Problem Definition
Kestrel Bank is a U.S. regional bank with approximately $8.5 billion in assets and 140 branches across three Great Lakes states. Under the Federal Reserve's Small Business Credit Survey definitions (a small bank has less than $10 billion in assets), Kestrel counts as a "small bank". It serves approximately 26,000 small-business customers, and small-business loans make up about $1.9 billion of its loan book.
Kestrel's small-business lending has always run on relationships. Owners apply at a branch, a relationship manager collects tax returns and bank statements, and a credit analyst underwrites each loan by hand. For loans under $250,000, the median time from application to credit decision is 21 days. Kestrel funds approximately 1,500 of these smaller loans a year, and the number has been flat for three years.
Across the U.S. market, the share of small-business loan applicants who go to online lenders rose from 17% in the 2020 survey to 29% in the 2025 survey (Federal Reserve Small Business Credit Survey). Kestrel's relationship managers report that customers increasingly say "I needed the money this week" and take offers from online lenders, even at higher rates.
The Chief Operating Officer has a proposal: a digital origination platform for loans up to $250,000, with an online application, automatic retrieval of bank and accounting data, and automated credit decisions for simpler cases. It would cost approximately $9 million to build and $1.5 million a year to run. The CEO has asked your team: should Kestrel invest in digital small-business lending, and if so, how should it roll it out?
Additional InformationAsk for dataInterviewer’s data
If asked, please share that:
- Kestrel's net interest margin is approximately 3.4%, close to the industry average of 3.32% reported by the FDIC for Q2 2026.
- Kestrel's return on assets is approximately 1.1%, below the industry's 1.37% in Q2 2026.
- 85% of the smaller loans Kestrel funds go to existing deposit customers.
- Kestrel is an approved SBA 7(a) lender, but it uses the program for only about 10% of its small loans. Nationally, the SBA 7(a) program guaranteed a record of approximately 77,600 loans worth approximately $37 billion in FY2025.
- There are approximately 36.2 million small businesses in the U.S., employing about 45.9% of private-sector workers (SBA Office of Advocacy).
- The board is cautious about credit risk after a spike in commercial real-estate losses two years ago.
Question 1Structuring
How would you structure your analysis of whether Kestrel should invest in digital small-business lending?
Hint · Structuring
Build 3–4 branches that are specific to this client and question, not a generic framework. Check they don't overlap and together cover the problem.
Additional InformationAsk for dataInterviewer’s data
- Share the $9M build cost and $1.5M annual running cost if asked.
- If the candidate asks about competitors: large national banks and several online lenders are active in Kestrel's states. No other local bank has a digital small-business loan product yet.
Try it first, then checkCheck my answerModel answer
A strong framework would cover four areas:
1. Customer and market
- a) What do small-business borrowers value, and where is Kestrel losing them (speed, ease, approval, price)?
- b) How big is the leakage to online lenders and large banks in Kestrel's footprint?
- c) Which segments matter most: existing deposit customers vs. new-to-bank firms, loan size, industry?
2. Economics per loan and in total
- a) Today's cost to originate a small loan vs. with a digital process
- b) Profit per loan over its life (interest spread, credit losses, servicing, origination)
- c) Volume uplift from faster decisions; payback on the $9M build
3. Risk
- a) Credit risk of automated decisions, especially for new-to-bank borrowers
- b) Fraud and data risk in an online channel
- c) Regulatory expectations (fair lending, model governance)
4. Execution
- a) Build, buy or partner for the platform
- b) Role of relationship managers after the change
- c) Phasing: which customers and products go first
What the interviewer is looking forInterviewer’s viewInterviewer’s view
A good structure goes beyond "costs vs. benefits of the platform". It should cover why customers are leaving, what a digital process changes in the economics of each loan, and what new risks it creates.
- Good candidates cover investment cost, revenue upside and implementation risk.
- Strong candidates split the benefit into (a) lower cost per loan and (b) more loans (win back speed-sensitive borrowers), and bring in credit risk.
- Excellent candidates start from the customer: what small-business owners value (speed, approval certainty, fair pricing, relationship). They ask what Kestrel's right to win is against online lenders, then decide the scope of automation (which customers and which loans).
Question 2Judgement & Insights
Exhibit 1 shows results from the Federal Reserve's latest Small Business Credit Survey. What does it tell you about where Kestrel can win?
Hint · Judgement & Insights
Read the exhibit title, axes and units first. Lead with the ‘so what’, then back it with one or two numbers.
Exhibit 1Small-Business Borrower Experience by Lender Type (2025 Survey)
| Metric (% of applicants at each lender type) | Large banks | Small banks | Online lenders |
|---|---|---|---|
| Chose lender because of existing relationship | 61% | 62% | 31% |
| Chose lender because of speed of decision or funding | 29% | 28% | 64% |
| Fully approved | 43% | 57% | 38% |
| Satisfied with lender (approved applicants) | 64% | 65% | 35% |
| Challenge: long wait for credit decision or funding | 21% | 16% | 8% |
| Challenge: difficult application process | 26% | 18% | 11% |
| Actual borrowing cost higher than expected (borrowers) | 32% | 37% | 60% |
Source: Federal Reserve Banks, 2026 Report on Employer Firms: Findings from the 2025 Small Business Credit Survey. Share of loan applicants applying at online lenders: 17% (2020 survey) → 29% (2025 survey).
Source: Kestrel Bank case file
Additional InformationAsk for dataInterviewer’s data
- Share Exhibit 1.
- If asked: the survey covers U.S. employer firms with 1–499 employees, and 38% of them applied for a loan, line of credit or merchant cash advance in the prior 12 months.
- Clarify if needed: "Satisfied" is measured among applicants approved for at least some financing.
Try it first, then checkCheck my answerModel answer
Where small banks already win:
- Approval: 57% fully approved at small banks vs. 43% at large banks and 38% at online lenders.
- Satisfaction: 65% satisfied vs. only 35% at online lenders.
- Relationships: 62% of small-bank applicants chose their lender because of an existing relationship.
- Price transparency: 60% of online-lender borrowers paid more than they expected, vs. 37% at small banks.
Where small banks lose:
- Speed: 64% of online-lender applicants chose them for speed. At small banks, 16% of applicants complain about a long wait, twice the online-lender rate (8%).
- Ease: 18% of small-bank applicants find the application process difficult vs. 11% at online lenders.
What this means for Kestrel:
- Borrowers are not leaving because Kestrel says "no" or charges too much. They leave because Kestrel is slow. The 21-day median decision time is the problem to fix.
- Kestrel's natural advantage is its existing relationships. It already holds deposit and payment data on 85% of the firms it lends to, and that data can support fast automated decisions.
- Kestrel should not copy online lenders' pricing or risk appetite. Their borrowers are the least satisfied, so a "fast but fair" bank offer is a real gap in the market.
What the interviewer is looking forInterviewer’s viewInterviewer’s view
The candidate should see that small banks already win on approval, satisfaction and price transparency, and lose only on speed and ease of applying. Online lenders win on speed, but their borrowers are the least satisfied and the most often surprised by costs. The insight is that Kestrel does not need to become a fintech. It needs to remove the speed and paperwork gap while keeping its pricing and relationship advantage.
Apply the "So What?" cascade when evaluating answers:
- Level 1 (surface): "Online lenders are growing, and 64% of their applicants chose them for speed" — this is reading the data
- Level 2 (implication): "Borrowers accept worse pricing and service from online lenders only because banks are slow; small banks already beat them on approval and satisfaction" — this shows pattern recognition
- Level 3 (actionable): "Kestrel should compete on speed for the customers it already knows, using its own deposit data to decide in days, not weeks, while keeping bank pricing" — this is strategic judgement
Question 3Numeracy
Using Exhibit 2, calculate the lifetime profit of one small loan under today's process and under the digital process. Then calculate the payback on the $9M platform. Finally, how does the answer change if loan volume does not grow?
Hint · Numeracy
Write the formula before you plug in numbers, keep units and zeros explicit, and sanity-check the order of magnitude at the end.
Exhibit 2Economics of a Small-Business Loan Under $250,000
| Metric | Today (branch process) | Digital process (COO estimate) |
|---|---|---|
| Average loan size | $80,000 | $80,000 |
| Average balance over the loan's life | $50,000 | $50,000 |
| Loan life | 3 years | 3 years |
| Net interest spread (per year, on average balance) | 4.0% | 4.0% |
| Expected credit losses (per year, on average balance) | 0.8% | 0.8% |
| Servicing cost per year | $300 | $300 |
| One-time origination cost (staff time, underwriting, documentation) | $3,600 | $1,100 |
| Staff hours per loan | 30 | 8 |
| Loans funded per year | 1,500 | 2,600 |
Kestrel internal data and COO business case. For comparison, a widely cited industry estimate puts the cost to originate, set up and monitor a small-business loan of $100,000 or less at $3,000–4,000 (ProSight Financial Association, 2015).
Source: Kestrel Bank case file
Additional InformationAsk for dataInterviewer’s data
- Share Exhibit 2.
- Assume all figures are per loan and that each year's loans form one "cohort". Treat the lifetime profit of a year's cohort as that year's benefit (a simplification; ignore discounting).
- The platform costs $9M upfront and $1.5M a year to run.
Try it first, then checkCheck my answerModel answer
Step 1: Lifetime profit per loan
| Item | Today | Digital |
|---|---|---|
| Net interest: $50,000 x 4.0% x 3 years | $6,000 | $6,000 |
| Credit losses: $50,000 x 0.8% x 3 years | -$1,200 | -$1,200 |
| Servicing: $300 x 3 years | -$900 | -$900 |
| Origination | -$3,600 | -$1,100 |
| Lifetime profit per loan | $300 | $2,800 |
Today, origination cost uses up 60% of the net interest earned ($3,600 / $6,000), so the loan barely breaks even.
Step 2: Annual cohort profit
- Today: 1,500 x $300 = $0.45M
- Digital: 2,600 x $2,800 = $7.28M
- Incremental benefit: $7.28M - $0.45M = $6.83M
- Less running cost: $6.83M - $1.5M = $5.33M net per year
Step 3: Payback
- $9M / $5.33M = approximately 1.7 years
Step 4: Sensitivity with no volume growth (1,500 loans)
- Digital cohort profit: 1,500 x $2,800 = $4.20M
- Incremental: $4.20M - $0.45M = $3.75M; less $1.5M running cost = $2.25M net per year
- Payback: $9M / $2.25M = 4.0 years
Step 5: Staff capacity freed
- Today, 1,500 loans x 30 hours = 45,000 hours. Digital at the same volume: 1,500 x 8 = 12,000 hours. That frees about 33,000 hours a year, which could go to relationship managers spending more time on larger, more complex loans.
So what: The platform pays back on cost savings alone, but slowly (4 years). The investment case depends on volume growth, which depends on whether faster decisions actually win back borrowers. The pilot data should tell us that.
What the interviewer is looking forInterviewer’s viewInterviewer’s view
Key numbers: lifetime profit per loan of $300 today vs. $2,800 digital; incremental annual benefit of ~$6.8M before running costs; payback of ~1.7 years with volume growth, and ~4.0 years without it. The insight is that today small loans barely break even, because the fixed cost of origination eats almost all the margin. The digital case is attractive on cost alone, but it becomes compelling only if Kestrel also wins more loans. Strong candidates also check whether the volume uplift is realistic.
Question 4Judgement & Insights
Kestrel ran a six-month pilot of the digital process in one state. Exhibit 3 compares the pilot with the branch process. What do you conclude, and what would you change before a full rollout?
Hint · Judgement & Insights
Read the exhibit title, axes and units first. Lead with the ‘so what’, then back it with one or two numbers.
Exhibit 3Pilot Results (per 1,000 applications started)
| Stage / metric | Branch process | Digital pilot |
|---|---|---|
| Applications started | 1,000 | 1,000 |
| Applications completed | 610 | 540 |
| Approved | 350 | 300 |
| Loans funded | 290 | 270 |
| Median time to credit decision | 21 days | 2 days |
| Existing Kestrel customers, share of funded loans | 85% | 55% |
| 30+ days delinquent after 12 months (all funded loans) | 1.8% | 3.1% |
| — of which existing customers | 1.7% | 1.6% |
| — of which new-to-bank | 2.4% | 5.0% |
Kestrel pilot data. Each process tracked separately. The digital pilot was open to existing customers and to new applicants from online advertising.
Source: Kestrel Bank case file
Additional InformationAsk for dataInterviewer’s data
- Share Exhibit 3.
- If asked: most new-to-bank applicants in the pilot came through online search advertising.
- If asked: the pilot auto-decided every application with the same credit model, whether or not the applicant already banked with Kestrel.
Try it first, then checkCheck my answerModel answer
1. Speed wins customers. In the pilot, 90% of approved applicants took the loan (270 / 300) vs. 83% in the branch (290 / 350). Deciding in 2 days instead of 21 stops approved borrowers from going to a faster competitor.
2. The online application loses people. Only 54% of digital applications were completed vs. 61% in branches. Asking business owners to upload documents is still hard. Kestrel already holds bank data for existing customers, so it should pre-fill their applications.
3. The risk is concentrated in one segment. Existing customers perform the same in both channels (1.6% vs. 1.7% delinquent). The higher overall delinquency (3.1%) comes from new-to-bank borrowers at 5.0%, about three times the rate for existing customers. Check: 55% x 1.6% + 45% x 5.0% = 0.88% + 2.25% = approximately 3.1%. The same credit model was used for borrowers Kestrel knows well and borrowers it does not know at all.
What to change before a full rollout:
- Existing customers: fast lane. Pre-approved offers and automated decisions using Kestrel's own deposit and payment data. This is the lowest-risk, highest-conversion segment.
- New-to-bank: guarded lane. Require connected bank or accounting data (cash-flow underwriting). Use lower starting limits (e.g., up to $50,000) and send larger or unclear cases to a human credit analyst.
- Fix the form: fewer fields, document upload from phone, "save and continue later", and relationship-manager follow-up for applications left unfinished.
- Watch the numbers: track delinquency by segment monthly, with a threshold that triggers tighter rules for new-to-bank lending.
What the interviewer is looking forInterviewer’s viewInterviewer’s view
The candidate should see three things: (1) speed converts: a higher share of approved applicants take the loan; (2) the online form loses applicants before they finish; (3) the risk problem sits entirely with new-to-bank borrowers, where delinquency is about three times that of existing customers. A strong candidate recommends different treatment by segment rather than stopping the project. Excellent candidates propose concrete fixes (pre-filled applications for existing customers, cash-flow data, manual review or lower limits for new-to-bank applicants).
Question 5Synthesis
The CEO has to take this to the board next week. What is your recommendation?
Hint · Synthesis
Answer first: the recommendation, two or three reasons with numbers, then risks and next steps.
Additional InformationAsk for dataInterviewer’s data
- If needed, remind the candidate: lifetime profit per loan $300 today vs. $2,800 digital; payback approximately 1.7 years with growth and 4.0 years without; pilot delinquency 1.6% for existing customers vs. 5.0% for new-to-bank.
Try it first, then checkCheck my answerModel answer
Recommendation: Invest the $9M, but roll out in two phases: existing customers first, new-to-bank customers only once risk controls are proven.
Why invest:
- Small loans barely make money today. Each earns only about $300 over its life. The digital process raises that to about $2,800 per loan, so it pays back in about 4 years even with no growth.
- Speed is the one thing Kestrel is missing. Small banks already beat online lenders on approval (57% vs. 38%) and satisfaction (65% vs. 35%). Borrowers leave because of slow decisions and paperwork, and the pilot cut decision time from 21 days to 2.
- With volume growth the payback is about 1.7 years. The pilot already shows higher take-up of approved loans.
Phased rollout:
| Phase | Scope | Target |
|---|---|---|
| Months 0–9 | Build the platform; launch for existing deposit customers in all three states, with pre-filled applications | Median decision ≤ 3 days; completion rate ≥ 65% |
| Months 9–18 | Add cash-flow underwriting; open to new-to-bank firms with limits up to $50,000 and manual review above that | New-to-bank delinquency ≤ 3% |
| Month 18+ | Raise new-to-bank limits if performance holds; use freed staff time (about 33,000 hours a year) on larger relationship loans | 2,600+ small loans a year |
Key risks and mitigations
- Credit losses from new customers: segment-specific credit rules, lower starting limits, monthly monitoring against a hard threshold.
- Volume uplift does not happen: the investment still pays back on cost savings in about 4 years; set a volume gate at month 12 before spending on marketing to attract new-to-bank firms.
- Relationship managers resist: give them credit for digital loans in their portfolio and move them to advisory work, not processing.
- Regulatory and model risk: independent validation of the credit model and fair-lending testing before launch.
What the interviewer is looking forInterviewer’s viewInterviewer’s view
Look for a clear "yes, with conditions" or similar answer, supported by numbers from the case, and a phased rollout that deals with the board's concern about credit risk. Excellent candidates connect the recommendation to Kestrel's right to win ("bank pricing at fintech speed for the customers we already know") and propose clear targets for the board.
Data Sources
Company figures for Kestrel Bank (assets, branches, loan volumes, costs per loan, pilot results) are invented for this case. Market facts come from the sources below. Figures are rounded for interview math.
- Share of applicants at online lenders 17% (2020) → 29% (2025 survey); lender choice for relationship (61% / 62% / 31%) and speed (29% / 28% / 64%); full approval 43% / 57% / 38%; satisfaction 64% / 65% / 35%; long wait 21% / 16% / 8%; difficult application 26% / 18% / 11%; higher-than-expected cost 32% / 37% / 60%; 38% of firms applied for a loan, line or cash advance; small bank = under $10B in assets → Federal Reserve Banks, 2026 Report on Employer Firms: Findings from the 2025 Small Business Credit Survey, 2026, https://www.fedsmallbusiness.org/reports/survey/2026/2026-report-on-employer-firms
- Industry return on assets 1.37% and net interest margin 3.32% (Q2 2026) → FDIC, FDIC-Insured Institutions Reported Return on Assets of 1.37 Percent and Net Income of $90.1 Billion in Second Quarter 2026, 2026, https://www.fdic.gov/news/press-releases/2026/fdic-insured-institutions-reported-return-assets-137-percent-and-net
- Cost to originate, set up and monitor a small-business loan of $100,000 or less approximately $3,000–4,000 → ProSight Financial Association, Making small business loans profitably, 2015, https://www.prosightfa.org/insights/making-small-business-loans-profitably/
- SBA 7(a) program: approximately 77,600 loans for approximately $37 billion in FY2025, a record → U.S. Small Business Administration, SBA Delivers Record Capital to Small Businesses in FY25, 2025, https://legacy.sba.gov/article/2025/09/30/trump-sba-delivers-record-capital-small-businesses-fy25
- Approximately 36.2 million U.S. small businesses employing 45.9% of private-sector workers → SBA Office of Advocacy, Frequently Asked Questions About Small Businesses 2026, 2026, https://advocacy.sba.gov/wp-content/uploads/2026/02/FINAL_FAQsAboutSmallBusiness_2026_012826.pdf
Free in the DrillCase case library · Editorial policy