Project Norbury — 3-Statement Model with Growth Equity Returns

Revenue in a staffing business is gross billings, most of which is contractor pay passing straight through. Net fee income — permanent fees in full plus the margin on contract hours — is the gross profit of the business and the line every operator, lender and buyer underwrites. It is the product of the average number of fee earners and the net fee income each of them produces, and nothing else. That is why this archetype exists: both halves are things diligence measures directly rather than inferring from a comparable multiple, so a plan built this way is testable in a way that a revenue growth rate is not.

This is the largest trap in the case. A staffing business invoices the client for the whole of the contractor's time and collects the whole of it; the receivable is on revenue whatever margin the business keeps. Drive days sales outstanding off net fee income instead and the balance sheet books a fraction of the receivable it should, the cash flow statement hands the model cash it never collected, and the whole file still ties and still balances. Nothing complains. Get the denominator right before you get anything else right, and check the days you were handed against what the last audited balance sheet implies.

Contractors are paid weekly; clients pay in about sixty days. Every pound of contract growth is therefore funded before it is collected, and the faster the business grows the worse its cash conversion looks. On top of that, a fee earner takes nine to twelve months to reach full productivity, so a year of heavy hiring carries the cost of the desks for most of a year before they bill. A plan whose first projected year gets better on conversion is not a hiring plan. Reading that dip as a problem rather than as the mechanic is a misread; not noticing it at all is worse.

The pool exists to hold fee earners, and the fee earners are the entire asset. An earnings figure that excludes the cost of holding them is valuing a business that does not exist, so the charge is taken inside Adjusted EBITDA here rather than added back. Adding it back is not forbidden; adding it back without restating the multiple is. Multiples struck on post-charge earnings applied to a pre-charge earnings base credit a real cost at a real multiple and manufacture enterprise value out of a definition. Same price, different basis: state which one you are on, every time.

Deeply subordinated, unsecured, no fixed repayment date, settled only on a liquidity event — under the company's accounting policy the instrument sits in shareholders' funds rather than in debt. Its coupon is therefore an appropriation within equity, a transfer from retained earnings to a reserve, not an expense: it does not reduce net income, it does not reduce the tax charge, and it changes total equity by nothing at all. That is the whole trap. The accretion is invisible on every balance sheet the candidate builds and decisive at the exit gate, because the accreted balance is settled out of exit equity value before the ordinary and the preferred shares divide anything.

Three articulation rules carry the balance sheet. Retained earnings are the prior balance plus net income less the shareholder loan accretion. The share-based payment reserve is the prior balance plus the year's charge, which is why the charge reduces net income, increases equity by the same amount and is added back on the cash flow statement. And the cash line is the cash flow statement's closing balance reached a second way. The check row is the only cheap proof that any of the three is right. Encode it in the number format so a zero renders as a word rather than as a zero, and read it in every year — a balance sheet that balances in four years out of five balances in none of them, because the error has only moved.

Each year's intangible additions amortize straight line over their life with half a year's charge in the year of purchase and a full year thereafter, layered on top of an existing book that runs off partway through the plan and then stops. Build it as one row per vintage. The convention itself is worth almost nothing to the answer, so a mechanic that changes almost nothing never corrects the candidate who gets it wrong. What is being graded is whether the schedule was built or simply typed in, and the tell is that the total amortization line does not rise smoothly across the plan. A D&A line that climbs neatly every year has not come from a schedule.

There are three plausible ways to compute what the fund owns and only one of them is a definition. Capital over pre-money is wrong. Primary over post-money is wrong. Shares acquired over shares outstanding is right, and it is right because the secondary tranche bought existing shares from the founders at the same price the primary subscribed at. Two consequences follow. Part of the money only ever reaches the founders, so a return that divides exit proceeds by the primary alone overstates the answer badly. And the option pool sits inside the pre-money count, so its dilution is borne by the founders before the fund subscribes at all.

Each preferred class independently takes the greater of its stated preference and its as-converted share, so a waterfall is a comparison at every exit level rather than a fixed split. Two facts make that live here. The classes are ranked, and the later one ranks ahead of the earlier one — a senior claim that did not exist when the earlier preference was negotiated changes what that preference is worth at low outcomes. And a 1.0x non-participating preference converts exactly when the company is worth what the round paid for it, so the crossover for each class is a number you can state and check rather than guess. Find both crossovers and say where the base case sits relative to them.

A minority holder cannot block a round a majority board approves. The plan carries one, already priced and already sized, and it changes the fund's ownership without the fund doing anything. The right way to price that is not the percentage of ownership lost multiplied by the exit value: that arithmetic ignores the cash the round put into the company, which is still there at exit and lifts the value the remaining ownership is a share of. The cost of dilution is the difference between two worlds — one in which the round happened and one in which it did not — and the gap between the two ways of counting is the most common error in growth-equity returns work. Ask the prior question too: run the plan without the round and compare the lowest cash balance it reaches against the stated minimum, and against what is still undrawn on a facility already committed at a fraction of a growth-equity cost of capital. A candidate who prices the dilution but never asks whether the cash was needed has answered half the question.

The two statistics everyone reaches for are the wrong ones here. Net leverage is negative in every projected year and interest cover runs away to numbers nobody quotes, so both say the same uninformative thing: there is essentially no debt. Neither speaks to the case's defining feature, which is that growth consumes cash. The two that do are cash conversion — operating cash flow over Adjusted EBITDA, worst in the year the hiring lands — and the borrowing base against the commitment on the discounting facility. Compute the advance rate on gross receivables each year and say which of the two binds, because the answer flips early, and the constraint that matters is fixed by a conversation with the bank rather than by a financing round.

Heads of terms are signed on one date and the deal completes on another, so the multiple quoted at signing and the multiple at completion are struck on different periods. Build both straddling periods on the stated weights and print the multiple on all four bases. Then check the shape rather than the arithmetic: a next-twelve-months figure must sit strictly between the two annual periods it is built from, and above the trailing one when the plan grows. A series that does not do that has not been calendarized, it has been mistyped. Quoting one basis alone is quoting the flattering one, whether or not you meant to.

In a buyout the third bucket is deleveraging. Here it is not: the company is in net cash from completion, only the term loan is ever repaid, and most of the movement in net bank debt is cash the business generated or that later investors put in. So label it for what it is. The full decomposition has five parts — the change in net fee income at the entry conversion capitalized at the entry multiple, the change in conversion applied to exit-year net fee income at the same multiple, the change in multiple applied to exit-year earnings, the movement in net bank debt, and the accretion on the shareholder loan — and the first two together must reconcile to the change in Adjusted EBITDA times the entry multiple. Splitting that into volume and margin is what turns an EBITDA bridge into an operating one. Then run the whole thing a second time on the fund rather than on the enterprise, because an ownership percentage that moves mid-hold separates the two, and reporting only one of them reports somebody else's outcome.

A fund that underwrites to both a money multiple and an internal rate of return has set two constraints that do not agree at every hold period. A fixed multiple over a longer hold is a lower annual return, so beyond some hold length no outcome can satisfy both standards at once, and whichever one gets quoted at the investment committee the other is being waived. Work out where that crossover sits for the standards you have been handed, and say so. It is a two-line calculation and almost nobody volunteers it.

The completed workbook and a short written answer. There is no deck, no slides and nothing presented. The written answer has to cover the three things the prompt asked for — the money multiple and the internal rate of return at each exit year, what the fund owns at exit rather than today, and where the return came from — plus what you would pay and what the model leaves out. The sixty minutes is a hard limit on the workbook and a soft limit on the answer, which is a distinction worth taking literally: a candidate who balances the balance sheet at minute fifty-eight and says nothing about ownership has completed the test and failed the exercise.

No, and treating it as one is the fastest way to lose the exercise. This is a minority growth investment. The company is in net cash from the day the round funds, there is no cash sweep, no revolver, no minimum-cash plug and no debt schedule in the buyout sense — the one amortizing facility repays on a fixed schedule and nothing else moves. Deleveraging is not a source of return. What does the work instead is ownership, dilution and a payout waterfall, and the returns attribution has no debt paydown line at all.

Because the price is given as a pre-money equity value and the exit is given as a multiple, so there is nothing to triangulate. There are no comps, no precedent transactions, no football field, no terminal value and no weighted average cost of capital in this exercise. Adding any of them to a build that already carries three statements, a layered amortization schedule and a waterfall would not fit sixty minutes, and making one up would misteach the archetype.

Not here, and the reasoning matters more than the answer. In a recruitment business the option pool is consultant compensation in a different wrapper: the pool exists to hold fee earners, the fee earners are the whole asset, and earnings struck without the cost of holding them describe a business that does not exist. The charge is also reflected in the ownership arithmetic, since the pool sits in the fully diluted count — but that is a different cost of the same instrument, not a double count. And if you do add it back somewhere, you must restate every multiple you apply onto the matching basis, or you have credited a real cost at a multiple derived from businesses that charge it.

Under the company's accounting policy they sit in shareholders' funds: deeply subordinated, unsecured, no fixed repayment date, settled only on a liquidity event. That has three consequences a careful candidate spells out. Their coupon is an appropriation within equity rather than an expense, so it never touches profit before tax, the tax charge or net income. They are excluded from net bank debt at every date in the case. And they are still a claim that is settled out of exit equity value before the ordinary and the preferred shares divide anything — invisible on every balance sheet you build, decisive at the exit gate.

Both, in different places, and the prompt says so in one line: tell me what we own at exit, not what we own today. The completion cap table sets what the fund bought and is what the first stage of the attribution runs at. The exit cap table sets what the fund receives, and it differs because a later financing round and a pool top-up land in between. A model that runs the exit waterfall on the completion cap table produces a clean, defensible and wrong number, and nothing in the file complains.

At the pattern before the formulas. Off by the same amount in every year means the error came out of the pro forma opening position, so go back to the transaction tab — the usual culprits are the issue costs charged to the wrong place, or the secondary consideration put into sources and uses when it never touches the company. Off in one year only means it is that year's flow, and the usual culprits are the sign on the working-capital movement, the share-based payment charge added back on one statement but not the other, or the shareholder loan accretion routed into the reserve without the matching reduction in retained earnings.

Beginning, on everything — both facilities, the shareholder loan accretion and interest income on cash — and the instruction sheet says so along with an instruction not to switch on iterative calculation. Average-balance interest is the market convention and is more precise, but it makes the model circular, and a circular model can settle on a wrong answer without warning when someone edits a formula. If you normally build it the other way, say so on the tab. Knowing why a convention exists is worth more than the convention.

Proceeds over invested capital, raised to the power of one over the hold period, less one. Neither IRR nor XIRR appears anywhere in the model or the template, and neither should appear in yours. With a single outflow at completion and a single inflow at exit there is no cash flow series to discount, so the geometric form is not an approximation of the right answer — it is the right answer, and it is auditable by anyone reading the cell.

It is stated as graded, as it is on real modeling tests. Blue for a hardcoded input, black for a formula on the same tab, green for a link to another tab — and in a three-statement build the green links are the articulation, which is the thing being graded, so they are not handed to you. A hardcoded number buried inside a formula is marked down even when it is right. One related habit pays for itself immediately: make every reference to a scalar assumption absolute, because an unpinned scalar cannot be filled right across the year columns at all.

Cut the two sensitivities first and the price-discipline solve second. Do not cut the check row, do not cut the exit waterfall's deduction of the accreted shareholder loan, and do not cut the attribution — write it in prose if you cannot build it. Three statements that tie, one exit column that is right, and an owner who can say where the value came from beats three exit columns computed off a waterfall that skipped a claim.

Yes, and naming it is free marks. Seven items are excluded on purpose and none of them is priced: the tax treatment of the share-based payment charge, which really needs a deferred tax schedule; a flat base rate with no curve and no floor; the discounting facility held flat rather than flexed against its borrowing base; tax installment timing; exit costs, warranty escrow and management transaction bonuses; options assumed exercised at nil cost; and no secondary sale by the fund before the exit date. State the direction of each and move on. Also absent by design: no purchase price allocation and no goodwill, because a minority primary subscription is not a business combination; no participating preferred, no ratchet, no anti-dilution and no cumulative dividend; and no Excel data tables anywhere, so every sensitivity is real formulas whose center reproduces the base case and can be proven to.

It is the standard second-round screen at growth funds, and it sits one tier above a quick buyout test and one below a two-hour full model. It exists to find out whether three statements articulate in a candidate's hands without being told how, which is hard to fake: a balance sheet either closes in every year or it does not, and a cash flow statement built from a blank tab either reproduces the balance sheet's cash line or it does not. Thirty seconds of scrolling tells an interviewer which. The growth-equity layer on top is what distinguishes it from a banking modeling test — ownership that moves, a round the investor cannot block, and a waterfall with a senior class that did not exist when the earlier one was negotiated.