3 Key Types Of Data Projects You Will Need To Take On As A Head Of Data Analytics
Data and analytics teams are often responsible for several key pillars in a company.
This can pose a challenge when you’re the head of data and analytics and you need to pick your next data project.
Where do you even start when you’re constantly bombarded with requests from your management team.
In this article I want to discuss how you can start selecting your key projects to work on as a head of data analytics.
First, I started by breaking down your projects into 3 key types of projects.
Analytics, automation and data products.
Below we will discuss what each of these projects are and some considerations you should have when taking these projects on.
Analytics Projects
Analytics projects on their own can often be a great quick win. Starting with a few key core questions that you’ve gathered from the business and ending with some form of insights.
But how do you pick a solid data analytics project?
And what do analytics projects look like?
What Do Analytics Projects Look Like?
Let’s answer that last question first. Data analytics projects can look like an ad-hoc data report, a cleaned-up Jupyter Notebook, a dashboard or (shudders) a Powerpoint.
Each of these mediums can be utilized to tell a narrative to your stakeholders about what is happening in their business.
Generally speaking you will likely always have some form of ad-hoc analysis phase as you work to figure out if there is value in digging into the data further. Once you have finished your initial analysis of the data, then you will need to package your analysis.
But how you package your analysis will be up to you as well as your stakeholders in the end. The key point here is creating a narrative with data. Meaning, based off your ad-hoc analysis, you should have 1–2 key take aways you want your management to glean from your analysis and build around them.
Have a concise message and make sure your data, charts, graphs, and other pictograms support it.
How To Pick Your Analytics Projects
How you present your project is important.
But you will also need to make sure you’re working on projects that drive clear business value.
To identify the highest-value projects for your company, you will need to meet with your stakeholders to figure out what questions are most pressing at the moment as well as what would answering them impact in the business.
In order to figure this out you will need to talk with directors and business owners that manage the key departments in your company.
What questions should you ask?
Well for that, I will take a few questions provided by Ethan Aaron, CEO of Portable and Ex-Head Of BI from his article The 10 Steps To Building A Great Data Team.
These were:
- What Key Performance Indicators (KPIs) do you use to run your business?
- Where do you find them?
- What metrics do you wish you had at your fingertips every morning?
- What is the action you take based on each?
- How do you measure the impact on the business? Is it critical?
Another important point Ethan had when it came to figuring out which analytics project to take on was that:
The goal of analytics is not to present data for the sake of data. The goal is to inform actions that have a material impact on the business. — Ethan Aaron
Once you have a strong understanding of what questions your business leaders have you can start to create a list of high-value projects. From there it will be about making sure you align your projects with your managements goals.
Automation Projects
Automation has several key outcomes.
- Increase the scalability of a a process,
- Reduce its costs
- Reduce the amount of human caused error
The problem with automation is much of the work happens behind the scenes so to some degree, management doesn’t really care.
Unless the specific work is directly impacting their bottom line in a huge way, the C-suite probably won’t be as enamored with the output.
However, in order for all data driven businesses to run, they will need solid and robust automated systems.
Automated data pipelines.
Automated data QA.
Automated model deployment.
and so on.
Finding the right projects to automate can be difficult. The truth is, yes, automation can save time. Of course, creating automated systems also creates tech debt. Even when using low-code solutions.
So what should be automated.
What To Automate?
If you’ve been in the tech industry for a while, then you have probably seen the image below.
It’s far from gospel but I think it covers the concept of what processes you should automate clearly.
You need to look for projects that are either currently taking a lot of time or will, by nature of scale, take a lot of time in the near coming future.
In addition, if you don’t need the task done frequently, then automation, again becomes more costly than its worth.
Automating projects can be a lot of fun for engineers because it’s cool to watch processes flow smoothly. The problem is automation equals code and code equals another code base to maintain.
So what you automate needs to make sense from a time trade off perspective.
In terms of finding automated projects. Don’t worry, the good ones will find you(as will the bad ones).
Although the C-suite might not care about automation projects, analysts, project managers, data scientists and everyone else in between all have created processes that can be automated.
The question is, should they.
Data Product Projects
Once your team has mastered analytics and automation they can start to consider the idea of building a data product. You can’t jump straight into data products before at least having a working knowledge of your teams analysis process and having a baseline automated data infrastructure because what will you be building the data product off of?
Your data products can often take the form of a processed data set that automatically categorizes the data inputs using natural language processing.
Another data product might be a dashboard and insights.
Still another might be an API end-point used to manage your companies dynamic pricing.
All of these data products could be built internally by your customers IT teams. However, often due to limitations of time, budget or expertise, many companies look externally for this type of data products.
So a data product can be a great way to increase revenue. In fact, some companies only create data products. I know. I have worked for them.
What these data products look like can be very simple. You first ingest data, you have some form of mapping to make sure regardless of who provides you the data, the data will always look the same by the end of the initial processing.
From there your team has likely developed some form of model or standardized business logic that further processes the data and then serves up the data in some form of dashboard, model, targeted list, API end-points, etc.
Personally this is my favorite type of project because it’s a lot of fun to develop a data product and there is a clear ROI attached the revenue created by the product.
If you really want to learn more about data products themselves, then I would say check out Eric Webers Newsletter.
Here is a quick excerpt.
“I think we’re about ready to share this set of dashboards. The team has spent a few months getting them ready and made sure everything works really well.”
[2 months go by]
“It doesn’t make sense. The dashboards and metrics seemingly covered everything product would need, in addition to meeting a bunch of finance requests. Why is no one talking about it or using it?”
[another month goes by]
“This other team built a hacked together view of the data and suddenly it is all that finance can talk about. It keeps showing up in conversations and it doesn’t do much of anything other than the finance metrics.”
Where To Start?
Picking the right projects to take on as a head of data analytics is crucial to making sure you succeed in your role.
It can be easy to get distracted by a lot of questions such as which tools should you use and whether you should use Python or R.
But in many cases your C-suite won’t care about how the sausage is made. Instead, they want results. Yes, do spend some time answering questions about standards. However if you’re multiple months in and still not delivering some form of output, then more than likely your business leaders will start to become frustrated very quickly.
To avoid this build your baseline data stack, answer key questions and continue to iterate and improve your process.
Read/Watch more about data stacks and engineering below.