What I do as a Data Analyst: no, it’s not only about analysing data
The reality of being a Data Analyst in a fast-paced business, juggling multiple reports, having your findings challenged, and dealing with the inevitable mistakes you will make.
Whether you’re considering a career in Data Analysis or you’re curious to see if other Data Analysts have a similar routine to yours — you’ll like this article.
Before we start, a bit of context…
I work as a CX Analyst, meaning I analyse customer experience data for a consulting company within the Luxury industry. Our clients are Luxury brands, and my main responsibility is to analyse the customer experience they are offering and suggest next steps to improve it. This analysis goes into reports presented by the account managers and are an integral part of what the company delivers to the client. Hence, they have to be insightful and of impeccable data quality.
The tools I use for analysing data and building reports are mainly Excel, Power Query, and Powerpoint. Part of my job is also to use Power BI to create automated reports.
Finally, I am currently fully remote and a digital nomad. I move countries every 3 months and mostly live in co-living spaces with other remote workers. My office is usually a coworking space, and my hours are flexible. I need to be somewhat available between 9–6pm CET, but I can start and finish work whenever I want as long as I meet my deadlines and am present for important meetings.
So, what do I actually do as a Data Analyst?
I manage my pipeline of projects and negotiate deadlines:
I’m part of a team of analysts, and we all have our pipelines with all the reports we need to deliver and their deadlines. Usually, we work on 1 report at a time, but we have peak months in which we sometimes have to create a report while another one is being reviewed by the account managers and a third is being proofread by another analyst. Things can get quite hectic!
This is one of the misconceptions I had before I became a Data Analyst. Usually, when we’re practicing and building our portfolio, we get used to having all the time in the world to focus on that analysis without interruptions. But, the reality might be very different — especially if you work for a private company. You’ll likely be working on many reports at the same time, and you’ll need to master time management in order to meet your deadlines.
Speaking of deadlines, managing the expectations of the people waiting for your report is also an essential skill you’ll need to master. I had never thought negotiation and communication skills would be so essential in data analysis, but trust me when I say it is! If you give your stakeholders an unrealistic deadline, you’ll end up working under a lot of pressure and delivering something that isn’t of good quality which will then disappoint them. At the same time, trying to bargain for more time in a fast-paced environment is sometimes challenging. So, drop that Google Certification and start acquiring some negotiation skills, my friend!
Well, yes, I do ACTUALLY analyse some data:
Even though there are other skills involved in a data analyst role, unless you’re part of a poorly structured data team who misses the balance between analysis and admin and drowns you in meetings, you should ideally spend most of your time analysing data. So, let’s talk about the fun part: how do I conduct my analysis and build the reports?
Firstly, I download the data from our internal servers and clean it up. Regardless of where you work, when data is generated and handled by multiple hands, you will always end up with something to clean or normalise in your data. Therefore, my next step is using Power Query to clean the data and create bespoke calculations or normalisations that I will need for my analysis. This can include eliminating discrepancies in wording, creating custom groups that will help me in my analysis (top and bottom performing stores or grouping customers by generation, for example), and normalising product categories with multiple confusing entries. Just to name a few.
Once I’m happy with the data, I create PowerPivot tables for my dataset on Excel. Even though nobody will see my dataset, I like to keep it organised and tidy for my own mental peace. Don’t get me wrong, I’ve seen brilliant reports created with only 1 pivot table. But my brain doesn’t work like that, so my dataset always has multiple pivot tables in several tabs, dividing the data into sections and angles. I actually work from a template I created, and the beauty of Power Query is that it enables me to easily update it with new data for every analysis.
Then, it’s time for the analysis itself. I always start by understanding the conversations we’ve been having with the client recently. I do this by making notes from previous reports we presented to them and thoroughly analysing the brief we receive from the account managers (I could write a whooooooole new article about briefing, it’s so important!). Then, I look at how they’re performing in the areas they’ve been focusing on.
Once I’ve reviewed the main KPIs, I try to find the reasons that led them to achieve that performance. This is time-consuming as you need to explore different angles. To take it to a practical example (and apologies if I have to keep it a bit vague, but I need to respect confidentiality): if a brand decreased the quality of its customer experience, I need to check if that decrease came from a specific region, from customers purchasing a particular product category, from customers of a certain generation, or from a group of stores, or which moment of the experience led to that decrease. Sometimes, the reason is very clear from the start, but it can also take curiosity and imagination to explore different aspects of the data until you find a well-rounded explanation for the performance.
Once you identify the issue and what led to it, it’s time to define what you’re going to recommend. For your action plan to be credible, you need to find reasons in the data that will show your audience why it’s worth following your advice. This can take many forms. I sometimes show that when Advisors take a specific action in-store, a high percentage of customers claim to feel a certain way. I also like to use customer verbatims as they can be more impactful than any number.
Once I have a very clear story that I want to tell and a nicely justified action plan, I choose the framework I want to tell my story in. I wrote a more thorough article on this, but this is essentially how you’ll want to package your analysis. Frameworks work in favour of your story and help it stand out and be easily understood.
The final step is to finally create the Powerpoint deck. Depending on how much time I have for the analysis, I usually start with a deck of blank slides with a text box saying what I want in each slide and even some visualisation ideas. This helps me ensure that every slide has a reason for being there and contribute to the story. In terms of visualisation, I use charts that the client is already used to, but I do try to bring something new in 1 slide or 2.
This is another reality of working in a fast-paced business: sometimes, you have to hold back on big ideas and deliver something simpler that will do the job. However, I do try to bring at least a little bit of innovation — a different way to display data, or a different angle, or even just nice little icons to make a slide less boring — to keep things entertaining for me and also improve our reports little by little.
After I deliver the report, it’s time to have my findings challenged and answer questions about the data — and that’s a good thing!
The beauty of working with data is that your work will cause reactions. Data raises questions, starts discussions, and can even unearth sensitive topics for the audience. The challenge for us, analysts, is to make sure that our findings are causing the reaction we intended and that we can answer questions about it to people with all levels of data literacy. This is the ability to humanise the data, and this can be especially hard for extremely technical analysts.
In a previous Data Analyst role for a manufacturer, I’d usually have the providers of our sales data come in and present findings about the market. Their analysis was always brilliant, extremely technical, and made use of intrinsic analysis techniques. But, I was often left frustrated because of their inability to humanise the data. They just could not answer my questions about how their findings related to our world and, at times, used technical jargon to explain more complex measures and visualisations. As a result, all their brilliant technical skills were wasted as they failed to cause the reaction they wanted on us and failed to answer our questions.
Being on the other side now, where I’m the provider of the data, I realise how important it is to be able to speak the client’s language and not take it personally when they challenge my findings. Sometimes, account managers or even clients will question a correlation I did or a visualisation and it leads to a very interesting discussion that ultimately brings the findings closer to the client’s world and improves my analysis skills.
On that topic, knowledge sharing between analysts is also hugely beneficial. An outside perspective can gift you with new analysis techniques for your tool kit and make you a better Data Analyst. In the company I work for, we are encouraged to brainstorm ideas with each other and share best practice reports. By doing that, they end up creating an environment where we don’t feel like we must be perfect all the time and like there’s room for experimenting and being innovative.
The importance of allowing room for error: we’re Data Analysts, not robots.
I have worked in companies in the past where no measures were put in place to account for the fact that analysts are not robots, and a small mistake is bound to happen on occasion. Instead, they made you feel like you were not a good enough analyst unless your reports were all perfect, even if you were juggling many of them at the same time.
Obviously, it’s fair enough to expect excellent data quality from your team of analysts. But, you will not get that from punishing the analysts or expecting them never to miss a data label or miscalculate something. What leads to good data quality is putting a process in place that allows mistakes to be caught before they reach the final recipient.
The company I work for ensures that every report is proofread by another analyst to ensure that all data points, charts, titles, and footnotes are accurate and make sense. This doesn’t make me feel like they are questioning my abilities. On the contrary, I feel more relaxed in knowing that if I let something small slip, it will be caught by someone else before it reaches the client. And thanks to our great teamwork, an excellent, error-free report is delivered to the client, and everyone is happy.
Whether you’re an aspiring or a current data analyst, be mindful that it is not normal, healthy, or productive for a company to expect you to be perfect and deliver error-free analysis 100% of the time. Don’t ever question your data analysis skills and all the time you spent learning the craft because your manager gave you a hard time for a wrong data label or a missing footnote. If you can, look for companies that don’t have this mindset or try to bring change to your team’s mindset.
Final thoughts:
When I started writing this article, I was literally listing tasks I had to do in a given workweek. But I realised that perhaps it would be more thought-provoking to bring discussions to the table. This is my attempt to contribute to forming more mindful new analysts and to offer a new perspective to experienced analysts.
If it did provoke your thoughts, please share them with me! And if you’d still like a more play-by-play article about what I do in my job, let me know in the comments too!
