Navigating tech interviews isn’t easy. You need to be across your chosen tech stack, be up to date with current trends and come across as the perfect fit for a team you haven’t met.
This post hopes to help those new to tech interviews interpret what the interviewer is asking with ten questions that may be asked in a data analyst interview.
Business and Career
What unique skills do you think can you add to our team?
Good:
You may hear ‘skills’ and think that you need to start reciting how you know your JOINs and can Index a table like a pro. This is an ok answer but doesn’t answer why you are unique.
Better:
The interviewer is asking what makes you ‘unique’ and why they should consider you over other candidates. Go back to the job description and look for what problem this new hire will be solving.
- Are you going to be part of a client-focused team that will mean you need to have strong stakeholder management skills?
- Will you be working on projects?
- Are they in an industry where you have worked before?
Highlight how you have read the job description and understood the problem they are trying to solve with this new hire. You are a problem solver after all.
What do you do to keep your technical skills up to date?
Good:
Let the interviewer know what you’re reading right now, what your favourite Tech Twitter accounts are and any tutorials or projects you’ve been working on.
Better:
Expand on this by talking about the interesting things that have been happening in the industry right now.
- Cloud computing and moving services to the Cloud.
- Security, Data Governance and privacy breaches.
- How a new framework or tool compares to the existing options.
Relate these current trends in technology and tools to what the company is working on.
- Has their move to the Cloud been in the news recently?
- Will they have to take more precautions with GDPR legislation given the business they are in?
- Ask if they are considering moving to a new framework or taking on a new tool?
Describe an example where you played an active role in solving a business problem through an innovative approach
Good:
Describe the project:
- What you worked on,
- The challenges you faced,
- The approach you came up with.
Better:
Focus on your specific contribution, rather than that of the team. Show how you were:
- Diplomatic and attentive with stakeholders and team members
- Wrote documentation and were able to back up your approach.
Talk about customer or manager feedback and how you’re interested in continually improving. Whether that’s through your own innovation or by taking on suggestions from your team.
Understanding Requirements
Provide an example where you had a customer extend the scope of work after the scoping had been completed and signed off?
Good:
Describe an example of how you managed to work under pressure successfully. The interviewer is asking this question to gauge your ability to handle it.
Better:
Be specific and realistic about how you deal with customers. The interviewer wants to see that you show empathy and assertiveness.
- You can anticipate change and push back on timelines
- Understand that there is a tradeoff between quality and timeliness.
Data Cleansing and Analysis
Is more data always better? Do you prefer raw or enriched?
Good:
Describing the trade-offs between quality and quantity is a good start to show your understanding in this area:
- Enriched, cleansed data is easy to work with depending on your tools of choice.
- More data can produce more complete data models if outliers are accounted for.
- Cleansed data takes longer to get into a normalised and enriched state – deduping, joining to other datasets, adding to a relational model.
- Raw, granular data is expensive to store and move around.
Better:
It really depends on the project you are working as to whether you go with quality or quantity. However, the tradeoffs can always be tackled with tools:
- AWS Kinesis, and similar tools can send data in streams to avoid batch loading and the maintenance that comes with that.
- Data Lakes can be built with S3, Glue and Athena to keep costs down.
- Machine Learning models can be deployed to do initial data cleansing and perform data quality monitoring when building new data sets.
What steps do you go through when processing and cleansing data in a typical project?
Good:
Describe the different steps of a typical data analyst process:
- Exploration,
- Preparation,
- Modelling,
- Validation,
- Visualisation
Focus on why data cleansing is important:
- To find any anomalies and outliers.
- Remove duplication and incorrect data.
- Makes the data set easier to work with.
Better:
Take this one step further by talking about the best practices for data cleansing:
- Taking an iterative approach and cleansing in logical chunks.
- Developing a plan to identify where errors are occurring and to identify the root cause.
- Verifying data is correct before it is signed off and allowed to flow into a model.
- Script out as much as possible so the process can be repeated or rolled back when required.
Which tools are you familiar with? What’s your preference?
Good:
This question isn’t just about which tools you use, it’s also an opportunity to talk about your experience with each tool. Analysts should be familiar with Excel, SQL, a visualisation tool and if required a statistical analysis tool or scripting language.
Show the interviewer that you are familiar with a suite of tools, even better if they are the prefered tools for the role you are interviewing for. But also make sure the examples you use in interviews show the kind of tasks you use them for.
- Excel – projects to aggregate data using Pivot Tables, and visualise the results using conditional formatting and graphs.
- SQL – projects to JOIN multiple datasets together and schedule them to run with a stored procedure.
- Visualisation tool – projects to track the progress of sales over time using multiple graphs. Thought needs to be put into the colour, graph type and what the end-user is trying to get out of it.
Better:
Take this one step further by talking about any up and coming tools you have read about or tried using. Big data tools like Hadoop and Spark, scripting languages like Python, libraries like D3 that make visualisations more interactive.
Describe an example of a complex analysis that you ran that you are particularly proud of, your approach and the insights gained
Good:
The interviewer is looking for examples where you are not only proud of the work but can enthusiastically describe what you did and what the result was. This is hopefully the kind of work that you enjoy and may enjoy in the new role.
Make sure your example is:
- Relevant to the role
- Something you are genuinely proud to have worked on
- Not sheer luck or where you only contributed a small part of a team project
- True! Don’t embellish a story.
Better:
Talk about how this project helped push you forward:
- Could you make changes to a process that then saved time or money?
- Did your analysis feed into future work?
- Did this success help you discover what you enjoy and what you are good at?
Data Visualisation
What tools have you used to publish data to end users?
Good:
This is an opportunity to show your understanding of the range of options for data visualisation and when they are appropriate.
- Excel – if you have worked in a startup, small organisation or prefer Excel as a one-stop-shop. There is nothing wrong with using Excel for smaller datasets that don’t contain sensitive information.
- Enterprise Tools – if you have worked in a bigger organisation you may have used Tableau, PowerBI or MicroStrategy. These are more expensive with associated licensing and training costs but provide a secure way to connect from database to visualisation layer.
- Statistical Tools – if you have worked in academia or scientific fields you may have used SAS, R, Jupyter notebooks or SPSS to present data. These are much more specialised tools but are relevant for roles in these fields.
- Web-based Tools – frameworks and libraries like D3 and HighCharts are increasingly common for infographics and web-based data visualisations.
Better:
Show that you know when to use one over the other and the drawbacks given each scenario:
Excel
- Great for quick analysis that is accessible and user-friendly.
- Isn’t a secure way to share sensitive data and multiple copies may end up on individuals machines.
Enterprise Tools
- Provide a secure, scalable way to connect the database to the visualisation tool.
- Expensive licensing arrangements and can be time-consuming to set up and train users.
Statistical Tools
- Specialised tools that allow the code for aggregation and visualisation to be run in the same place.
- Conflicting libraries and package versions make it hard to share, less technical users might find it difficult to get started.
Web-based Tools
- Generally beautiful to look at and interactive.
- Requires a different set of skills to set up and maintain, not always an appropriate way to publish sensitive data.
What form of supporting user help would you include?
Good:
The interviewer is looking for your preference in delivering support once you have completed a project. Do you prefer:
- One-on-one support.
- Presenting to a big group.
- Writing or blogging findings and ongoing insights.
Better:
Show you can evaluate what is more appropriate for the user and what ongoing help they might need.
- Would you present your findings in a different way when dealing with senior managers?
- Would you consider the role of your end-user? A fellow analyst may have different questions than a colleague in sales or marketing?
- What would you do if your audience looked bored in your presentation?
As hard as it is to believe but interviewers are not expecting perfectly scripted answers. The interviewer knows you are human and is looking for not only technical knowledge but cultural fit and how you communicate and think on your feet.
Good luck with your interview preparation!
Photo by Drigo Diniz from Pexels
