The uncomfortable truth about selling on credit
In the last two guides you saw a sale become real — under realization, you can book revenue once you have done the work and hold a reasonably collectible claim — and you watched that claim sit on the balance sheet as accounts receivable, the money customers owe you for goods already delivered. There was a quiet promise buried in the word 'collectible.' This guide cashes it in. Because the hard truth about selling on credit is that not every customer keeps their promise.
Picture a print shop with two hundred business accounts. Most pay within the month. But one customer goes bankrupt, another quietly disappears, a third disputes the bill into oblivion. The shop never finds out in advance which names those will be — yet from years of experience it knows, with near certainty, that some small fraction of every batch of credit sales will never turn into cash. This is not carelessness. It is the price of offering credit at all: the same generosity that wins you customers guarantees that a handful of them will stiff you.
So accounting faces an honesty problem. The balance sheet says receivables are 100,000, but the company already knows, today, that it will collect less than that. The question this whole guide answers is the question of timing: *when* do you admit the loss, and *how much* do you admit, given that you cannot yet name the customers who will fail you? Two methods compete for the answer, and only one of them tells the truth.
The tempting-but-wrong way: direct write-off
The obvious approach is to do nothing until you are certain. Wait until a specific account is truly hopeless — the customer is bankrupt, the phone is disconnected — then cross their name off and call the money gone. This is the [[direct-write-off-method|direct write-off method]]. When the print shop finally gives up on a 1,200 invoice, it records bad debt expense of 1,200 and erases that one receivable. No estimates, no special accounts, no guessing. It feels admirably precise: you only book a loss you can actually point to.
But precision in the wrong period is a kind of lie. Suppose the print shop made that credit sale last October and booked 1,200 of revenue then, in a year that looked nicely profitable. The customer collapses this February. Under direct write-off, last year's income statement proudly shows the full 1,200 of revenue with no offsetting cost, while *this* year's statement eats a 1,200 expense for a sale it never made. The cost of the bad customer has been ripped away from the revenue that created it. That violates the matching principle — the rule that an expense should sit in the same period as the revenue it helped earn — and it distorts the profit of two separate years at once.
There is a second flaw, and it lives on the balance sheet. Between October and February, everyone could plainly foresee that some receivables were rotting — yet direct write-off keeps reporting them at full face value until the very moment of surrender. The asset is overstated the whole time. For these two reasons — broken matching and inflated receivables — GAAP and IFRS will not allow the direct write-off method for real financial statements. It survives only in two corners: income-tax returns (where the tax authority wants a confirmed loss before granting a deduction) and tiny businesses whose bad debts are too small to matter.
The honest way: estimate first, name names later
The fix is to stop waiting for certainty. Under the [[allowance-method|allowance method]], at the end of each period you estimate the total bad debts hiding in your receivables — without yet knowing which specific customers they are — and you record that estimate now, in the same period as the sales. This is the same move you already met with depreciation: recognize a cost as the benefit is consumed, not when cash finally moves. Here the 'cost' is the slice of credit sales you will never collect, and recognizing it up front is what keeps the matching principle whole.
The estimate becomes a single adjusting entry with two sides. You debit bad debt expense — an operating expense that lands on this period's income statement and lowers profit. You credit the allowance for doubtful accounts — and this account is the clever part. It is a *contra-asset*, the very same device you saw in accumulated depreciation: an account with a credit balance that sits beneath accounts receivable and is subtracted from it. You never touch the Accounts Receivable account itself in this entry, because you cannot yet point to a single customer to remove. The allowance is a holding pen for an estimate, not a list of names.
On the balance sheet the two numbers stand together: gross receivables of 100,000, minus an allowance of, say, 4,000, equals receivables reported at their [[net-realizable-value-receivables|net realizable value]] of 96,000 — the cash the company honestly expects to collect. This is the conservatism principle in action: an asset is carried at the realistic cash it can bring in, never flattered above it. The print shop is now telling the truth in two places at once — its profit absorbs the cost of bad customers, and its receivables show what they are really worth.
What happens when a specific customer finally dies
Now the moment the direct-write-off crowd was waiting for: a named customer, say one owing 800, is confirmed hopeless. Here is the surprise that trips up almost every beginner. Under the allowance method, writing off that 800 records no expense at all. The expense was already taken — back when you set up the allowance, as an estimate. Now you are merely identifying which slice of that pre-funded cushion this particular failure consumes. So the entry is: debit the Allowance for Doubtful Accounts 800, credit Accounts Receivable 800. Both shrink by 800. The income statement is not touched.
And here is the quietly beautiful consequence. Gross receivables fall by 800, and the allowance falls by 800 — so the net realizable value, gross minus allowance, does not budge at all. It was 96,000 before the write-off and it is 96,000 after. The asset's honest value never flinched, because the loss had already been baked in. The write-off is pure housekeeping: it tidies away a dead account against the reserve that was always meant to absorb it.
Two ways to size the estimate
All of this rests on one number — the estimate — so how do you actually get it? There are two classic approaches, and they differ in what they aim at. The first, the percentage-of-sales method, takes an income-statement view: it asks what fraction of *this period's credit sales* will go bad. If history says 2% of credit sales sour, and the company made 500,000 of credit sales, then bad debt expense is simply 500,000 × 2% = 10,000. You debit 10,000 of expense and credit 10,000 to the allowance — straight to the point, because this method computes the *expense* directly.
The second approach, [[aging-of-receivables|aging of receivables]], takes a balance-sheet view and is more precise. It sorts every unpaid balance into buckets by how overdue it is, on a simple intuition you know in your bones: a debt from last week barely worries you, but a debt from a year ago, where the customer has stopped answering calls, feels nearly hopeless. The older the balance, the higher the loss rate you apply. Add up the estimated losses across all buckets and you get the *target balance* the allowance should hold.
Aging schedule Balance Loss % Est. uncollectible ------------------ ------- ------ ------------------ Not yet due 80,000 1% 800 1-30 days past due 12,000 5% 600 31-60 days past due 5,000 20% 1,000 Over 60 days 3,000 50% 1,500 ------------------ ------- ------------------ Total 100,000 = 3,900 (target) Allowance already has a 1,000 credit balance. Bad debt expense to record = 3,900 - 1,000 = 2,900
That subtlety in the table is the single most common exam trap, so let it sink in. Percentage-of-sales computes the expense and ignores whatever is already in the allowance. Aging computes the *target balance* of the allowance, so the expense is only the difference between that target and the balance already there. Same allowance method, two lenses: one aims straight at the income statement, the other at the balance sheet. Many companies use percentage-of-sales for quick monthly entries and then true everything up with an aging schedule at year-end.
Honest about an estimate built on judgment
The allowance method is honest about the past, but be honest about its limits too. The whole thing rests on a forecast, and a forecast is a judgment, not a fact. Loss rates come from history and from a manager's read of the economy — and reasonable people can disagree. That softness has a dark side: nudge the estimate down and this year's profit quietly swells; nudge it up and profit shrinks, building a reserve that can be released to flatter a future bad year. The allowance is therefore a classic spot where reported earnings can be smoothed or even manipulated, which is exactly why auditors probe it so hard.
One last twist for completeness: sometimes a customer you had given up on pays after all. You do not just pocket the cash quietly — you first reverse the earlier write-off, putting the receivable and the allowance back, and then record the collection normally. It is the write-off run in rewind, so the records stay coherent. With that, you can read a real balance sheet honestly: when you see receivables presented as a gross figure 'less allowance for doubtful accounts,' you now know it is a company admitting, out loud, that not all of what it is owed will come home.