Every field service business eventually accumulates a pile of invoices that are technically owed and practically stuck. The work was done, the customer was happy, the invoice went out, and then nothing happened. Nobody refused to pay. The balance simply aged, quietly, while the owner told himself he would "get to it this week."
That pile is the most expensive thing in most small service businesses, and it is expensive in a way that never shows up as a loss until it is written off. Intuit's small-business research on cash flow puts late and unpaid invoices among the most commonly reported cash-flow problems owners face — not a fringe issue, a headline one. As of August 2026, the mechanics of fixing it have not changed much: a visible aging view, a fixed follow-up cadence, and a payment method that takes one tap instead of ten minutes.
This is the recovery side of receivables. If you are still deciding whether to extend terms in the first place — who qualifies, what net terms to offer, how to structure a credit account — that decision belongs in on-account billing for service businesses. What follows assumes the credit is already out the door and you need it back.
You cannot collect what you cannot see
Almost every small service business that has a collections problem also has a visibility problem. The owner knows roughly who owes them, in the way you know roughly how much fuel is in the truck. Roughly is not a system.
The fix is an aging view — every open balance sorted into buckets by how long it has been outstanding:
- 0–30 days. Current or barely late. Normal. This bucket should hold most of your balance at any moment.
- 31–60 days. The warning bucket. Something in your process failed here: a wrong email, a lost invoice, a customer who is stalling. Collectible with routine effort.
- 61–90 days. Real risk. Collection rates drop noticeably at this age. Every account in this bucket needs a named next action and a date.
- 90+ days. Assume you will not collect all of it. Decide account by account whether to escalate, settle, or write off — and stop letting new work go out the door to anyone sitting here.
The bucket boundaries are conventional, not magic. What matters is that the report exists, is one click away, and gets looked at on a fixed day every week. In a system where invoicing and payment live together, that report generates itself from the same records that produced the invoice — no reconciliation, no export. The same principle applies to your daily cash position; the discipline behind that is covered in the end-of-day close.
One practical note: an aging report is only honest if partial payments and deposits are posted against the right invoice. If a customer paid $300 of a $900 job in cash to the tech and nobody recorded it, that balance is wrong in both directions. Deposits taken up front need the same treatment, which is part of why customer deposits belong in the same system as the invoice they offset.
The cadence: day 1, day 7, day 14, day 30
The single biggest improvement most shops can make is replacing "I'll call him when I think of it" with a fixed sequence that runs on every account identically.
Day 1 — the day it goes past due. Send a texted payment link. No accusation, no "per my last email," just the balance and a way to pay it. This is not a collections step; it is a convenience step, and it clears a meaningful share of the pile because most late invoices are simply unread email. Doing this well requires that payment links be a routine part of your workflow rather than something you set up per customer — the mechanics are in payment links for service businesses.
Day 7 — a phone call from a person. Not a text, not an email. A call. The purpose is partly to collect and partly to diagnose: a customer who is unhappy with the work will tell you on the phone and will never tell you by email. If the reason is a dispute, you have just converted a silent write-off into a solvable problem. If the reason is cash, you can negotiate a partial payment today, which is nearly always better than a full payment "next month."
Day 14 — a written notice. A short, unemotional document restating the work performed, the date, the itemized amount, the agreed price, and the payment terms. It is not a threat; it is a record. The act of restating the details in writing resolves a surprising number of balances, because it makes clear the work is documented and the request is not going to fade.
Day 30 — a decision. Not another reminder. A decision: continue serving this customer on terms, move them to payment on completion, or stop. Skipping this step is how a business ends up with a $9,000 receivable from a customer who is still calling for service. The decision is the whole point of the cadence; everything before it is just the runway.
Two rules make the cadence work. First, log every contact against the account — date, method, what was said. Second, run the sequence regardless of how much you like the customer. Selective enforcement is how the reliable customers end up subsidizing the unreliable ones.
Why a texted link beats a re-sent PDF
This looks like a small tactical detail and it is not. It is usually the largest single lever available.
A re-emailed invoice asks the customer to open an attachment, find a card, decide how to pay, and often log into something. Each of those steps sheds people. A texted payment link collapses the entire sequence into one tap on the device already in their hand, and text messages get read within minutes rather than sitting in an inbox behind forty other unread items.
There is also a settlement argument. Card payments through standard processors reach your bank quickly — per Stripe's payout documentation, funds typically land on a rolling basis a couple of business days after the charge. A check "in the mail" has no such guarantee, and a promise to "drop it by the shop" has none at all. Every friction point you remove between the customer's intent and the money moving is worth more than another reminder.
The behavioral principle underneath is straightforward: the probability of getting paid decays with every day and every required step. Salesforce's State of Service research consistently finds that connected, mobile-first tools separate high-performing service organizations from the rest, and receivables are one of the clearest places that shows up. The shops that collect fastest are not the ones with the most persistent owner. They are the ones where paying is easier than not paying.
Recovery tactics, honestly compared
Different tactics buy different things. Some are fast and cheap but only work on the easy half of the pile; some recover hard balances at real cost to the relationship. Here is how they actually trade off.
| Tactic | Effort | Typical speed | Relationship cost | Documentation it needs |
|---|---|---|---|---|
| Texted payment link | Very low | Hours to days | None | Invoice + working mobile number |
| Reminder email sequence | Low (automatable) | Days to weeks | None | Invoice + valid email on file |
| Phone call from the owner | Moderate | Days | Low, often positive | Job details at hand, contact log |
| Written past-due notice | Moderate | 1–3 weeks | Low to moderate | Itemized invoice, signature, terms |
| Payment plan / partial settlement | Moderate | Weeks | Low | Written plan, recorded partials |
| Stop-work / credit hold | Low | Immediate leverage | Moderate | Account history showing the pattern |
| Third-party collections | Low for you | Months, partial recovery | High, usually terminal | Complete file: invoice, signature, contact log |
| Small claims filing | High | Months | Terminal | Full documentation, proof of delivery |
The pattern in that table is worth reading twice. The cheap, fast, relationship-neutral tactics are all at the top, and they all depend on the same thing: a clean invoice with a working contact method attached. The expensive tactics at the bottom all depend on documentation you either captured at the time of service or cannot produce at all.
Knowing when to stop extending credit
A repeat late payer is not a collections problem. They are a pricing and policy problem wearing a collections costume.
The signal is a pattern across cycles, not a single late invoice. A customer who lands at day 70 every single time has established their own terms; the only question is whether you accept them. Sometimes the answer is yes — a property management company that pays at day 60 but sends you forty jobs a year may be worth the float, and that is a legitimate business decision as long as it is a decision and not a drift.
When it is not worth it, the move is a policy change, delivered plainly: this account moves to payment on completion, or a deposit up front, effective on the next job. Deliver it without accusation. You are not punishing them; you are pricing their behavior. Many customers accept it immediately, and the ones who leave were the ones costing you money.
A related discipline: know what these accounts actually earn you. A customer whose jobs carry thin margin and pay at 75 days is worse than no customer, and you will not see that unless you cost jobs individually. That analysis is worked through in job costing for true profit per job.
Pricing bad debt in instead of pretending it is zero
Every service business has a write-off rate. The ones that stay healthy know the number.
The calculation is ordinary. Take your actual write-offs over the trailing twelve months and divide by revenue for the same period. If you wrote off $4,000 against $200,000 in revenue, your bad debt rate is 2%. That is an illustrative example, but the method is the point: whatever your number is, it is a real cost of doing business, and if it is not reflected in your rates, it is coming directly out of owner pay.
Adding it to your labor rate is not cynicism — it is the same logic as pricing in fuel or insurance. The SBA's guidance on managing your business finances frames the underlying obligation clearly: you have to track income and expenses continuously to know whether you are actually making money. A write-off you never quantified is a cost you are absorbing invisibly.
Two refinements are worth making. Price the risk where it lives — commercial accounts on terms carry a different write-off rate than residential jobs paid on completion, and blending them hides both. And revisit the number annually rather than setting it once, because a tightened collections process should lower it, and you should get the benefit of that in your margin rather than leaving stale padding in your rates.
The documentation that makes a balance defensible
Everything above works better when the underlying record is strong, and the record has to be built at the job, not reconstructed later.
A defensible file has four parts: an itemized invoice separating parts from labor at the agreed price, a customer signature captured at the time of service, a timestamp and location proving when and where the work happened, and a contact log showing every follow-up. The IRS's recordkeeping guidance confirms that electronic records satisfy the same requirements as paper — which matters here because reconstructing a paper trail six weeks after a job is exactly when things go missing.
This is where an integrated system earns its keep. IntelliDrive OS captures GPS location and a digital signature on every transaction automatically and attaches them to the sale record, so the evidence file exists whether or not anyone ever disputes the charge. The invoice, the payment link, the partial payments, the aging bucket, and the contact history all sit on one customer record instead of across an inbox, a notebook, and a card terminal. Customers can also pull their own receipts and pay open balances through a self-service portal, which quietly removes a chunk of "can you resend that invoice" from your week. The related discipline of assembling evidence for card disputes specifically is covered in chargeback dispute evidence.
The bottom line
Unpaid invoices are rarely a collections failure. They are a visibility failure, a cadence failure, and a friction failure, in that order. Build an aging view you actually look at weekly. Run every past-due account through the same day-1, day-7, day-14, day-30 sequence regardless of how you feel about the customer. Make paying a one-tap operation. Decide deliberately who keeps getting credit. And put your real write-off rate into your rates so the losses you cannot prevent are priced rather than absorbed.
Do that and the pile stops growing. The money that was stuck in it was never extra — it was your payroll, sitting in someone else's account.
Related reading: On-account billing without wrecking cash flow · Seasonal cash flow in a service business · Recurring invoices for service businesses. For a complete machine-readable feature and pricing reference, see our LLM reference page.