revolut

Over the last 3 months, Revolut's paid-plan sign-ups increased by 20%, but monthly subscription revenue fell by 12%. We want to understand why and fix it.

1. Reframe

"So sign-ups are growing, but subscription revenue is falling. We need to identify what is driving the decline and improve revenue. Is that correct?"

Interviewer: Yes.

---

2. Clarify

"What is our target and timeline?"

Interviewer: Return revenue to its previous level within 6 months.

"Any constraints?"

Interviewer: Do not increase headline subscription prices.

"Have we already tried anything?"

Interviewer: No.

---

3. Structure

I would start with:

Revenue=Paid customers×Revenue per paid customer

So I want to investigate:

1. Number of active paid customers

2. Revenue per paid customer

---

4. Ask for data

"How have those two metrics changed?"

Interviewer:

• Active paid customers: -5%

• Revenue/customer: -7%

Quick check:

0.95×0.93=0.8835

So revenue falls about:

1−0.8835=11.7%≈12%

Both explain the problem.

---

5. Analyse A. Why are paid customers falling? Paid customers=Existing+New−Churned

"Since sign-ups increased, has churn increased?"

Interviewer: Yes.

Monthly churn:

3%→7%

"Which customers are driving that increase?"

Interviewer: 70% of the increase comes from customers acquired through a recent discount promotion.

"What happens to them?"

Interviewer: They receive 50% off for 2 months. Many downgrade to the free plan when full pricing begins.

Good. That explains much of the customer decline.

B. Why did revenue per customer fall?

"Has pricing changed, or has the mix of plans changed?"

Interviewer: Prices are unchanged. More customers are choosing the cheapest paid plan.

Share choosing cheapest plan:

40%→65%

So customer mix is lowering average revenue.

---

6. Root causes

I see two main causes:

Root cause 1

Discount promotion attracts customers who later downgrade.

Promotion→Low retention→Churn↑→Paid customers↓

Root cause 2

More customers choose the cheapest paid plan.

Cheaper plan mix↑→Revenue/customer↓

So the 12% revenue decline is explained by both churn and lower ARPU.

---

7. Solutions Solution 1: Improve the promotion

Instead of a large temporary discount, test offers that encourage longer-term usage.

Why?

Because promotional customers drive 70% of increased churn.

KPI:

3-month retention rate

---

Solution 2: Improve conversion before the discount ends

Show customers the paid features they actually use and remind them of their value before full pricing starts.

KPI:

Promo→full price conversion

---

Solution 3: Increase adoption of higher-value plans

Target customers with relevant higher-tier benefits rather than generic upselling.

KPI:

Revenue per paid customer

---

8. Prioritise

1. Fix the promotion

It causes most of the churn.

2. Improve promo-to-full-price conversion

Directly protects paid customers.

3. Improve plan mix

Then address the 7% ARPU decline.

Main trade-off:

Reducing the discount may reduce new sign-ups.

So I would track:

New signups

alongside:

Retention

and optimise for long-term revenue, not sign-ups alone.

---

9. Executive summary

"Our objective was to recover the 12% revenue decline within 6 months.

I found two main causes. Active paid customers fell 5% because churn rose from 3% to 7%, with promotional customers causing 70% of that increase. Revenue per customer also fell 7% because more customers moved toward the cheapest plan.

I would first redesign the promotion, then improve conversion when discounts expire, and finally improve adoption of higher-value plans.

I would track retention, promo-to-paid conversion and revenue per customer, with new sign-ups as a guardrail."

Revolut has seen a 20% increase in the time taken to onboard new business customers over the last 2 months. We want to understand why and reduce onboarding time.

1. Reframe

“So business customer onboarding is taking 20% longer than before. Our goal is to identify the causes and reduce onboarding time. Is that correct?”

Interviewer: Yes.

2. Clarify

“Do we have a target and timeline?”

Interviewer: Reduce average onboarding time from 5 days to 3 days within 4 months.

“Any constraints?”

Interviewer: We cannot weaken compliance checks.

3. Structure

I would break onboarding time into:

Total onboarding time=Customer waiting time+Revolut processing time

Then within Revolut:

1. automated checks

2. manual reviews

3. rework / repeated checks

4. Ask for data

“Which part has increased the most?”

Interviewer: Manual review time.

“How much?”

Interviewer: From 1.5 days to 3 days.

So this is likely the main driver.

5. Analyse deeper

“Why are more cases going to manual review?”

Interviewer: A new compliance rule sends more applications for manual checks.

“What percentage went to manual review before and now?”

Interviewer:

Before:

20%

Now:

45%

“Are all these additional reviews finding real issues?”

Interviewer: No. Around 70% of the extra reviewed cases are approved without changes.

So many reviews appear unnecessary.

6. Root causes

Main causes:

1. Manual review rate increased from 20% to 45%

2. Many additional reviews are low-value false positives

3. This doubled manual review time from 1.5 to 3 days

So:

More false positives→More manual reviews→Longer onboarding 7. Solutions

Solution 1: Improve the compliance rule

Reduce false positives while keeping risky cases in review.

KPI:

Manual review rate

and

Risk cases missed

---

Solution 2: Prioritise reviews by risk

High-risk applications first, low-risk cases through a faster path.

KPI:

Average manual review time

---

Solution 3: Automate repetitive checks

Automate simple document and data checks before an analyst sees the case.

KPI:

Manual minutes per application 8. Prioritise

I would prioritise:

1. Fix the compliance rule

Because it removes the main root cause.

Then:

2. Automate simple checks

Then:

3. Improve review prioritisation

Main trade-off:

Reducing reviews too aggressively could increase compliance risk.

So guardrail:

Compliance breach rate↑ 9. Executive summary

“Our objective was to reduce onboarding time from 5 days to 3 days.

The main issue is manual review. The review rate increased from 20% to 45%, and 70% of the additional cases are approved without changes.

I would first improve the compliance rule to reduce false positives, then automate simple checks, and finally prioritise reviews by risk.

I would track manual review rate, review time and manual minutes per application, while ensuring compliance risk does not increase.”

Over the last 3 months, Revolut’s customer support cost has increased by 25%, while the number of customers has only increased by 10%. We want to understand why and reduce the cost.

1. Reframe

“So support cost has risen faster than customer growth. Our goal is to understand the causes and reduce support cost. Is that correct?”

Interviewer: Yes.

2. Clarify

“What is the target and timeline?”

Interviewer: Reduce support cost by 15% within 6 months, without reducing customer satisfaction.

“Have we already tried anything?”

Interviewer: No.

3. Structure

I would break cost into:

Support Cost=Number of contacts×Cost per contact

So I want to investigate:

1. More customer contacts

2. Higher cost per contact

4. Ask for data

“Has the number of support contacts increased?”

Interviewer: Yes, by 30%.

So contact growth is likely a major driver.

“Has cost per contact also changed?”

Interviewer: It fell slightly.

So the main problem is more contacts, not agents becoming more expensive.

5. Analyse deeper

I would break contacts into reasons:

“What are customers contacting support about?”

Interviewer:

• Card issues: +10%

• Transfers: +15%

• Account verification: +80%

• Other: roughly flat

So verification is the biggest increase.

“Why have verification contacts increased?”

Interviewer: A new verification process causes many customers to fail the first attempt.

“Do we know the failure rate before and after the change?”

Interviewer:

Before:

8%

Now:

22%

That looks like the key root cause.

6. Root causes

I see two main causes:

1. Verification failure increased from 8% → 22%

2. Failed verification creates extra support contacts

So:

Higher failure rate→More contacts→Higher support cost 7. Solutions

Solution 1: Fix verification failures

Improve the verification flow and error messages.

Why first?

Because verification is the fastest-growing support reason.

KPI:

Verification failure rate

Target: move from 22% closer to 8%.

---

Solution 2: Add self-service for common verification issues

For example, explain exactly why verification failed and how to retry.

This reduces unnecessary agent contact.

KPI:

Support contacts per failed verification

---

Solution 3: Automate simple verification support requests

Use automated support for repetitive questions, while sending complex cases to agents.

KPI:

Cost per support contact 8. Prioritise

I would do:

1 → Fix verification flow

Largest root cause.

2 → Improve self-service

Fast and relatively cheap.

3 → Automate support

Useful, but it treats the symptom more than the root cause.

Main trade-off:

Making verification easier cannot weaken fraud controls.

So I would monitor:

Verification success↑

while ensuring:

Fraud rate↑ 9. Executive summary

“Our goal was to reduce support cost by 15% within 6 months without harming customer satisfaction.

The main issue is not cost per contact. Support contacts increased by 30%, driven mainly by verification issues.

Verification failure increased from 8% to 22%, creating additional support demand.

I would first fix the verification flow, then improve self-service, and finally automate repetitive support requests.

I would track verification failure rate, contacts per failed verification and cost per support contact, while keeping fraud and customer satisfaction as guardrails.”

Only 5% of Revolut customers currently use international transfers. We want to increase this to 10%.

1. Reframe

“So currently 5% of customers use international transfers, and we want to double adoption to 10%. I’d like to understand what prevents customers from using the product before recommending solutions. Is that correct?”

Interviewer: Yes.

---

2. Clarify

“What is the timeline?”

Interviewer: 6 months.

“Is the goal purely adoption, or should we maintain profitability?”

Interviewer: Increase adoption while maintaining positive contribution margin.

“Any markets in scope?”

Interviewer: UK, US and Europe.

---

3. Structure

First:

Usage rate=Customers using international transfers​ / Active customers

To understand low usage, I would look at the customer funnel:

Need→Availability→Start→Complete

So:

1. Demand: do customers need international transfers?

2. Availability: can Revolut serve their destination?

3. Conversion: if available, why don't they use Revolut?

4. Completion: are users dropping out during the process?

---

4. Ask for data

“What percentage of customers actually make international transfers somewhere, including competitors?”

Interviewer: 15%.

Good.

So demand exists:

15%>5%

We only need to capture another 5 percentage points.

“Of the 10% using competitors instead of Revolut, why?”

Interviewer:

Reason	Customers
Destination not supported	5%
Prefer competitor's price/rate	3%
Start Revolut transfer but abandon	2%

So all three areas matter.

---

5. Analyse

A. Destination availability

“Which unsupported destinations create most of the demand?”

Interviewer:

Five countries account for 80% of unsupported demand.

So:

5%×80%=4%

Potentially 4 percentage points are concentrated in five corridors.

“Is there anything preventing us from supporting those countries?”

Interviewer: Two require major regulatory work. Three can be launched relatively easily.

So I would focus on the three feasible corridors first.

---

B. Price

“For the 3% choosing competitors on price, are our fees higher?”

Interviewer: Headline fees are similar, but competitors show the final amount earlier.

Interesting.

So this may be price transparency, not actual price.

“Do customers abandon after seeing the final exchange rate?”

Interviewer: Yes. That step has a 35% drop-off rate.

---

C. Transfer journey

“Where does the remaining 2% abandon?”

Interviewer:

• Recipient details: 20%

• Verification: 55%

• Confirmation: 25%

“Why is verification causing the biggest drop?”

Interviewer: Some existing verified Revolut customers are being asked for additional information again.

So unnecessary repeated verification is creating friction.

---

6. Root causes

I see three main causes.

1. Missing transfer corridors

Five destinations create most unsupported demand, with three relatively easy to launch.

2. Poor price transparency

Customers only see the final exchange outcome late in the journey.

3. Repeated verification

Already-verified customers face unnecessary extra checks.

So:

Low adoption=Availability+Price transparency+Process friction

---

7. Solutions

Solution 1: Launch the 3 feasible corridors

These address the largest quantified opportunity.

KPI:

Transfer users from new corridors

---

Solution 2: Show total cost upfront

Show:

• fee

• exchange rate

• recipient receives

before the customer progresses.

KPI:

Quote→Transfer conversion

---

Solution 3: Remove unnecessary repeat verification

Reuse existing verified information where regulation allows.

KPI:

Verification completion rate

Guardrail:

Compliance failures↑

---

Solution 4: Target existing international-transfer users

Once the product issues are fixed, target customers already likely to transfer internationally.

I would not do this first because advertising a broken funnel wastes money.

KPI:

New international transfer users

---

8. Prioritise

I would sequence them:

1. Add the 3 feasible corridors

Largest opportunity.

2. Fix repeated verification

Clear process bottleneck.

3. Improve price transparency

Low-cost conversion improvement.

4. Promote the product

Only once the funnel works better.

Main trade-offs:

• New corridors require operational and compliance work.

• Less verification must not increase financial-crime risk.

• Better pricing transparency may expose uncompetitive pricing, which we would then need to address.

---

9. Executive summary

“Our objective was to increase international-transfer adoption from 5% to 10% within 6 months while maintaining profitability.

We found that demand already exists because 15% of customers make international transfers somewhere.

The three main barriers are unsupported destinations, poor price transparency and repeated verification.

I would first launch the three feasible high-demand corridors, then remove unnecessary verification friction and show the full transfer cost earlier. Once those issues are fixed, I would target customers with existing international-transfer needs.

I would track overall transfer adoption, new-corridor usage, transfer conversion and verification completion, while monitoring contribution margin and compliance risk.”

The cost of resolving card-payment disputes has increased by 35% over the last quarter, while dispute volume increased by only 10%. Average resolution time has also increased from 4 to 7 days. What should we do?

1. Reframe

“So dispute cost is growing much faster than volume, and resolution time has increased from 4 to 7 days. We need to identify why and improve both cost and speed. Is that correct?”

Interviewer: Yes.

---

2. Clarify

“What are our targets and timeline?”

Interviewer:

• Reduce cost per dispute by 20%

• Resolution time below 5 days

• Within 6 months

• Do not increase customer losses or compliance risk

---

3. Structure

First:

Total Cost=Disputes×Cost/Dispute

Since:

Disputes↑10%

but:

Cost↑35%

cost per dispute must also have risen.

I would investigate the dispute process:

Submission→Classification→Review→Decision

For each step:

• volume

• time

• manual work

• rework

---

4. Ask for data

“How much has cost per dispute changed?”

Quick calculation:

1.101.35​=1.227

So:

Cost/dispute↑23%

“Which part of the process has changed most?”

Interviewer: Manual review.

Manual review rate:

40%→65%

That looks important.

---

5. Analyse A. Why more manual reviews?

“Why are cases being sent to manual review?”

Interviewer: Many applications are missing merchant evidence.

“How significant is that?”

Interviewer: 60% of the extra manual reviews have missing information.

So the submission process is creating avoidable work.

---

B. Why is resolution slower?

“Where did the extra 3 days come from?”

Interviewer:

Stage	Before	Now
Submission	0.5 d	0.5 d
Queue	1 d	3 d
Review	2 d	2.5 d
Decision	0.5 d	1 d

Biggest issue:

Queue: 1→3 days

“Why has the queue increased?”

Interviewer: All disputes enter the same queue regardless of complexity.

---

C. Is there rework?

“How many cases require repeated work?”

Interviewer:

Reopened cases:

10%→25%

“Why?”

Interviewer: A new classification system incorrectly routes some disputes.

So we now have another root cause.

---

6. Root causes

Three main issues:

1. Incomplete submissions Missing evidence→Manual review↑→Cost↑ 2. One queue for every case Simple+Complex cases→Queue↑→Resolution time↑ 3. Incorrect classification Wrong routing→Rework→Cost+Time↑

---

7. Solutions Solution 1: Fix dispute submission

Require the necessary information before submission.

Example:

“Please upload merchant evidence before continuing.”

KPI:

Incomplete submission rate

---

Solution 2: Fix classification

Improve routing rules using historical dispute outcomes.

KPI:

Correct classification rate

and

Reopen rate

---

Solution 3: Create separate queues

For example:

• simple cases

• complex cases

• high-risk cases

Simple disputes should not wait behind complex ones.

KPI:

Average queue time

---

Solution 4: Automate simple disputes

Only after fixing the process, automate clearly low-complexity cases.

KPI:

Cost/dispute

Guardrail:

Incorrect decision rate

---

8. Prioritise

I would do:

1. Fix submission

Because it causes 60% of additional manual reviews.

2. Fix classification

Because reopens increased:

10%→25%

3. Separate queues

Targets the biggest time increase:

1→3 days

4. Automate

After fixing the underlying process.

Main trade-off:

Too much automation could create incorrect dispute decisions.

So monitor:

Cost/dispute↓

while:

Customer loss rate↑

---

9. Executive summary

“Our objective was to reduce cost per dispute by 20% and resolution time below 5 days within 6 months.

I found three main causes: incomplete submissions increased manual reviews, incorrect classification increased rework, and a single queue increased waiting time.

I would first improve the submission process, then fix classification, separate simple and complex queues, and finally automate suitable cases.

I would track incomplete submissions, reopen rate, queue time and cost per dispute, while monitoring incorrect decisions and customer losses.”

Revolut’s hiring process is taking too long and too few candidates are reaching offer acceptance. How would you improve it?

1. Reframe

“So we have two issues: hiring takes too long, and candidate conversion is too low. We need to identify where the process is failing and improve both. Is that correct?”

Interviewer: Yes.

2. Clarify

“What are the current numbers and targets?”

Interviewer:

Current:

Time to hire=42 days

Application→Accepted offer=1.8%

Target within 6 months:

30 days

and

3%

“Any constraints?”

Interviewer: No major increase in hiring headcount or hiring cost.

3. Structure

I would split this into:

Speed : Time to hire=Time at each hiring stage

Conversion: Overall conversion=Conversion through each stage

So I want to examine the funnel:

Apply→Screen→Assessment→Interview→Final→Offer→Accept

4. Ask for data

“Can we see time and conversion at each stage?”

Interviewer:

Stage	Time	Conversion
Application → Screen	3 d	30%
Screen → Assessment	4 d	70%
Assessment → Interview	10 d	35%
Interview → Final	12 d	45%
Final → Offer	5 d	60%
Offer → Accept	8 d	55%

Two things stand out:

• Interview stages create 22 days

• Offer acceptance is only 55%

5. Analyse

A. Why are interviews slow?

“Why do Assessment → Interview and Interview → Final take so long?”

Interviewer: Scheduling.

“What specifically causes the scheduling delay?”

Interviewer: Candidates wait for interviewer availability. Each candidate sees 3 different interviewers.

“Do all 3 interviews provide different information?”

Interviewer: Two interviews overlap heavily.

So there is duplicated assessment.

B. Why is offer acceptance low?

“Why are 45% rejecting offers?”

Interviewer:

Main reasons:

• 40% accepted another offer first

• 35% salary expectations differed

• 15% role expectations differed

• 10% other

“When are salary and role expectations discussed?”

Interviewer: Usually late in the process.

That creates avoidable late-stage drop-off.

C. Is there another conversion issue?

“Is there any stage where we reject many good candidates?”

Interviewer: Yes. Assessment pass rate is 35%, but hiring managers later say many rejected candidates may have been suitable.

“Has the assessment been validated against employee performance?”

Interviewer: No.

So the assessment may also be filtering too aggressively.

6. Root causes

I see three main causes.

1. Duplicate interviews More interviews→Scheduling delays→Time to hire↑

2. Salary and role alignment happens too late Late mismatch→Offer rejection↑

3. Assessment may reject suitable candidates Overly strict filter→Conversion↓

7. Solutions

Solution 1: Remove duplicated interview content

Combine overlapping interviews or make each one test a clearly different skill.

KPI:

Interview cycle time

Solution 2: Align salary and role expectations at screening

Confirm early:

• salary range

• location

• responsibilities

• working model

KPI:

Offer acceptance rate

Solution 3: Improve scheduling

Use shared interviewer slots and automatic scheduling.

KPI:

Days waiting for interview

Solution 4: Validate the assessment

Compare assessment scores with actual employee performance.

If weak correlation exists, redesign the cut-off or test.

KPI:

Assessment→Interview conversion

Guardrail:

Quality of hire

8. Prioritise

I would do:

1. Fix salary and role alignment

Cheap, fast, directly targets the 45% offer rejection.

2. Remove duplicate interviews

Directly targets the 22-day interview bottleneck.

3. Improve scheduling

Reduces waiting further.

4. Validate the assessment

Important, but requires more data and testing.

Main trade-off:

Fewer interviews and a less strict assessment could reduce hiring quality.

So I would monitor:

Time to hire↓

and

Conversion↑

while keeping:

Quality of hire↓

9. Executive summary

“Our objective was to reduce time to hire from 42 to 30 days and improve application-to-acceptance conversion from 1.8% to 3%.

I found three main issues: duplicated interviews create scheduling delays, salary and role expectations are aligned too late, and the assessment may reject suitable candidates.

I would first align expectations during screening, then remove duplicated interviews and improve scheduling, followed by validating the assessment.

I would track time to hire, offer acceptance and stage conversion, while monitoring quality of hire.”