Defending a number

A request for money is a request for someone else to accept a risk. The figure tells them how much risk. It tells them nothing about what kind, how long, what would make it worse, or whether you would notice. Those are the questions that actually get answered before a decision, and if you do not answer them the other side will invent answers, which are usually more pessimistic than yours.

So the working unit is not a number. It is a number with four things attached to it, and the four are what make the number checkable. A checkable number is fundable. An unchecked one is a guess that happens to have decimal places.

• Facility. An agreed amount you may draw on, as opposed to a lump sum handed over on day one. Interest is normally charged on what is drawn.
• Drawdown. The act of taking money from a facility. A gap that opens slowly is usually drawn slowly.
• Headroom. The distance between where you expect to be and where a limit sits. Headroom of nothing is a plan that requires everything to go right.
• Covenant. A condition attached to money, usually a ratio that must stay on one side of a line, tested at fixed dates.
• Trough. The deepest point a cash balance reaches across a modelled period. Marchmont's is minus £115,501 in month 19.
• Buffer. The amount asked for above the trough, to absorb being wrong. It is a judgement, and the judgement has to be shown.
• Trigger. A condition agreed in advance that obliges you to come back and talk, before the position changes rather than after.

None of this is guidance on obtaining credit for any particular business. It is the vocabulary used in the room.

A figure without a period is unreadable in exactly the way session 1 described. £115,501 could be a monthly shortfall, an annual one, or a single deep point that lasts a fortnight. For Marchmont it is the last of those: the trough of a curve that goes under in month 10 and comes back in month 28. That shape decides what instrument fits. A gap that opens gradually and closes on its own is a different product from a permanent deficit, and the two get priced differently.

Saying "eighteen months" also commits you to something. It gives the other side a date on which to ask how it is going, which is uncomfortable and is the reason the request is credible.

Marchmont's figure rests on two things: revenue growing 4% a month, and customers paying in 75 days. Neither is a fact. Both are inputs, and the whole number moves when either moves. Putting them on the front page costs nothing and removes the worst moment in any funding conversation, which is the other side discovering an assumption you did not mention.

The test is simple. If somebody changed one input and the headline figure changed materially, that input belongs in the request. On those grounds growth and collection days belong in Marchmont's, and the depreciation life does not.

£115,501 is the bottom of the modelled curve. Asking for exactly that is asking for a plan with no headroom, where one slow-paying customer in month 18 breaks it. So the request is the trough plus something, and the something is a judgement about how wrong the assumptions could be.

What makes it defensible is showing the working rather than the answer. The growth ladder is already measured: at 2% a month there is no gap at all, at 4% the trough is £115,501 and closes in month 28, at 6% it is £897,962 and has not closed after four years. A buffer justified by that ladder is arguable. A buffer that is a round number chosen because it felt safe is not, and it invites the other side to pick their own.

A funding request with no end is a request to be carried. A request that names the month the money comes back is a request to be bridged. They feel similar from the inside and they are completely different from the other side of the table, because the second one has a defined exposure and the first does not.

Marchmont's version: the business is profitable in every month, the gap is created by growth it has already committed to, the worst point is minus £115,501 in month 19, and the balance is positive again in month 28 without anything changing. There is no turnaround in that sentence, no cost programme, and no promise to do better. The recovery is arithmetic, and that is what makes it checkable.

Somebody in the room will say the gap is self inflicted, and they have a case. Marchmont is owed money. Collecting in 45 days instead of 75 removes the shortfall entirely: cash never goes below zero and the month 12 balance is £168,521 instead of minus £47,261. Even 60 days is enough, at £60,630. Paying suppliers in 60 days rather than 30, or holding 30 days of stock rather than 60, each land at £77,892. And none of those four changes moves net income by a single pound, so on the income statement they are free. The obvious question follows: why borrow money at 9% when the cash is already yours and sitting in someone else's account?

The answer that finance people give is that a lever you do not control is not a plan. Collecting in 45 days means going back on terms customers have already agreed. Paying suppliers in 60 means spending goodwill you may want for something else. Holding 30 days of stock means a stockout on an instrument a laboratory ordered. The model prices the cash those levers release and says nothing at all about what they cost, and the cost is real even though it does not appear in any of the three statements.

This is not a disagreement that gets settled tonight, and it is not settled in practice either. One side holds that the cheapest capital available is the money you are already owed and that borrowing before collecting is a governance failure. The other holds that a working capital target is an intention while a facility is a certainty, and that funding a known, ending gap with debt is the honest move. Both positions are held by serious people. What is not defensible is arriving with only one of them prepared.

The request rests on 4% monthly growth. At 6% the trough is £897,962 and has not closed by month 48, which is a different company with a different problem. So the condition that matters is growth running above the rate the request was built on, and the useful move is to say in advance what you will do when it does: come back and talk, before the position moves rather than after.

Writing your own trigger is not a concession. A condition you proposed is one you have already thought about, and it removes the suspicion that you would let the position drift and present it as a surprise. It also makes the difference between the 4% case and the 6% case part of the record, so nobody has to rediscover it.