development cycle for your models, requiring a lengthier review and promotion process." This means the cost of not getting it right in the first pull request is a lot higher.

Kelly advocates for a holistic and strategic approach to data, advising that "business priorities should determine the order in which you're ingesting your sources, modeling concepts out, even which tools you add to the stack." She argues that this can help prevent the cultural gap that often opens up in organizations that don't think about their data holistically or strategically.

The importance of mockups in bridging the communication gap

Kelly highlights mockups as a critical tool in bridging the communication gap between analysts and engineers. Not only do they provide a tangible reference point for discussion, but they also allow for iterative feedback and fine-tuning before any actual development work begins.

Kelly advises, "Create a mock dashboard. Pair it with a mock data set, and then go put that in front of the stakeholders before you start any work." She goes on to emphasize that this process helps uncover valuable requirements and potential pitfalls early on while providing a clear definition of what needs to be done.

Referring to the usefulness of this method, Kelly reveals, "Once you're done, this is what gets passed to the analytics engineer: the mock data set, the mock dashboard, and the definition of ‘done.’ And then, that allows me to do my own mockups. I love planning out my modeling before I even start. Again, mockups have kind of been the theme of the day."

The Impact of a holistic and strategic approach to data

Kelly shares the positive impact a strategic, holistic approach to data has had on their small team at Wellthy. Despite the initial time investment, the approach has resulted in scalable, impactful models and dashboards that have driven measurable business results.

She explains, "Our capacity models power our most used and most visible company metrics in dashboards. We have not had to modify or update these models since rolling them out earlier this year."

Regarding business impact, Kelly explains, "Our version of that capacity dashboard helped drive a 21% increase in our gross margin this year." She concludes that this kind of measurable and meaningful impact is what being an effective and impactful data team is all about.

Insights surfaced

• Communication gaps often lead to problems in data modeling
• Aligning with business priorities is crucial for successful data modeling
• Creating mock dashboards and data sets can help in understanding the requirements better and building the right data model
• It's important to work backward from stakeholders to analytics to data sources for effective data modeling
• The role of analytics engineers is to recreate the mock data set at scale and design and deploy the model with detailed requirements
• The process might involve some upfront work, but the payoff is significant in terms of scalability, immediate impact, and measurable results