Loading
By writing in your own voice rather than a collections voice, making paying trivially easy, and stopping the sequence the moment anything is disputed.
Published December 16, 2025
Chasing money from customers you want to keep is genuinely uncomfortable, which is why it is the task most often postponed. The discomfort is also informative: it tells you that the tone of the message matters as much as the fact of sending it.
The mental model that makes this easier is that the overdue list is mostly administrative failure rather than unwillingness to pay. A card expired. An invoice went to somebody who left. A reminder arrived during a busy week and got buried. These get paid the moment someone asks clearly and makes it easy.
A genuine refusal to pay is a different conversation and it should reach a person quickly. Treating the whole list as though it were made of refusals produces exactly the aggressive tone that damages relationships, and it aims that tone mostly at people who simply forgot.
The practical test is whether the message would embarrass you if the customer forwarded it to somebody. A reminder that states the invoice number, the amount, the date it was due and a link to pay passes that test. Anything with escalating language or implied consequences usually does not, and it buys you very little on invoices that were going to be paid anyway.
Every step between the reminder and the payment loses people. Find the invoice, log into the portal, reset the forgotten password, find the card. Each of those is a place where someone who fully intended to pay stops and means to come back later.
A link that opens a hosted checkout with the amount already filled in removes all of it. This is unglamorous and it is the single highest-leverage change available to most accounts receivable processes.
The second one is worth dwelling on. Continuing to chase somebody who has told you when they will pay is the fastest way to turn a cooperative customer into an uncooperative one, and it happens constantly in sequences that only listen for the word 'paid'.
Where you operate under rules about when and how often you may contact somebody about money, those limits go into the configuration before the first message goes out. They are not guidance for the model to weigh up. This applies regardless of whether the rules come from regulation, from a contract, or from your own standards.
What was sent, when, on which channel, and what came back. If a customer ever says they were contacted too aggressively, the record either supports you or it does not, and reconstructing it afterwards from an inbox is not a position you want to be in.
The side benefit is that the trail makes the sequence improvable. When you can see which message in the sequence actually gets responses, you can shorten the ones that do nothing, which is better for your customers and for your cash position at the same time.
On the site
Keep reading