Mach7 Blog | From Pixel to Point of Care

You're only as good as your longest delay: what teleradiology teaches about turnaround time

Written by Sean Kowaliw | November 19, 2026

If you buy reading services, or run them, the number everyone watches is average turnaround time. The number that decides whether you keep the contract is a different one: the slowest exam of the night. I learned that running a teleradiology operation, and it changed how I think about every imaging workflow since.

Why the slowest exam is the one that counts

For several years I ran client services, and then clinical and technical systems, at a teleradiology provider. On a typical night that meant roughly a thousand exams, around sixty hospital and imaging center customers, every night of the year.

Here is what that teaches you quickly. Nobody at a customer hospital remembers the four hundred studies that came back on time. They remember the one stat head CT that sat in a queue while their emergency department waited. A month of good numbers can be undone by one bad night, because the person on the phone the next morning is not looking at your average. They are looking at their worst moment.

That is why I say a reading service is only as good as its longest delay.

The clock only moves one way

When I started, our target was 30 minutes or less per exam. We worked that down to an average of 22 minutes. However, the faster you deliver, the faster customers expect the next exam. Every improvement becomes the new floor, and nobody gives you credit for the old number.

There is a second pressure that people outside the business rarely see. With most teleradiology contracts we had, nothing obliged a customer to send their overnight volume to you. A hospital could decide on Tuesday morning that you had not performed, and by Tuesday night the studies were going somewhere else. You are the vendor who can be replaced tomorrow, and you run the operation that way.

I say that with affection, not complaint. That pressure is exactly what made the good teleradiology groups so disciplined, and it is why I think anyone running imaging operations can learn from them.

Where the delay actually hides

When we dug into our slow exams, most of the lost time was not reading time. It happened before a radiologist ever opened the study.

The pattern looked like this. An order would arrive for a stat study, and the images would not. Or the images would arrive and the relevant priors would not. Or the history on the order did not match what the radiologist needed to read it safely. Each of those sends a person hunting: calling the site, checking what was sent, waiting for a resend. On a busy night, I would estimate that chasing studies that had been ordered but had not actually arrived accounted for something like a quarter of the volume.

That is an interoperability problem, but nobody on the overnight shift experiences it that way. They experience it as a clock running on an exam they cannot read yet. Our clock started when we had all the information. The customer's clock started when they ordered the exam.

Measure the tail, not the average

If the slowest exams are what customers remember, they are also what you should be studying. An end-to-end average blends every delay together and hides where the time went.

What worked for us was breaking each exam into segments and watching each one: order received, images complete, study available to the reader, read finished, report delivered. Once you can see the segments, the slow exams stop being mysteries. You can say that a given night's worst cases were three missing-image chases and one study that waited for a reader with the right subspecialty, and those are different fixes.

We built our own turnaround reporting to do this, because nothing we could buy at the time matched the way the business actually ran. It took about three years to get it where we wanted it. I would not recommend that path to everyone, but I would recommend the habit behind it: every morning, look at the slowest handful of exams from the night before and write down why each one was slow. Within a few weeks, the reasons repeat, and the repeats are your priority list.

The people who absorb the delay

Every one of those delays lands on someone before it reaches a radiologist. In teleradiology that is the workflow coordinator: the person matching orders to images, chasing missing priors, confirming who needs a call and keeping the reading queue honest.

The coordinators I worked with were always stretched thin, yet I saw no way to replace them: radiologists had too many other responsibilities to take on that work. When people talk about automating that role away, I would ask a different question: what would it take for the coordinator to see the problem ten minutes sooner? Visibility, not replacement, is what bought us time.

Where to start

If you purchase reading services, ask your provider three questions. How long did your slowest exams take last month, and why? How do you classify the cause of a delay? What happens, step by step, when an order arrives without its images? A provider that can answer those clearly is managing the tail. One that can only quote an average may not be.

If you run reading services, inside a hospital or as a group, start with the morning review. Pull the slowest exams from the night, write one line on each, and look for the repeats. You will learn more from ten bad exams than from ten thousand good ones.

And if you are one of the people who keeps an overnight operation running, thank you. At two in the morning, the reader on your schedule may be the only radiologist that emergency department has, and the work you do before the study opens is part of the care.

If your team reads across sites and is working through the same questions, you can see how Mach7 works with teleradiology groups.

Mach7 Technologies builds vendor-neutral enterprise imaging: the Vendor Neutral Archive, the Communication Workflow Engine, the eUnity enterprise viewer, AI enablement, and Flamingo, the next Mach7 product set.