The data-driven organization’s hierarchy of needs
What it means to be data-driven
Every company wants to be more data-driven, but how does one go about making it happen? As they grow, many startups (large/small, consumer/enterprise) will travel on a well-trodden path of data-related challenges:
- What type of data professional should I hire?
- What tools do I need? Should I build or buy?
- Which metric should I use to measure X?
- What is the best team structure given my current business needs?
The list goes on and on. Each question probably deserves its own blogpost, but first I wanted to share a general framework for thinking about the evolution of a data-driven company, based on my own experiences and conversations with other analytics leaders.
Hierarchy of needs
The framework is based on Maslow’s hierarchy of needs for human growth. The basic idea is that you have to meet peoples’ most primitive, fundamental needs before satisfying higher needs. For example, if you cannot get access to food and shelter, things like friendship and self-confidence are irrelevant.
Applying the analogy, I believe there are foundational building blocks of a data-driven organization that must be in place before you can move on to the next layer:
Reliability. Before anyone can start working with data, they need to trust that it’s accurate. Otherwise it’s garbage-in, garbage out. This means data sources, data pipelines, and data stores perform as expected. When these systems go down, analysis grinds to a halt. If data infrastructure, maintenance, or quality problems are a consistent theme, everything else takes a backseat.
Literacy. It’s difficult to deliver on the full value of a piece of analysis until people are speaking a common language. So it’s important to cultivate a quantitative understanding of how the business runs, characterized by the company KPIs. Depending on the business model, this may be transactions, daily active users, or monthly recurring revenue. Whatever they are, all employees should expect to have basic reporting of these stats on a dashboard every day. There may not even be anyone doing actual analysis, but at least people will see how numbers are trending. You’d be surprised how far a little visibility into the data can go.
Insight. Once people can see high-level metrics, they will want to understand their drivers. If monthly user retention is decreasing, the next order of business is figuring out the underlying cause. This next level of the pyramid is the ability to answer the “why” for a wide range of business questions. The goal is to identify the causal relationships between the levers the business controls and the metrics that matter to the business. Examples include segmenting users by lifetime value, or building a model to predict likelihood of churn. These insights are used to inform business decisions — which product features to ship, which growth channels to invest in, and which shiny objects to avoid.
Adoption. If you’re good at answering questions, the next step is to do it an integrated manner. This stage represents the widespread adoption of a data-driven approach to the business, where decisions are backed by data and teams are held accountable to measurable goals. Instead of fielding ad-hoc requests, analysts partner closely with business stakeholders as part of everyday workflows. Instead of running simple queries, analysts proactively push out nuanced analyses, while non-technical employees answer their own questions in true self-serve fashion.
Automation. The highest level of the hierarchy is the full realization of the business as a well-oiled machine running on autopilot. In this stage, data is deeply rooted in the fabric of the company culture. Analysis itself is easily executed and highly reusable due to well-designed tools that streamline and remove the pain in the process. Experimentation and algorithms automate the analyses used to make product and marketing decisions. Analytics resources are widely embedded throughout the organization, and significant engineering resources are dedicated to data infrastructure. Everything just works!
Conclusion
The hierarchy of needs is a checklist to ensure your data organization and where you need to focus your time and resources. To summarize:
- Data is clean, accurate, and reliable.
- There is sharing understanding and reporting on important metrics.
- One can perform meaningful analysis with the data you have.
- Analysis can be done within integrated workflows and the right contexts.
- Frictions are removed to automate the decision-making lifecycle.
Finally, if you’re interested in helping build the next great data-driven company, Lyft is hiring!
