The Four Components the Interview Tests
1. Opening the case
Clarifying Questions
That's one of the main differences from consulting prep.
Revolut's prompts are usually very short and need additional data for the assignment to be fully comprehended. See this example below where I compare real prompts from McKinsey and Revolut.
Considering this, before touching the problem, a strong candidate asks questions to better define the problem.
Clarifying questions serve a specific purpose: they establish what the problem actually is, what success looks like, and where the analysis should focus. Without them, you risk building a beautiful structure around the wrong problem.
There are four dimensions worth covering in every case:
- Context → what is the broader situation? What does the business do, and what has changed recently that makes this problem worth solving now?
- Efficiency drivers → what are the key metrics that define the problem? How have they moved, over what period, and across which segments?
- Objective and success definition → what does a good outcome look like? Is there a specific target, timeline, or metric the solution needs to hit? what's the baseline?
- Constraints → what limitations exist that will shape the solution space? Budget, timeline, regulation, internal policies, or anything the interviewer has flagged as off the table?
The data you collect here is not just for orientation. It directly shapes your structure, telling you which branches to go deep on and which to deprioritise before you even start building and also guide you later in the case, during the brainstorm session.
2. Structuring
Structure is what turns an ambiguous question into a solvable problem. The goal is to build an issue tree: a breakdown of the problem into mutually exclusive, collectively exhaustive (MECE) components, each of which can be confirmed or ruled out with data.
Think of it like a doctor diagnosing a patient. When someone presents with unexplained weight gain, a good doctor doesn't guess, they break the problem down logically. See one example of how to breakdown this problem below:
Each branch is independent. Together they cover all possible explanations. The doctor then prioritises based on what's most likely given the patient's profile and consultation (the clarifying questions), goes deep on that branch first, and only expands if needed.
After defining how you would break down the structure, you will fill it with hypotheses of what may contribute to the problem. These hypotheses must be testable, which means you need to ask for data to either confirm them or rule them out. See below one example of hypotheses and their corresponding data requests.
There are three types of issue trees you'll need:
Logic-based: Used when the problem requires causal reasoning. Why is something happening? Each branch represents a distinct logical possibility. Example: churn is increasing either because users are choosing to leave, or because they're being forced out.
Quantitative (equation-based): Used when the metric can be expressed as a meaningful equation. Profit = Revenue − Costs. If profit has declined, the cause must be in one of those two variables, and the equation tells you exactly where to look and prioritize after deep diving the first layer.
Flow-based (process or funnel): Used when the problem sits within a sequential process. Usually found in ops or product problems. If conversion of a particular operational process is low, for example, the problem may be concentrated at specific stages. The data will tell you which one.
See that prioritization is required for you to have an efficient structure that will allow you to solve the case on time.
A common mistake is building a perfectly symmetrical tree, equal depth on every branch, regardless of what the data says. That signals a lack of judgement. A strong candidate deliberately prioritises branches that the clarifying questions have already signed as the ones that may be hiding the problem, and explains why.
3. Interview Math
Math in the PS interview is not about mental arithmetic, you'll be allowed to use a calculator and in some cases, even excel (make sure to ask your interviewer about it). It's about knowing when a calculation is needed, structuring it correctly, and using the result to make a decision.
Math problems typically appear after a root cause or potential solution has been identified. The question being answered is almost always one of: how significant is this? or is this worth solving?
A repeatable 5-step approach:
- Clarify what needs to be calculated → define the exact question before touching any numbers
- Define the calculation logic → explain how you'll structure the result (equation, funnel, table) before using data
- Ask for the required data → request specific numbers one at a time, in the order your calculation requires
- Organise and calculate → restate the data to confirm accuracy, then work through it step by step. You can narrate as you go or keep in silence, whatever makes you feel more comfortable (I personally suggest doing it in silence, so that if you make any minor mistake, you can fix it before voicing it over).
- Interpret the result → explain what the number means for the goal, and whether it's significant enough to act on.
The fifth step is where most candidates lose points. A number without an interpretation is not an insight. Always connect the result back to the case objective: does this close the gap? Is it large enough to prioritise?
4. Generating Solutions
Once a root cause is confirmed and sized, the case moves to solutions. Most candidates revert to free brainstorming here, listing whatever comes to mind. This produces generic ideas that are hard to prioritise and often don't address the actual root cause.
A strong candidate structures their ideas before generating them, organizing them in common themes or full structures.
Regardless of the structure, two additional steps are essential:
Prioritisation and validation: Not all ideas are equal. Rank your top ones based on potential impact relative to the constraints established during clarifying questions. Always check with your interviewer if that initiative is valid and bring risks and potential constraints before going deep on any single idea, treating them as hypotheses.
Implementation: That's one step that Revolut really values, as they want to see if you're able to take an idea out of the paper and make it happen. Before closing the case, a strong candidate moves beyond the 'what to do' and addresses the 'how to do it', naming the team or function responsible, the concrete action plan, and a realistic timeline given the constraints already established. For example, if one of the solutions prioritised to improve conversion rate of a product is to place its widget on the front page, you may say: “[Considering we have a green light to move with this initiative] I would recommend running a two-week A/B test with a valid sample size, half seeing the widget on the front page, half in its current position. If CTR improves in a statistically significant way, we have sufficient evidence to roll out to the full base.”
KPIs: For each prioritised solution, define at least one primary KPI (does this achieve the goal and is sustainable in the long term?) This signals that you're thinking beyond the immediate fix.
What Separates Good from Great
After coaching dozens of candidates, these are by far the most common mistakes that prevent some to move forward the process:
- They don't prioritise. They go equally deep on every branch, regardless of what the data says. A structured candidate uses clarifying questions data to identify where the problem is concentrated before building, and deprioritises everything else explicitly.
- They stop at the symptom. They identify that "revenue per transaction declined" without asking why. The root cause is always one more question away, it is something specific that can have a targeted solution or set of solutions. Great candidates keep asking why until they reach something actionable.
- They calculate without interpreting. They produce a number and wait. The interviewer is not waiting for a number, they're waiting for a decision. What does this mean for the goal? Is it significant enough to act on? What's the next step?
- They don't think about execution. They stop at the list of ideas. Great candidates validate ideas and then suggest how they would implement and monitor them.
How to Prepare
The PS interview is highly learnable. The frameworks are not complex, but applying them consistently under pressure, with real data, in a candidate-led format, requires deliberate practice.
However, most candidates preparing for Revolut face the same challenge: very limited preparation time, little to no role-specific material, and uncertainty about what interviewers actually expect.
If you want a structured, efficient end-to-end preparation resource built specifically for Revolut's PS process covering clarifying questions, the most common types of issue trees, math problem types, and solution generation with worked exercises, find a full Revolut course here:
👉 Revolut Problem-Solving Interview — Complete Preparation Course
If you'd prefer personalised guidance, I also offer 1:1 coaching sessions tailored to your specific profile, target role, and current level.
👉 Book a 1:1 coaching session
Written by Mariana, former McKinsey consultant and Nubank Senior Product Ops Manager.
#1 Revolut PS interview coach on PrepLounge, with a Problem Solving pass rate 4X higher than the UK benchmark.
