Choosing the Right Data Projects as Head of Analytics

As the leader of data analytics, you'll be tasked with managing various crucial aspects of a company. This can make it difficult to decide which data projects to prioritize, especially when faced with a barrage of requests from management. This article will guide you in selecting your key projects effectively.

To start, it's helpful to categorize your projects into three major types: analytics, automation, and data products. Let's delve into each type and discuss considerations for tackling them.

Analytics Projects

Analytics projects can provide quick wins by starting with core questions from the business and yielding insights.

But how do you identify a worthwhile data analytics project? What do these projects entail?

Characteristics of Analytics Projects

Analytics projects can take various forms, such as ad-hoc reports, refined Jupyter Notebooks, dashboards, or presentations. Each format can convey a narrative to stakeholders about business performance.

Typically, an initial phase of ad-hoc analysis will help determine the value of deeper data exploration. Once you complete the initial data review, you'll need to present your findings.

The way you present your analysis will depend on both your preferences and those of your stakeholders. The essential aspect is to create a compelling narrative. Based on your ad-hoc analysis, aim to extract one or two key takeaways for your management to focus on.

Ensure your message is clear and that your data visualizations reinforce it.

Selecting Your Analytics Projects

How you showcase your projects is vital, but it’s equally important to engage in initiatives that deliver tangible business benefits.

To pinpoint high-impact projects for your organization, consult with stakeholders to identify pressing questions and understand how addressing these will affect the business. Engaging with department heads and directors is crucial for this.

Consider asking the following questions, inspired by Ethan Aaron, CEO of Portable and former Head of BI:

Which Key Performance Indicators (KPIs) do you track for your business?
Where do you find these metrics?
What metrics would you like to access daily?
What actions do you take based on these metrics?
How do you gauge the impact on the business? Is it essential?

Ethan emphasizes that the purpose of analytics is not merely to present data; it's about driving actions that significantly affect the business.

Once you understand the questions your leaders have, you can compile a list of valuable projects that align with their objectives.

Automation Projects

Automation leads to several key benefits:

Enhanced scalability of processes
Cost reduction
Minimization of human error

However, because much of the automation work occurs behind the scenes, it may not capture management's attention unless it significantly influences their bottom line.

Nonetheless, robust automated systems are vital for any data-driven organization, including automated data pipelines, quality assurance processes, and model deployment.

Finding suitable automation projects can be challenging. While automation saves time, it can also introduce technical debt, even with low-code solutions.

What Should You Automate?

If you've been in tech for a while, you might be familiar with the common guideline for automation. It's not a definitive rule, but it effectively illustrates which processes should be automated.

Focus on projects that are currently time-consuming or that will become burdensome as your organization scales. If a task isn't frequently required, automating it may not be cost-effective.

Automation can be exciting for engineers who enjoy observing seamless processes, but it also adds another codebase to manage. Therefore, the decision to automate should be based on time efficiency.

While C-suite executives might overlook automation projects, analysts, project managers, and data scientists often identify processes ripe for automation. The question is whether these processes should be automated.

Data Product Projects

Once your team excels at analytics and automation, you can explore building data products. It's essential to have a solid grasp of your analysis processes and a reliable automated data infrastructure before venturing into data products.

Data products may include processed datasets that categorize inputs with natural language processing, insightful dashboards, or APIs for managing dynamic pricing.

Organizations often turn to external providers for data products due to limitations in time, budget, or expertise. Creating data products can substantially boost revenue, and some companies focus solely on this.

The development of a data product usually starts with data ingestion, followed by mapping to ensure consistency, and applying standardized business logic for further processing. The end result might be a dashboard, a model, or an API endpoint.

This type of project is particularly rewarding due to its clear ROI linked to generated revenue. For more insights on data products, Eric Weber's newsletter is a great resource.

Where to Begin?

Choosing the right projects as a head of data analytics is crucial for success. It's easy to get sidetracked by questions about tools or programming languages like Python or R. However, executives often care more about results than the details of your methods.

While it's important to establish standards, if you're months into a project without delivering results, you risk frustrating business leaders. To avoid this, build your foundational data stack, address key queries, and continue to refine and improve your processes.

For further reading on data stacks and engineering, consider the following resources:

The Baseline Data Stack — The Different Types Of Data Stacks — Part 2
What To Consider When Building Data Pipelines
How to Become a Data Engineer in 2022