Net Terms and What They Do to a Small Shop's Cash Position

Why this matters

A school district will pay you. A hospital will pay you. A city will pay you. Institutional receivables are among the most collectible in the trades, which is why shops relax about net terms on institutional work and why the failure, when it comes, is never a bad debt. It is a shop that could not make payroll while holding a receivable everyone agreed was good.

Net terms on institutional work are not a credit risk. They are a working capital requirement: a fixed amount of your own cost permanently parked inside the account, sized by how much of your capacity that account consumes and by how long the money takes to come back. That parked amount does not shrink when the account performs well. It grows when the account grows. Which means the account rewarding you with more work is the same event as the account demanding more of your cash, and a shop that does not see those as one event will say yes to the reward and get blindsided by the demand.

The gate

Here is the whole rule, and the rest of this article is two shops run against it.

Parked weeks = the share of your total operating capacity the account consumes, multiplied by the collection lag in weeks. Keep parked weeks below your cover, where cover is your cash reserve plus your undrawn line of credit, both expressed in weeks of total operating cost.

Both sides are in the same unit, weeks of total operating cost, which is what makes them comparable. The unit of analysis is the whole account, not a job, and it is measured at steady state rather than at a moment. This formulation assumes the account's cost intensity roughly matches your shop average; if the institutional work is materially heavier on purchased parts than your residential work, the parked figure is understated and you should weight the share by cost rather than by revenue.

Collection lag is not the stated terms. It is the number of weeks between the day your cost leaves and the day the cash lands. Count all four segments:

Net 45 is almost never 45 days of lag. A clean net-45 institutional account with weekly invoicing and a semi-monthly check run runs closer to 9 or 10 weeks end to end. Measure yours from your own records rather than reading it off the agreement.

Two shops, same terms, opposite outcomes

Both shops hold the same net-45 agreement with the same institution, both bill weekly against a standing order, both measure their real collection lag at 11 weeks, both run a cash reserve worth 6 weeks of operating cost and an undrawn line worth another 4, so both have 10 weeks of cover. Their margins are the same. The only difference is share.

Shop A runs the institution at 25 percent of capacity. Parked weeks: 0.25 times 11 equals 2.75 weeks of operating cost. Against 10 weeks of cover, that consumes 27.5 percent of the cover and leaves 7.25 weeks. Comfortable.

Shop B runs the institution at 60 percent of capacity. Parked weeks: 0.60 times 11 equals 6.6 weeks. Against the same 10 weeks of cover, that consumes 66 percent of it and leaves 3.4 weeks. Still solvent, still nobody late, and Shop B's owner reads the account as their best customer because it is: it pays reliably and it never argues about price.

Now apply one ordinary event. At the institution's fiscal year end the accounts payable department slows while it closes out old encumbrances, and effective lag stretches from 11 weeks to 15.3 weeks for about two months.

• Shop A: 0.25 times 15.3 equals 3.8 weeks parked, against 10 weeks of cover. Cover remaining, 6.2 weeks. Shop A does not notice.
• Shop B: 0.60 times 15.3 equals 9.2 weeks parked, against 10 weeks of cover. Cover remaining, 0.8 weeks. Shop B is now less than one week of operating cost away from the wall, has drawn its line to the limit, and cannot take on any other work, because every new job would add parked cost it has no room for.

Read what actually broke. Shop B's margin never moved. Its collections never failed. Its customer did nothing unusual - a year-end payables slowdown at a public institution is a scheduled event, not a crisis. What broke was that Shop B's parked position was already using two thirds of its cover at steady state, so a 39 percent stretch in lag consumed the remaining third. Shop A absorbed the identical stretch because it started with headroom.

The direction of that is worth stating both ways, because it is the sentence that gets remembered wrong. Higher share and longer lag both increase parked weeks. Lower share, or a shorter lag, decreases them: a shop at 25 percent share with a 6-week lag parks 1.5 weeks and could absorb the lag doubling and still hold 7 weeks of cover. There is no share so small that a long enough lag is safe, and no lag so short that an unlimited share is safe, because the two multiply.

The failure mode. Shop B does not discover this by reading a report. It discovers it on a Thursday when it looks at Friday's payroll. And by then the corrective levers are all bad ones: slow its own suppliers and lose terms, factor the receivable and give up margin, or stop selling into the account, which reduces the parked position only at the speed of the lag - eleven weeks of already-delivered work still has to come back before the position falls. That last point is the one shops miss. You cannot exit a parked position quickly. Stopping new work today leaves the parked cost fully in place for the length of the lag.

When the account is material-heavy, weight by cost and credit your supplier terms

The gate as stated uses the account's share of capacity, which is fine when the account looks like the rest of your work. When it does not, two corrections apply and they push in opposite directions, so running only one of them is worse than running neither.

Take a third shop whose institutional account is 35 percent of revenue but, because the work is unusually heavy on purchased equipment, 48 percent of delivered cost. The revenue-share shortcut gives 0.35 times 11, which is 3.85 parked weeks. Weighting by cost instead gives 0.48 times 11, which is 5.28 parked weeks, 37 percent higher. That correction alone is the argument for measuring share on cost.

Now the second correction, which runs the other way. Material on this account is 45 percent of its delivered cost and is bought on supplier net-30, so those dollars leave about 25 days after the reference point rather than at it, and are funded for roughly 7.4 weeks rather than the full 11. Labor and everything else stay at 11. The effective lag is 0.45 times 7.4 plus 0.55 times 11, which is 9.38 weeks, and parked weeks become 0.48 times 9.38, or 4.5.

So: weighting by cost raised the figure 37 percent, crediting supplier terms gave back about 15 percent of that, and the corrected answer sits about 17 percent above the revenue-share shortcut rather than 37 percent above it. A shop that applies the cost weighting and not the supplier credit overstates its exposure by enough to decline work it could carry. A shop that applies neither understates it by enough to accept work it cannot.

What actually shortens the lag

Two of the four segments are yours and two are not, and the two that are yours are usually the larger opportunity.

Invoice cadence is the biggest single lever and it costs nothing. A shop invoicing monthly holds an average of half a month of completed work before an invoice even exists, which adds roughly 2.2 weeks to the lag on top of everything else. Weekly invoicing against a standing order removes almost all of that. On the Shop B numbers above, cutting lag from 11 weeks to 8.8 weeks drops parked weeks from 6.6 to 5.3, which is 1.3 weeks of cover recovered for no cost and no concession.

Rejection rate is the second. Each rejection cycle is typically two to four days, and a scope-versus-encumbrance rejection is far worse. See related: What a Vendor Portal Wants and Why Invoices Come Back.

The stated term is the hardest to move and it is not the place to start. An institution's payment terms are usually set by policy or statute rather than negotiated per vendor, and asking for net 15 on a public entity's standard net 45 mostly signals that you do not understand who you are dealing with. Where there is room, it is more often in the check-run cadence or in enrolling for electronic payment than in the term itself.

What would change the answer. A shop with genuinely seasonal capacity can carry a higher share, because parked weeks measured against annual capacity understate the pressure in peak season and overstate it in the trough - in that case run the calculation on peak-quarter capacity rather than annual, which is a stricter test. And a shop whose cover is mostly an undrawn line rather than cash should discount the line, because a line is callable and a lender reading a concentrated receivable ledger is a lender who may reduce it exactly when it is needed.

How to verify you got this right

Pull twelve months of invoices for the account and compute lag directly: for each, the days from the last day cost was incurred to the day cash cleared. Take the median, not the average, because a few disputed invoices will distort a mean badly. Convert to weeks. Then take that account's share of your delivered cost, not its share of revenue, and multiply.

If the answer is more than half your cover, you have a concentration position rather than a customer, and the next thing to price is not the work. It is the pricing of the carry. See related: How to Price Work That Will Be Paid in Sixty Days.

References

• U.S. Small Business Administration, working capital and cash flow management guidance for small business
• See related: How to Price Work That Will Be Paid in Sixty Days, What a Vendor Portal Wants and Why Invoices Come Back, Cash vs Profit: Why They're Different, The Cash Reserve a Service Business Should Protect