Every page instructs the same two point average. Eight of the ten also warn about what it hides, in a different section, without changing the instruction.
The interesting part is that eight of those ten also warn you about what that hides. The warning is just never in the same section as the instruction, and not one page changes the instruction because of it.
What the ten pages say
Two things the board above leaves out. All ten convert the ratio into days by dividing 365 by it, which is how the number reaches most people: not as turns per year but as a collection period. And nine of the ten insist on net credit sales on top. Only CreditPulse allows the looser one: "Total net revenue is an acceptable approximation if you cannot separate credit from cash sales cleanly."
Two of the ten name the two dates themselves as the thing that distorts the ratio, Billtrust and Tipalti. Six more raise the distortion in another register, as seasonality, as customer spread, or as a caution against reading a single point in time. Two say nothing about limits at all.
Tipalti is the one to read closely, because the diagnosis is in its second sentence. It opens with the usual line, that "any business model that is cyclical or subscription-based may also have a slightly skewed ratio", and then explains why: "the start and end points of the accounts receivable average can change quickly, affecting the ultimate accounts receivable balance". That is the whole problem, stated correctly, in a limitations box, three sections below a formula that it does not alter.
Sage does the same thing on a larger scale, and it is the only page of the ten that defines the honest version. Average receivables, it says, are "the average of all recorded AR balances throughout the analysis period". That is exactly what this article recommends. Then its own step-by-step tells you to "Add your beginning and ending accounts receivable balances and divide by two". Both sentences are on the page, and nothing reconciles them.
What it costs on a real file
The sample workbook that ships inside our own dashboard has 724 invoices across twelve months. Computing the month-end receivable balance the way the tool does, invoiced by that date less paid by that date, the book runs from $28,454 in the first month to $175,413 in the last. It is a business whose receivables pile up all year.
Net sales for the year are $550,425.
A denominator 18.8 per cent too large makes the ratio too small: 5.40 turns against a true 6.42. In days, which is how eight of these pages would have you read it, that is 68 days reported against 57 actual. Eleven days of collection performance, lost in the arithmetic.
The direction is not fixed either. Here the closing balance is the peak of the year, so the midpoint sits above the real average and the business looks slower than it is. On a book that peaks in the middle and settles by December, the same shortcut flatters. The reported number cannot tell you which case you are in, because both of the numbers it is built from are the same two numbers.
The case the article has to answer
There is a good reason this convention exists, and none of the ten pages says it out loud.
The opening and closing receivable balances are the two figures that appear in a published comparative balance sheet. An analyst comparing two companies has no access to either one's month-end ledger, so a two-point average is not a shortcut for them, it is the only average available. That is why the most academic page in the sample, Corporate Finance Institute, teaches it without limitations: its reader is holding a filing, not an export.
Which means the advice below comes with a cost. Every published benchmark band, the five to ten turns a year that these pages quote, was computed on two-point denominators from filings. Switch to a twelve-point average and your own number gets more honest while your comparison to those bands gets worse. On the workbook above, the shortcut says 5.40 and lands inside that band; the honest figure, 6.42, also lands inside it, but that is luck rather than a rule.
So there are two metrics wearing one name. One is for reading somebody else's filing. The other is for watching your own book. The pages that teach the first to readers doing the second are where this goes wrong.
What to do instead
Average the month ends you already have
If you are reading your own ledger rather than a filing, the twelve balances exist. Averaging twelve numbers is the same arithmetic as averaging two.
Keep the two-point version alongside it if you benchmark
The published bands were built that way, so comparing to them needs the same denominator they used. Report both and label which is which.
Look at the shape before you trust either
Plot the month-end balances. If they sit near the line joining the first and last, the two versions agree and the shortcut costs nothing. If they bow above or below it, the gap you saw above is what you are carrying.
Where this sits in my own tool
KPI Vitals does not compute this ratio. The word turnover appears nowhere in it.
What it does hold is one half of the answer: it builds a receivable balance for each month the file covers, for the card that charts cash owed month by month. Those are the balances an honest denominator needs, and getting them out of the tool today means hovering a sparkline point at a time, because there is no export.
The other half it does not hold. The numerator has to be net credit sales, and the file it reads has one amount column and no way to tell a trade sale from an invoice settled at the counter on the day it was issued. So the ratio is not free to add: the denominator is sitting there, the numerator is not, and any version I shipped without saying so would have the same fault this article is about.
The card the tool does have is days sales outstanding, and it is the reciprocal of this one computed on a snapshot balance. I wrote about everything loose in it, including the parts I got wrong, in the DSO piece. The monthly balances also drive the ageing report, and the clinical version of this same waiting time, with its own denominator argument, is days in A/R. The tool is KPI Vitals.
Sources and method
I opened thirteen pages across two searches on 8 September 2026. Three refused an automated fetch and are named rather than dropped: NetSuite, QuickBooks and Allianz Trade. They are also three of the largest publishers on this subject, so the sample leans away from exactly the pages most likely to discuss weighted averages, and the unanimity above should be read with that hole in it.
The ten are also not ten independent authorities. Accounting.Events reproduces Corporate Finance Institute's worked example and cites it as the source, so on the arithmetic they are one page counted twice.
Each page was asked the same questions before I read any of them: what it instructs, how it defines the average, what goes on top, whether it converts to days, and what it says about distortion. Counts: ten instruct the two-point average, one defines the twelve-point one, nine insist on net credit sales, ten convert to days, two name the endpoints, six warn in some other form, two say nothing.
The twelve balances are computed from the sample workbook that ships inside KPI Vitals, using the same month-end rule the product uses. An earlier draft of this article used a different series and described it as coming from that workbook; it did not. It came from the demo figures baked into the dashboard shell, and it understated the effect by a factor of about eight. The real file is in the numbers above.
Questions people actually ask
Net credit sales divided by average accounts receivable. All ten pages read here instruct you to build that average from the opening and closing balances, divided by two, and all ten convert the result to days by dividing 365 by it.
It is right for the job it came from. Opening and closing receivables are the two figures a published balance sheet gives you, so for anyone comparing companies from filings it is the only average available. For someone reading their own ledger, where every month end exists, it throws away ten of the twelve numbers.
On the sample workbook that ships inside our own dashboard, 724 invoices over twelve months, the two endpoints give an average of $101,934 against a true $85,794. That is 18.8 per cent too large, and it turns 57 days of actual collection into 68 reported.
The average of your month-end balances, if you are watching your own book. Keep the two-point version alongside it if you compare yourself to published bands, because those bands were computed from filings on two-point denominators.
The pages here quote roughly five to ten turns a year for many industries. The band is only meaningful against a denominator built the same way, which is the two-point one, so an honest twelve-point figure is better for you and worse for the comparison.
Reciprocals: 365 divided by the ratio gives the days, and every page in this sample prints that conversion. In practice the denominators differ, because turnover conventionally averages the balance and DSO conventionally does not.
Eight of the ten do, but never where the formula is. Two name the two dates themselves, Billtrust and Tipalti. Six raise it as seasonality, customer spread or a caution against single point in time readings. Not one of them changes the instruction it just gave.
WRITTEN BY
Olha · clinic data analyst
I build the reporting our managers open every morning at a multi-branch medical clinic — and package it so other practices don't have to start from scratch.
Published on 8 September 2026. Three limits worth stating outside the body text. The three pages that refused an automated fetch are three of the largest publishers on this subject, so the unanimity above has a hole in it exactly where it matters. Ten pages found by two search phrasings on one day is a sample, not a survey. And an earlier draft of this article measured the wrong file and understated its own finding by about eight times, which is said in the sources rather than quietly corrected.