Why You’re Failing Non-Technical Data Science Interviews and How to Fix It
Insider tips from a hiring manager who interviewed 100+ candidates
Today’s guest post is a deep-dive by
from . Tessa is a DS manager at LinkedIn, focusing on the Growth and Foundational AI areas. She also worked at McKinsey as a DS consultant and the autonomous driving startup Cruise as DS manager. She turns her unique perspective on how to do better work and grow your career into hands-on guides. You can also find her on LinkedIn.
When I was prepping for my data science interviews, I spent most of the time going through SQL, Python/R guides and memorizing different matplotlib functions.
But I barely spent any time thinking about the non-coding rounds.
Now that I’ve interviewed 100+ people as a hiring manager, I wish someone had told me that the technical rounds are merely what gets you a foot in the door.
Everyone at top companies is strong technically, so what really sets you apart are the non-technical rounds where your soft skills are put under a microscope.
Technical rounds are “guardrail metrics”. They weed out candidates who don’t meet the basic technical bar — that’s why they happen at the beginning of the interview process.
Non-technical rounds are the “True North”.
Technical skills get you the interview, soft skills land you the job.
However, in many interview guides, soft skills don’t get spot light they deserve, because they are more ambiguous, and as a result, harder to grasp.
But that doesn’t mean it’s impossible to teach.
In this article, we’ll dive into:
The top 3 soft skills data science hiring managers look for
The types of interview questions used to test these skills
How to confidently showcase your soft skills and stand out
Skill #1. Structured problem-solving and communication skills
Data scientists face abstract business problems day in and day out, like:
User engagement is down 10% today - can you figure out why?
Should we do what competitor X did and launch a Y feature?
When we launch Z product, how should we measure success?
It’s crucial for you to be able to scope the problem, solve the problem and communicate the results, all with a structure.
These skills are usually tested through case studies as well as certain questions in the behavioral portion of the interview.
How to ace case studies with a structure
I see a lot of candidates stumble here.
Not because they can’t get to the right answer (spoiler: there is none), but because they lack structure in their approach and end up rambling or going down an irrelevant rabbit hole.
I’ll show you how to avoid that.
The big categories of case study questions for DS interviews are:
Investigative questions like “X metric decreased by Y%, how would you investigate what happened?”
Measurement-focused questions like “How would you measure the success of XYZ initiative?"
Questions that require you to choose an appropriate type of analysis or model for a problem, like "The business wants to figure out which customers have the highest value; how would you go about that analysis?"
Remember, the case study is never meant to get to the “correct” answer. Some companies don’t even have an “answer key”, but rather a set of criteria, like:
Develop a structure for open-ended questions
Work through a problem from first principles, and
Communicate their reasoning and findings clearly
So here’s the key:
There are no right or wrong answers for case studies, there are only right or wrong approaches.
I wrote a deep dive on how to approach these types of questions, but here’re the high-level steps to tackle case studies:
1. Ask a lot of clarifying questions and actually listen (along the way)
Make sure you really understand the question and tailor your solution based on the interviewer’s hints and answers to your questions.
The #1 mistake you can make here is not paying attention to the interviewer’s hints and not adjusting their approach on the spot.
If you’re not comfortable thinking on the fly, you need to practice in mock interviews — it’s one of the most important interviewing skills. Nobody wants to hear rehearsed, generic answers and frameworks; you need to show that you’re “coachable”. The hints are supposed to help you, not trick you.
Remember: They want you to succeed (that’s why they invited you to the interview), so work with them, not against them.
2. Communicate with a structure
If there’s only one thing you remember, let it be this: Any structure is better than no structure.
If you have a structure and take your interviewers along with you, then they can ask questions and guide you in the right direction if they disagree with any step you take.
But they can’t read your mind if you don’t tell them what you’re thinking.
How do you show and communicate your structure?
Jot down the “inputs” you are given by the interviewer (any facts about the situation, constraints etc.)
Write down any assumptions you might be making
Sketch out the high-level outline of your structure. This doesn’t need to be complete; for example, if you’re asked to investigate why revenue dropped on Facebook Feed, you could start by breaking down:
Show your interviewer your high-level structure, and walk them through your approach & collaborate with them
Through this structured approach, you show that you are a structured thinker who can solve any problem, not just this specific one
Skill #2. Ability to introspect, learn and improve
Candidates often overlook this skill, but a lot of hiring managers mention this as one of the key skills that make them say “yes”.
Most people think you should use every minute of the interview to showcase your achievements and strengths. So when being asked questions like “What’s your weakness/area of improvement?” or “Tell me about a time when you failed / a project didn’t go as planned”, they think it’s a trap and try to come up with answers that are actually strengths disguised as “weaknesses”.
In fact, there are still tips online suggesting people should say things like:
“My biggest weakness is that I’m too much of a perfectionist…”
or “My biggest weakness is that I work too hard…”.
Please don’t do that.
The truth is, I personally have failed every single candidate who answered those questions like this, because I don’t believe (or at least they have not shown me) they have the ability to admit their failures and learn from them.
To learn from your failure, you must first admit you have failed.
To make it easier to remember what you should highlight, use the key word “SAFE”.
Summarize the weakness: Make it very clear upfront
Acknowledge the impact: Highlight why this is a weakness that’s worth mentioning and working on. Ideally it’s significant but not detrimental to the job requirements
Find examples of improvement: The ability to learn and improve is what every hiring manager wants to see. Make this the main part of your response. Highlight concrete examples.
Expand on future improvement plan: Provide a detailed plan for continuous improvement. Show that you are always looking to get better and never settle for “good enough”.
Skill #3. Ability to balance short-term and long-term solutions
This is a key skill you need in a fast-paced environment.
We constantly need to trade off between quality and speed. Even though everyone knows great things take time, often we need a good-enough solution to plug the gap while we are building the “great” solution.
A lot of analytics folks struggle with this: we’re trained to do things “right”, and any imperfection causes deep discomfort. This discomfort can cause big tension between the analytics org and business partners, so growing this skill will:
Make you stand out among your peers
Make your stakeholders very happy
Secretly, many behavioral questions are screening for this skill. Some examples:
Tell me about a time when you had to deliver a solution under time pressure. How did you handle the situation?
Have you ever had to deliver a less-than-perfect solution? What did you do?
Being able to balance short and long term solutions requires 3 things:
Ability to identify hacky short-term solutions
Ability to recognize the caveats of this short-term solution and improvements you can make in the future
Grit to iterate the hacky solution into a robust one
So you should highlight these 3 components. Here’s an example of using STAR:
Takeaways
Soft skills are harder to master because they’re not as clear-cut as technical skills. You can’t just grind through LeetCode problems to get better.
The best way to improve soft skills is through your everyday work—every project, meeting, and interaction is a chance to grow.
The second-best way? Practice before your interviews so you’re ready to showcase these skills when it counts.
Here’re the top 3 soft skills data science hiring managers look for:
Structured problem-solving and communication
Introspection and learning from mistakes
Balancing short-term wins with long-term strategy
 | A guest post by
|
Best read for the day ❤️
I love this. So many people lose out on opportunities because they don’t focus on soft/people skills.