Practical notes for FixPayment readers

Start with the cash-flow math, not the emotion

A 90-day-old invoice isn't a moral failing. It's a line item with a probability attached. Before you escalate anything, write down four numbers: face value, your fully loaded cost of goods or service, your internal cost to chase it (hours times loaded labor), and your realistic recovery rate at each stage.

Here's the arithmetic that matters. A $10,000 invoice at a 40% traditional collection-agency fee returns $6,000 gross to you, minus any internal hours you already spent. The same invoice resolved at an 18% success fee returns $8,200. That $2,200 gap is roughly the margin on a mid-size job. Multiply it across a year of aging AR and the fee structure stops being a rounding error — it becomes a working-capital decision.

Most small and mid-size B2B teams don't have a recovery problem so much as a sequencing problem. They send reminders too long, escalate too late, and pay too much for the last mile. Fixing the sequence usually beats fixing the vendor.

Segment before you chase

Not every overdue account deserves the same treatment. Split the ledger roughly three ways:

• Can pay, will pay, just slow. These respond to a clean reminder with a payment link and a due date. Automate them. No human needed.
• Can pay, won't pay without pressure. These need structured follow-up: a call, a payment plan, a documented paper trail. Human judgment matters here.
• Can't pay right now. These are negotiation cases — partial payments, settlements, extended terms. Chasing them with threats wastes everyone's time and can create legal exposure.

The mistake is treating all three the same. A blanket dunning sequence either annoys the first group or gets ignored by the second. Segmentation is where the operational leverage lives.

Why AI plus human beats either alone

AI is good at the boring parts: consistent touch cadence, reading payment signals, flagging accounts that changed behavior, drafting dispute documentation, and keeping a timestamped record. Humans are good at the parts that need judgment: reading a stall for what it is, negotiating a plan, and knowing when to stop.

In practice, an AI layer handles the first three or four touches and routes only genuine exceptions to a person. That cuts your internal chasing hours and keeps the file clean if it ever needs a formal demand. It also means your team spends its time on accounts worth $8,000 instead of $800.

If you want the mechanics of the fee model, our page on unpaid invoice recovery while keeping roughly 80% walks through how a success-fee structure works when there's no recovery.

Documentation is your real leverage

When an invoice goes sideways, the creditor with the cleanest file wins the argument. That means: signed agreement or accepted purchase order, delivery confirmation, the invoice itself, and a dated log of every reminder and response. Screenshots of a chat thread are better than nothing, but a structured dispute file is better.

Two habits pay off. First, capture acceptance at the point of sale — a signed quote, a PO number, an email that says "go ahead." Second, log every touch in one place. If a payment plan or settlement is agreed, write down the amount, the dates, and what happens if a payment is missed. Vague deals fall apart.

None of this is glamorous. It's just the difference between a recoverable file and a write-off.

Underwrite the account before it goes bad

Recovery is cheaper when you screen earlier. Traditional bureau scores tell you about a business's general credit history, but they're blunt instruments for B2B trade credit. They don't know your industry, your margin, or how a specific buyer behaves on net-30 terms.

A risk score built on payment behavior — how fast a company actually pays its suppliers, whether it stretches terms, how it responds to reminders — is more useful for setting credit limits. That's the idea behind the FP Risk Score: underwrite the counterparty against real payment patterns, not just a bureau number. Set a limit you can survive, then hold it.

Build the workflow once, then leave it alone

Good AR operations run on a fixed cadence: invoice on time, reminder at day 1 past due, follow-up at day 7, a call at day 15, escalation at day 30. Adjust the timing to your industry, but don't improvise it invoice by invoice. Predictability is what keeps a file clean and a customer relationship intact.

If you're wiring this into your own systems, most of the work is plumbing — statuses, webhooks, payment links. That's what our creditor MCP and API tooling is for. If you'd rather see the whole picture first, start at the FixPayment home page for B2B AR software.

One caution: keep the tone professional. Pressure works on people who intend to pay and are embarrassed. Threats work on almost no one and create risk. The goal is resolution, not a fight.