Nemt KPIs and the details that decide it

What to take away

Owners usually track trips and revenue, both of which can rise while the business gets worse. What is needed is a small set of measures with denominators that cannot be gamed by taking on more of the wrong work.

The eight

Trips delivered before the appointment time, as a share of trips with a fixed appointment. The single most important measure in this trade, because a patient delivered after their slot has been failed regardless of when they were collected. What it hides: nothing much, which is why it comes first.

Paid hours as a share of staffed hours. Your utilization. What it hides: it improves if you cut staffed hours below what the service needs, which shows up later as missed trips.

Occupied minutes per trip, by mode and assistance level. The number your schedule and your prices are built on. What it hides: an average conceals a few very long loads, so keep the spread.

Empty miles as a share of total miles. Deadhead, which is created by which trips you accept. What it hides: it falls when volume rises for reasons unrelated to routing, so read it alongside trip counts.

Late cancellations and no-shows, as a share of booked trips, by account. Capacity you paid for and were not paid for. What it hides: an account with a low rate but very late notice can cost more than one with a high rate and early notice, so track the notice period too.

Days from trip completed to money received, by payer. The measure that predicts a cash problem while there is still time. What it hides: an average across payers. Never aggregate this one.

Complaints by cause, not by account. What it hides: a cause that appears once at several accounts looks trivial and is usually systemic.

Driver turnover, and separately the number of drivers in their first eight weeks. The second is the leading indicator, because a crew heavy with new drivers runs slower loads and later trips. What it hides: turnover looks fine in a month where somebody has resigned but not yet left.

The denominators that ruin a measure

Cost per mile is the most seductive of these and the least useful. In a trade where a twenty-minute load produces no mileage at all, cost per mile improves every time you drive further, which is the opposite of what you want.

How to take the readings

Monthly, on the same day, from the same source, with last month beside it. The comparison that matters is your own trend, not a number from somewhere else.

Read them together rather than separately. Punctuality falling while first-eight-week driver numbers rise is a training and scheduling story. Utilization rising while complaints rise is a schedule that has been tightened past what the loads allow. Individually each of those looks like a small problem; together they are a diagnosis.

Where the readings come from

Times at both ends, mileage loaded and total, notice periods on cancellations, and remittance dates. All of it has to come out of the ordinary day, because a measure assembled specially at month end will be assembled less carefully each month until it stops.

Whether your system produces these is the practical question behind the finance and dispatch guide, and how to test a candidate against your own awkward trips is set out in the material on evaluating systems for this trade. If nothing you have can produce them, a month of manual tallying will, and it is worth doing once.

Keeping documents that support what was billed is the general expectation described by the Internal Revenue Service, and the same records feed both the invoices and these measures. Your payer's own requirements are stricter and come from the bodies mapped in the licensing and compliance guide.

Keep the reports where they can be trusted

These numbers describe passengers, addresses and appointment patterns, so treat the reporting the same way you treat the records themselves: limited access, no shared logins, a backup that has been tested. The NIST small business quick-start guides and the CISA resources for small and medium businesses are written for this size of operation and cover exactly that ground.

What to do with each reading

A measure that never leads to an action should be dropped. For each of the eight, decide in advance what a bad month would make you do.

Punctuality falling means checking load allowances before checking drivers. Utilization falling means looking at deadhead and at schedule gaps. Payment days lengthening for one payer means a call to that payer this week, not at quarter end. Turnover rising means a conversation about hours and equipment before one about pay, which is the argument in the hiring and training guide.

Common questions

How many measures is too many?

More than you review. Eight is already at the limit for most small operations; if you cannot get through them in half an hour, cut the ones you have never acted on.

Should drivers see these numbers?

The operational ones, yes, presented as facts about the schedule rather than as a scoreboard. Punctuality shown as evidence that a load allowance is wrong is useful; the same number shown as a personal ranking teaches people to hurry at the door.

What if a measure moves and I cannot explain it?

Look at the composition before the cause. Most unexplained moves in a small operation are one large account changing its pattern.

How long before the trend means anything?

Three readings. One month is noise, two is a line, three is a direction.