Surely with the data? When we've got so much of it. Together with all of these new visualization tools and algorithms at our fingertips. Plus there's generative AI.
But that is exactly the problem. Where to start then? Given that there is so much data, and so many ways available to us for slicing and dicing the information in a myriad different ways.
How do we not get lost?
I find it best to be guided by the world outside the data. What is happening in the environment? What are stakeholders asking for? What are their biggest challenges?
Start there, also see what data views are already available and what work has been done. What insights have already been gained? What has been covered that can just be refreshed? Where are the gaps in that?
Then only, do you dive into the data in a very targeted and focused way. Knowing where to look, even kind of having an idea of what you're looking for. With an open mind on what you'll find.
Start with the real world
Although data might come across as quite abstract, it is still supposed to be a representation of the real world. Interactions between people and systems is where the data comes from. It is there that you ultimately want to make a practical impact with your data analysis work.
Thus, when looking to tackle a data analytics project, it is invaluable to gain insight into the context of the people and the systems generating the data. When was the system developed? What is it for? Have there been any recent changes? What was the thinking behind the changes? How are people experiencing the system?
When operating within a business environment, there will be key commercial aspects to keep in mind, if you want to garner support for your analysis proposal. What is the business strategy? What performance targets have been set? Are those being met?
Sometimes, as organizations, there is the risk of being so busy with our own thing, that we lose touch with the broader market, what competitors are doing and how the industry is evolving. Now you don't want to just be chasing the tails of others, but there is definite value in staying abreast of relevant news, engaging in conferences and building out your networks. Not only for personal career purposes, but also for sharing and learning.
Build on what has been done
Data is not that new any more. Most companies have been collecting and working with data for multiple decades. Doing analysis, building reports and taking decisions based on what they see. The specific problem you are investigating, is most probably also not entirely new.
It is thus worthwhile investigating and asking around, to get a view of what has already been found in relation to the data analysis project you are embarking on. People who have spent time within the environment gather their own personal data points through experience. Having a sit down with them can thus be a rich source of context and understanding. Any existing reporting on the matter can potentially give you a nice picture of what is currently happening, with the bonus value of telling a story over time, where the report has been kept in place and running.
Others might very well also have done similar or related analysis into the area, which could give you a wonderful foundation from which to build or challenge with your own findings. It enables you to speed up your delivery significantly, not having to start from scratch.
You could actually even curate a first draft from what has already been done to share with stakeholders, giving them a nice summary of what we've come to know up until now, before we add further lenses. This then also enables them to ask further questions and refine their ask off the back of the foundational set of learnings you have compiled for them. A holistic story like this, taking multiple perspectives into consideration, is generally quite a rare treat.
Mine where the gold is at
Once you've spent the time engaging with the real world and looking at what has already been done, you'll have a much better idea of the remaining gaps you can potentially fill with your analysis. This allows you to focus on digging into the most relevant datasets, working towards specific views that will most likely be able to answer the remaining questions.
You can put together a clear project plan with what to tackle, also giving your stakeholders a better view of expected timelines, which is generally quite a challenge with exploratory analysis.
Try to break up the analysis into bite size chunks, that ideally takes 1 or 2 weeks at most to complete. This allows you to show ongoing progress, whilst also giving you something tangible to share for further feedback as you work through the analysis methodically.
Be ruthless in killing those avenues that just don't seem to be going anywhere. Focus on what is most applicable to the specific ask, and where you can gain the furthest initial traction. Nice to have explorations can be noted for when there is spare capacity or more time.
Where to start with data analysis?
Given how much we've got to play with in terms of tools and data, we can easily get lost in the leaves. It is thus critical to refine your approach and have a plan of action when tackling a piece of analytics.
In starting off with your work:
- Make sure to engage the real world, where the impact you hope to ultimately have actually lives.
- Do not be shy to build on what has already been done. When you give credit and allow others to input, they are generally quite open to sharing.
- Rather than digging up all over the place, making a mess, be targeted and informed towards spending your time where you are most likely to strike the gold called for.
There are generally many uncertainties when embarking on data analytics. By having a plan of action it is however possible to enhance your chances of success.