Think of authorized units as inventory. A payer approves, say, 520 units of 97153 over six months, that's your stock. Every session draws it down. Deliver a unit you don't have, and you've given away clinical labor for free: the payer won't pay for services beyond the auth, and you usually can't bill the family for the overage either. It's the purest form of revenue leak in ABA, because the cost is incurred with certainty and the revenue is zero.
The leak has a mirror image that's just as expensive: units that expire unused. An authorization that ends with 60 units on the table represents care a child was approved for and didn't receive, and revenue your practice earned the right to and didn't collect. Under-delivery usually isn't a clinical decision; it's a scheduling drift nobody noticed until the auth period closed.
For every active authorization, three quantities need to reconcile continuously: units authorized (what the payer approved, by code, with start and end dates), units scheduled (what's on the calendar between now and the auth end date), and units delivered (what's actually happened, from verified sessions). Healthy practices watch the relationships: delivered plus scheduled should land just under authorized by the end date. If delivered-plus-scheduled exceeds authorized, you're on track to give away free sessions. If it falls well short, you're on track to strand units, time to add sessions or document why not.
Auth tracking in a spreadsheet fails not because the math is hard but because the inputs move daily. Cancellations return units; make-up sessions consume them ahead of plan; a code change mid-auth splits the count; a renewal overlaps the old period by two weeks. Each event requires a human to remember the spreadsheet exists. Meanwhile the scheduler, who actually controls the burn rate, often can't see the balance at all, so they book what the family asks for and hope.
The balance must live where scheduling happens. When a scheduler books a session that would push scheduled-plus-delivered past the authorized total, they should be warned at booking time, not discovered by a biller at claim time. Burn-rate alerts should fire when an auth is depleting faster or slower than its calendar pace, and renewals should trigger sixty days out, since payer turnaround on reauthorization is measured in weeks. In ABAVault™ this is native: authorizations track against both schedule and delivery, the scheduling screen enforces the ceiling, the dashboard flags auths at risk, and billing validation refuses to build a claim against an exhausted or expired auth, the last line of defense, not the first.
Even with automation, one habit pays for itself: a weekly fifteen-minute review of every auth ending in the next sixty days, asking two questions, will we land within ten units of the total, and has the renewal been submitted? Practices that keep this ritual report the same thing: end-of-auth surprises simply stop happening, in both directions.