Data Analyst

Data Analysts draw meaningful insights from complex datasets with the goal of making better decisions. Data Analysts work wherever an organisation has data - these days that could be in any function, such as product, sales, marketing, HR, operations and more. Data Analysts will normally be well skilled in querying relational databases using SQL, wrangling & cleaning datasets, conducting analysis and creating simple to read reports & visualisations for the end decision makers. They’ll commonly use a programming language like Python or R to conduct their analysis, and also BI tools like Excel, Tableau, PowerBI and Looker.

Data Analysts are increasingly common positions in modern companies. The digitization of so many industries has seen an explosion in the supply of data generated, creating an enormous opportunity to make better, data-informed decisions. With it, companies then have had to hire people to make sense of the data.

Data Analysts are one of the core roles in any data team. The scope of the role depends quite a bit on the size & industry of the company, and their relative data maturity.

What are the responsibilities & duties of a Data Analyst

The duties and responsibilities of a Data Analyst do vary from role to role, industry to industry, company to company. It also depends on how senior or junior the role is. That said, these are the typical responsibilities of a Data Analyst:

• Collate, interpret and cleanse new datasets for new analyses & reports
• Creating and maintaining insightful and simple to interpret reports, dashboards and other visualizations
• Monitoring KPIs and proactively investigating any unusual trends
• Encouraging a data-driven culture with the organisation
• Creating and updating data models used in reports and dashboards
• Creating and maintaining statistical models
• Maintaining documentation for various data systems
• Various ad hoc analytical projects, for example creating forecast or classification models
• Ad hoc analysis into trends and ‘deep-dive’ analyses

Ultimately, the end goal is normally to generate ‘actionable insights’ - this is sometimes called the ‘so-what’ of the analysis. The analysis is considered valuable to the extent that it’s insightful (i.e. telling the audience something that they didn’t know) and ‘actionable’ (can the audience actually do anything with this insight, or is it just more of an academic curiosity).

What are the required skills & experiences of a Data Analyst?

Just like the responsibilities and duties, the required skills & experiences do vary from role to role. Here’s a typical set of requirements for a Data Analyst:

• Strong experience in using SQL to write complex queries
• Experience using Python or R to conduct data analysis
• Experience creating & maintaining dashboards and reports in a BI tool, such as Tableau or PowerBI
• Excellent written & verbal communication skills
• Ability to work independently and also to collaborate with other teams
• Experience communicating insights from complex & technical concepts to non-technical/non-data stakeholders
• Strong business acumen
• Experience in defining new metrics
• Strong all round analytical ability including a high attention to detail, scepticism and common sense

What are some other titles Data Analysts may also be called?

Titling in analytics is sometimes a little blurry. What one company calls a data scientist, another may call a data analyst, for example. That said, there has been a bit more consistency in titles over the last few years as the data industry has matured.

Depending on the organization, a Data Analyst may also be referred to as an Insights Analyst or a Data & Reporting Analyst.

A title is just a title, and for candidates, we’d recommend reading the job description and asking the hiring manager for details of what you’ll actually be doing to understand the role.

For organizations, we’d recommend aligning your job titles with what the market generally uses. E.g. over-selling a basic reporting role as a Data Analyst will lead to disappointment among candidates, and cause you to attract the wrong types of candidates.

What's a typical day in the life of a data analyst?

A day-in-the-life of a Data Analyst of course would vary a lot from team to team and organization to organization. Each organization has different datasets, work at different cadences with different methodologies, and have different technology set-ups.

But as a general guide, here’s what you might expect in a typical day:

• Start your day by checking existing dashboards for key metrics that you monitor.
• You notice the data is missing completely. After checking a shared Slack channel, you see a message from your data platform team - the ELT process failed, so the dashboards won’t be updated for another 20 minutes. Coffee time!
• Once the dashboards are up-to-date, you notice a negative trend in a core market - that’s something you’ll want to dig into further.
• Checking your email, you see one ‘urgent’ request for access to an existing report. You ping your colleague a Slack message who controls access to dashboards and ask them to sort it out.
• Time for the daily team meeting, where you run through what you did yesterday, what you’re doing today and what you’ll do tomorrow.
• You start digging into the negative trend, just as the market manager messages you ‘What’s going on here do you think? Please investigate it.’ You press on, using various existing dashboards and writing some specific SQL to dig deeper.
• After lunch you seem to have found one issue. Looks like performance took a hit on Android devices in the Korean market. Maybe an app update went wrong? You ask the product team if they released any big changes yesterday, but they also seem confused.
• After a bit of back and forth, the product team seems to have discovered a bug has been introduced that’s caused the sharp decline. Well spotted!
• You update a few people on what you’ve found and what’s going on.
• With ad hoc analysis completed, you’ve got just enough time to return to a modelling project - you’ve been working on a new way to forecast revenue more accurately. Time for some Python!

What skills does a Data Analyst need?

The scope of a Data Analyst role does vary a little bit by the size of the company, the industry, and the data maturity of the organization. This then dictates the skills that are (or are not) needed to be a Data Analyst.

While generally companies don’t expect tool-specific experience, the caveat to this is contracting roles, that might only last for 3-12 months. For these roles, the expectation is that you can hit the ground running, and so the company would expect you to already have familiarity with their stack, which is fair enough.

For lower data maturity organizations, you should expect to see less well-established data ecosystems. This means most likely an incomprehensive data warehouse, probably with limited access tied up with an ‘IT team’, and a lot of analytics being done in Microsoft Excel.

That said, for a typical Data Analyst role, these skills would be considered mandatory:

SQL

SQL is typically considered a must-have for a Data Analyst. Note, there are various subtle differences in SQL syntax, depending on the relational database management system (RDMS) that an organization is using (e.g. SQL Server, MySQL, etc.). Normally companies don’t require experience in any specific one of these, just that you have experience in any of them, as they’re all very similar. The differences in SQL syntax is comparable to the difference between, say, American English and Australian English.

A visualisation tool

Data visualization skills are considered a must-have. Most commonly, this will be some experience in putting together dashboards and reports in something like Tableau, PowerBI, Looker etc. Companies are looking not just for experience in the mechanics of creating & maintaining visualizations, but also the general ability to communicate graphically. This is because the goal of analytics is to influence decisions, with visualization being a critical output of the analysis.

Python or R

Python and R are the most commonly used programming languages as a Data Analyst.

Companies are generally agnostic as to which one a candidate would know well, as long as they know one of them. If you are a candidate and have to choose which to learn from scratch, it would probably be best to go for Python. It’s a general-purpose programming language so you can use it to do pretty much anything you like, while R is really made specifically for statistics and data science. The Python vs R debate is not something we’ll get into, but for what it’s worth, around 80% of candidates when taking tests on Alooba choose Python over R, when given the choice. The popularity has increased consistently over the last 5 years.

Analytical skills

Surprise, surprise - a Data Analyst needs to have analytical skills. Included in here is a very keen attention to detail for all types of data. Your stakeholders will be relying on you to scrutinize the data with extreme prejudice and a healthy dose of skepticism. You ultimately have to arrive at the right answer through correct analysis and good assumptions and deal with the ambiguity involved in real-life analysis.

Common sense is also a big part of this - for example, if you check your dashboard in the morning and it shows that there have been no sales at all in the last 12 hours, before you go blasting the ‘All Staff’ mailing list telling everyone the company is dying, you’ll probably want to check that the data warehouse ETL process completed successfully (hint: the data probably just is not up-to-date).

The line between general analytical skills and what’s termed ‘data literacy’ is perhaps a little blurred. Data literacy would include being able to know when to use a median vs a mean, not to take an average of an average (please, just don’t), and why putting time on the X-axis of a line chart probably isn’t a great idea.

Communication skills

The goal of any data analysis is to influence a better decision to be made. As a Data Analyst, you could create the most amazing analysis of all time, but it will all be wasted if you aren’t able to easily describe the key takeaways from the analysis. You will typically need to communicate these to people with less technical skills than yourself, who are less familiar with the data than you are. They’ll often be more senior, time-poor, and might not want to get into the nitty-gritty details. Data Analysts need to be able to explain the ‘so-what’ of their analysis quickly & easily. This could be through visualizations, a written email, a presentation, or simply an informal chat.

Business Acumen

Data Analysts are typically working in - or very close to - ‘the business’. Unlike Data Engineer roles that can be a little more abstract and disconnected, Data Analysts are right in the thick of it. There’s definitely an expectation that they’d be able to understand the key drivers of business success, understand what metrics should be tracked and why, understand underlying trends, interpret bigger-picture macro trends, etc. Maybe the best way to describe this skill is that they need to be able to think and act like an owner.

What are nice-to-have skills for a Data Analyst?

In addition to the above need-to-have skills for a Data Analyst, there’s also a lot of nice-to-have skills. Generally speaking, the smaller the organization you operate in, the wider the scope will be of your role, and so the more skills you’d be expected to have (or pick up).

Data modeling

Data modeling is basically the work of designing & maintaining data marts, datasets, views, etc. While data modeling may ultimately be reduced to some SQL, the modeling itself is distinct from this.

For larger organizations, data analysts are unlikely to be involved in building or maintaining datasets in the data warehouse. This would more commonly fall on a data engineer, business intelligence developer, or SQL developer. Nonetheless, the skillset is useful, especially as you will want to communicate with the person making a change to a data model and understand how it works.

E.g. Imagine you have a Tableau dashboard, and you’d like to create a new visualization that involves a new column not currently in your dataset. You’ll need to add this from the source, and understanding how the dataset is built will help you understand how this needs to be done.

Machine learning

Data analyst roles will often have machine learning - or in any case some kind of more advanced statistics - as a nice to have. If there is an expectation, it would normally be entry-level machine learning, such as regression models (linear or logistical), and some entry-level classification work. The expectation would be that you know how to implement, build & interpret basic models using some relevant package in R or Python. For the most part, though, the machine learning work would be done by data scientists, not data analysts.

Which tools and technologies do Data Analysts use most?

Data Analysts will typically use these technologies on a daily basis:

• Data warehouse or lake: Professionally organized datasets would normally be available within a data warehouse or data lake for a Data Analyst to then conduct analysis on. If it’s a data warehouse, they’ll normally query this database using the language SQL.
• Visualization tool: A visualization tool allows Data Analysts to create compelling visualizations to tell an easy-to-interpret story about their analysis.
• Programming language: Languages like Python & R are commonly used by Data Analysts to conduct their analysis.
• Documentation (email, Slack, wiki, PowerPoint): Ultimately Data Analysts will need to share the final output of their analysis somehow, either via email, Slack, some kind of internal wiki, or a presentation tool.

There’s some crossover in each of these technologies in the analytics process. For example, a Data Analyst might do all their data wrangling in SQL, and then pipe the data into a Python environment for their analysis and visualization.