When not to automate your chasing

Most software companies won't publish the conditions under which their product is the wrong choice. That's a shame, because the fastest way to find out you've bought too early is to buy, and then discover the tool solved a problem you didn't have.

So here is the honest version for automated chasing, and for ours specifically.

The test: can you personally ring everyone who's late?

That's the whole thing. Not revenue, not client count, not how modern your stack is. If you can personally ring every overdue customer, and you actually do, you don't need an AI making those calls. You will do it better, because you know the history and the relationship.

At fifteen clients that's usually true. You know every debtor's story, and "chasing" is a handful of calls you can fit into a Thursday afternoon. At sixty it usually isn't, and the honest description of what happens is not that the calls get made worse. It's that the bottom half of the list doesn't get rung at all. Not by you, not by anyone, not by a reminder that stopped at email five.

Four situations where we'd tell you to wait

1. Your book is small and you're on top of it. Covered above. Come back when you're not, and you'll know when that is because the aged report will stop matching what you've actually chased.
2. Your overdue column is mostly disputed or contractually withheld. If a large share of the ledger is progress claims, contested certifications or retention, the chaseable slice may be small enough that this isn't your bottleneck. Do the split first, then decide.
3. Nobody owns the outcome. Automated chasing produces exceptions: disputes, promises, callback requests. If there's no person who'll look at those within a day or two, the queue becomes another dashboard nobody opens, and you've bought a job rather than removed one.
4. Your invoices or contact data are unreliable. If phone numbers are stale or invoices routinely go to the wrong entity, fix that first. Automated contact scales whatever you point it at, including the errors.

What genuinely does change at scale

Above the line, three things break down at once, and they're all about consistency rather than effort.

• Chasing becomes uneven. It's the most deferrable task in any business. It's never urgent at 4pm on a Tuesday and nobody chases you for skipping it. So it happens in a burst when cash gets tight, by which point the invoices are 90 days old and the conversation is much harder than it would have been at day 30.
• Promises stop being tracked. Ten live promises to pay is a spreadsheet. Fifty is a memory test that nobody passes, and an untracked promise is just a deferral.
• The awkward ones get quietly written off. Not decided on. Avoided, then aged out. That's the most expensive habit in receivables and it's entirely invisible in the accounts.

You can keep the calls you want

If the objection is about specific relationships rather than volume, that's a settings question, not a reason to wait. Calling can be switched off for named customers while they still receive email and SMS, so the twelve relationships you'd never hand to software stay entirely yours and the other two hundred get worked. And when a call does hit a dispute or someone under real financial stress, the right outcome is a handoff to a human, not a negotiation. That's escalation working as designed, not a failure.

If you're a practice trying to work out which side of the line your book falls on, the savings calculator is built to answer exactly that, including the case where the answer is no. Put your own numbers in. If it doesn't clear the bar, it doesn't, and you'll have found out in two minutes rather than two meetings.

See it on your own Xero

Connect Xero and hear the AI teammate chase a quote and recover an overdue invoice by voice.

Keep reading