SkillsSkill 3 of 5
Underwrite the cash flows that could shrink
Forecast a shrinking line on its own, across a range of decline rates, and price the deal on that.
Worked example
Revenue grew 9.1%. Is the review business safe?
A healthy total can hide a business in decline. Split the bridge before you forecast it.
The question
Quillfield, a fictional claims business, sells software to insurers and also reviews claims documents for them with its own staff. Carriers are testing AI tools that extract data from claims documents. The management case grows every line from FY25.
| FY24 | FY25 | Change | |
|---|---|---|---|
| Claims software | 120 | 138 | 18 |
| Recurring document review | 100 | 88 | -12 |
| One-off backlog project | 0 | 14 | 14 |
| Total | 220 | 240 | 20 |
Invented data. Total growth 20 ÷ 220 = 9.1%; without the one-off project, 6 ÷ 220 = 2.7%. Software grew 15.0% and recurring review fell 12.0% (rounded to 0.1%). Review, including the project, was 102 of 240, or 42.5% of FY25 revenue.
- Management
- “Review volumes dipped as two carriers piloted AI extraction. We expect them back.”
- Contracts
- Review is priced per document, with no minimum volumes.
- Backlog project
- Bought by the two largest customers to clear a claims backlog. No repeat is contracted.
Invented scenario.
“Quillfield’s revenue grew 9.1% last year and management forecasts growth in every line. How would you underwrite the document-review business?”
Your turn
Say your answer out loud, or jot notes below. Both work. Or skip to the worked answer.
Not saved by this site. Your browser may restore text when you navigate back.
A strong answer, step by step
Open each step in turn. This is one strong answer, not the only one.
Step 1: Split the total
What would you do here? Open the move you’d make.
Forecast from the 9.1%: revenue grew $20m last year, so high single digits is a fair base case.
Tempting, but it misses something. $14m of the $20m is a one-off project. Underlying growth was 6 ÷ 220 = 2.7%, made of two opposite trends.
Note that review held roughly flat, $100m to $102m, so it isn’t a concern.
Tempting, but it misses something. The $102m includes the $14m project. Recurring review fell from $100m to $88m, down 12.0%.
Split it: $14m of the $20m is one-off. Underlying growth was 2.7%, with software up 15.0% and recurring review down 12.0%.
The strong move. Two opposite trends need two forecasts; one blended growth rate would hide the decline.
See how a strong candidate says it
“The 9.1% is $20m, and $14m of it is a one-off project. Underlying growth was 2.7%: software up 15.0%, recurring review down 12.0%.”
Why it works: Growth that comes from two opposite trends needs two forecasts.
Step 2: Ask why review fell
What would you do here? Open the move you’d make.
Test management’s reason against the contracts: per-document pricing and no minimums mean volume only returns if carriers stop using AI.
The strong move. It ties “we expect them back” to the terms. Nothing obliges the carriers that piloted AI extraction to come back.
Accept the dip as temporary: two pilots are a small share of the carrier base.
Tempting, but it misses something. Management ties the dip to those pilots, and recurring review fell $12m, 12.0%, in one year. Pilots that work tend to spread.
Ask management for a more detailed volume forecast for next year.
Tempting, but it misses something. More detail from the same source doesn’t test it. The useful question is what would bring volume back when the contracts protect none of it.
See how a strong candidate says it
“Per-document pricing with no minimums means volume is the whole story. If carriers that piloted AI extraction keep using it, ‘we expect them back’ is a hope, not a forecast.”
Why it works: It tests management’s explanation against the contract terms.
Step 3: Underwrite a range, not a point
What would you do here? Open the move you’d make.
Use management’s case: review recovers to $100m once the pilots end.
Tempting, but it misses something. It underwrites a hope. Nothing in the contracts brings volume back, and last year’s trend was −12.0%.
Model review declining: at −12% a year, $88m becomes about $46.4m in five years. Run faster and slower declines too.
The strong move. When cash flows shrink, the rate is the unknown. A range around 12% shows how much value depends on it.
Average management’s recovery with a 12% decline and use the midpoint.
Tempting, but it misses something. A midpoint of two stories isn’t a forecast. It still assumes a recovery the contracts don’t support, and it hides how bad the low case is.
See how a strong candidate says it
“I’d model review as declining. If it keeps falling 12% a year, $88m becomes about $46.4m in five years. I’d also run faster and slower declines, because nobody knows the rate.”
Why it works: When cash flows shrink, how fast they shrink is the unknown that matters.
Step 4: Price it and find the offset
What would you do here? Open the move you’d make.
Apply one blended multiple to total EBITDA, since 9.1% growth looks healthy.
Tempting, but it misses something. A blended multiple pays software prices for a review line falling 12.0% a year. Valuing the two separately shows what the decline costs.
Walk away: any business with revenue exposed to AI is too risky.
Tempting, but it misses something. It skips the price. Software grew 15.0% to $138m of $240m; the question is what to pay for the mix, not whether AI exists.
Value software and review separately, assume no multiple expansion, and test whether Quillfield can sell extraction to carriers it’s losing.
The strong move. It turns the risk into a price, plus one diligence question that could offset it: can the software line, up 15.0%, capture what review loses?
See how a strong candidate says it
“I’d value software and review separately, assume no multiple expansion, and test whether Quillfield’s software can sell extraction to the carriers it’s losing as review clients.”
Why it works: It turns the risk into a price and one diligence question that could reduce it.
The answer, in short
- Decision
- Underwrite recurring review as a declining line across a range of rates, not management’s recovery, and price the deal on that.
- Because
- Once the one-off project is removed, review fell 12.0% while software grew 15.0%, and the contracts protect no volume.
- What would change it
- Carriers committing to minimum volumes, or proof that Quillfield’s own extraction product keeps the revenue it replaces.
Common traps
- Forecasting from the total growth rate.
- Accepting ‘volumes will come back’ without a reason.
- Assuming every line grows because the company did.
- Naming AI risk without putting it in the numbers.
Check your answer
- Did I separate one-off from recurring revenue?
- Did I forecast each segment on its own trend?
- Did I model a range of decline rates?
- Did I say what would make me less worried?
Want feedback on how you’d answer? Practice “Underwrite the cash flows that could shrink” with an AI interviewer on Socratify.
Last reviewed Nov 9, 2026