It has been two years since I made a career shift towards the analytics world (or big data, or business intelligence — you name it). I was initially an IT project associate. I had spent my days gathering system requirements, developing test cases, working with service providers to have the right tools to support a business. With just a little exposure to one reporting tool that covered data queries and visualization, I was stunned to realize how interesting this field is.
There is a bunch of data available out there, and with the right mixture, data can help people make faster, more impactful actions and decisions.
My first six months were full of adaptation. New field of work, new company, new culture of work (from a multinational corporation to a local start-up!), and the other new things that follow a career shift. On the negative side — I didn’t do a lot of research on this new analytical field. I experienced a little part of it, I liked it, then I went (that wasn’t a very thoughtful decision, was it?).
4 Things I Wish I Had Known Before Shifting My Career Towards Analytics
I don’t regret this decision, and in fact, I’m grateful that I did it! But looking back, there are still things that I wish had known and considered before making this shift. Especially for my old self as a non-mathematics/statistics/actuary background.
1. Analytics is a broad field, and different organizations name this role differently
As mentioned above, my data analysis exposure before my career shift was very limited. I spent 3 months creating reports and dashboards that helped stakeholders make various business decisions. The stakeholders were happy and I received an internal award for it. I looked at a Business Intelligence Analyst role in a start-up that I knew would allow me to do it, and there I went.
A few months in the role, I figured out that it is part of a bigger field called analytics and it is so much more than only creating reports. However, there is no industry-wide standard of competencies or certifications on analytics. The department doing this also has various names across organizations — from Business Intelligence, Data Analytics, Big Data, even Data Science.
In general, analytics covers the following :
- Data analysis — synthesize raw data and manipulate them to generate actionable insights. Some projects on this might include identifying factors affecting the discovery pain-points; analyzing marketing campaign performance; or customer buying behavior
- Simple data modeling — create approaches to group data points for a better-informed strategy. Some projects on this might include customer segmentation, defining factors affecting conversion, etc
- Data foundations — identify possible new data points and feature to enrich analysis and data models. Raw data (depending on the size and form) might be too much too handle and interim data foundations can help to bridge this
- Report and dashboard — this is not new, but there are some principles to make each report/dashboard more valuable and actionable within a glance
Make sure you know what kind of analytics you would like to do, and match it with the scope of work of the analytical organization that you’re exploring.
As an implication, your scope of work as an analyst will differ depending on how the organization defines analytics. But regardless of what the analytics department is like there, make sure you have enough support from the product/data engineering team to ensure you have good data foundations for analysis. Without a good data foundation, you will find yourself spending more time preparing the data sets, instead of actually manipulating, interrogating, or even presenting it. And I tell you — it is such an inefficient process.
2. Basic statistical knowledge and data manipulation capability is required
Having problem-solving skills is good, but it’s not enough. Coming from IS/IT background, I know about enterprises and processes but very limited statistics. Correlation, clustering, or regression was not something that I am familiar with. This condition limits my data analysis so that it tends to be very exploratory — with very little definitive or scientific insight. While in some cases that might not be a bad thing, in others this acts like a boomerang, as insights are again subjectively-driven and not scalable.
If you come from a non-statistical/mathematical background as I do, I would recommend to first get familiar with descriptive analytics and some predictive analytics concepts. Descriptive analytics includes summarizing and describing data sets, looking at their frequency, central tendency, dispersion, relationship with other data sets, etc. Predictive analytics are techniques used to use historical data to predict potential outcomes. On doing these analyses, especially on large data sets, certain data manipulation tools are needed. There are a lot of tools out there, but my personal favorites are Python (yes it comes from my IT background :P) and R.
3. Slicing and dicing data the right way — not necessarily the way we expect the result to be
Traditionally, data was used as a retrospective measure of certain business performance, and justification of the next decision. Us as an analyst, and the decision-makers usually already have some preemptive hypothesis on what insights are going to be generated from the data. And we like to be right about it.
Once the analysis has been started, we work on the data, perform several descriptive exploratories and then work on statistical testing. From there, we got some insights that might cause our hypothesis to be rejected or might even contradict our assumption. There are cases when we or the stakeholders become anxious and curious about this result, causing us to again manipulate the data in such a way that we need to retest and prove the initial assumption.
While rechecking, and further slicing/dicing data across groups is good to validate our analysis, we should also be in a neutral position and embrace the fact told by the data.
Especially when we are sure that we already followed the right analytical process and there is less poor-quality data.
This might cause us ended up in a circle of rechecking, re-slicing-and-dicing the data sets, spending hours but still not getting any additional insights or getting results that shifted from our initial objectives. On handling this, we can :
- Set up a clear set of objectives and boundaries for things to be explored in the analysis. Whenever you’re going for a deep dive, ask yourself how it will help you answer the problem statements or achieve the analysis objectives.
- Keep in mind that data is one key input among other variables. It is not perfect, there are several other factors to be noted on generating insights, like limited sample groups/data features, or changing trends.
Check out this interesting article from HBR for guidance on using data as the baseline of decision making in the organization.
4. Importance of being comfortable with rapid and continuous improvement
The work on the analysis or data modeling doesn’t stop when the insights are presented or the model is deployed. From each analysis, there will still be follow-up questions to solve, new data sets to be worked on, and additional questions from the outside. In a tech start-up with high growth, things change very fast. It’s important for you as an analyst to pivot your way of work for this.
Some tips…
- More chunks of tactical analysis instead of directly grand-foundational one
- Make it reproducible! You might find yourself rerunning the analysis in the next months for validation or similar context
- Be adaptive and understand how new product features or external affairs might change the results from your previous analysis (and thus making it obsolete or need additional tweaks)
You might find the things that you work on now might be obsolete in the next 12 or even 6 months. And that’s the beauty of it. What we need to ensure is basically whether we’re still producing analyses that are aligned with the organization’s focus.
Final thoughts
The above things might help you make a more informed decision about shifting your career towards analytics and the preparations/requirements that follow. I recommend chatting with people from the industry to get a better idea of what the world is like out there.
On top of everything, making a career shift will still require determination, adaptation, and dedication to the work. With all the buzz on big data, analytics, data science, and machine learning, the opportunity in this field is endless and still evolving.