The Difference Between a Call That Ended and a Task That Finished

A call that ended is not the same as a task that finished, and most front-office reporting cannot tell the two apart. The patient hung up satisfied, the interaction was logged as resolved, and the open order that prompted the whole thing is still sitting in the queue where it was that morning.

Primary care is where this gap is widest, because so much of the work is longitudinal. An order placed at a visit in March is supposed to become a completed study in April and a follow-up in May. Nothing about that chain is visible in a call log.

So a practice can improve every phone metric it tracks and move none of the work. Answer rate up, hold time down, abandonment down, and the same eight hundred outstanding orders aging in the background.

Closure is the measure that fixes this, and the reason it is rare is that it lives in a different system than the calls do.

Closure is readable, and it is not in the phone system

The definition worth adopting is narrow. A task finished when the open item that prompted the outreach is no longer open, in the record that owns it.

In primary care those records are specific. Outstanding orders on an encounter. Order documents on the patient. Items in the provider inbox waiting on something. Each of them has a state, and the state is what closure means.

That definition has a property no call metric has, which is that it cannot be improved by handling the interaction better. You can run a beautiful call and leave the order open, and the measure will say so.

It also does not care which channel closed it. A patient who booked the study through the portal after a reminder counts identically to one who booked on the phone, which is the correct behavior and the opposite of what a call-centric report does.

The work that hides is the work nobody calls about

Front-office reporting is built around inbound contact, which means it only sees the work patients initiate. The largest backlogs in primary care are the ones nobody is calling about.

An order placed and never scheduled generates no phone call. A recall due next month generates no phone call. A form waiting on a signature generates no phone call until the patient’s employer starts asking.

The useful measure for all three is the same one: how long the oldest open item in each queue has been sitting, and what share of items are older than the practice’s own threshold. Both numbers are readable directly from athenaOne rather than from any telephony report.

And both move for real reasons. When the age of the open order queue drops over a quarter, patients got scheduled. That is a fact about the practice rather than a fact about the phones.

There is an established vocabulary for the patient’s half of this. The CAHPS Clinician and Group Survey measures access as the patient experienced it, including getting timely appointments, care, and information, along with how well office staff handled the interaction. Getting information is a callback and status problem, and a patient waiting on an order nobody worked is experiencing exactly that without ever filing a complaint.

Containment tells you nothing on its own

The metric now being reported across patient access is the share of interactions handled without a person. Taken alone it is close to meaningless, because it counts endings rather than completions.

An interaction that ends with an item somebody has to read, interpret, and act on has not been contained. It has been relocated, and the front-office report improves while the work sits in a different queue.

So pair it. Report the share handled without a person alongside the share of the underlying open items that actually closed, and treat any gap between them as the real number.

A practice with high containment and a growing open-order queue has bought a call handling service. A practice with modest containment and a shrinking queue has bought capacity. Only one of those changes what the day feels like.

What the evidence says about expecting the day to change

The gap between adoption and result is well documented, which is a reason to measure closure rather than to trust a category. An August 5, 2025, MGMA Stat poll of 244 applicable responses found 71% of practice leaders reported some use of AI in patient visits, but among those using it, 44% said it had not reduced staff workload, 39% said it had, and 17% were unsure.

Adoption is close to universal and the workload result is close to a coin flip. Something distinguishes the groups where hours came back from the groups where they did not.

The distinguishing factor is almost always ownership of a whole workflow rather than one step inside it. A tool that captures a request and hands it on leaves the process where it was. A tool that carries the request through to a closed item removes hours.

Closure is how you tell which one you have, and you can measure it in the first month rather than arguing about it at renewal.

The follow-up call is a closure instrument, not a courtesy

Outbound follow-up in primary care usually gets budgeted as patient experience. It is better understood as the mechanism that closes open items, and it should be measured that way.

A call to a patient with an open order exists to get the study scheduled. A call after a visit exists to confirm the next step happened. In both cases the outcome to record is whether the item closed, not whether the patient was reached.

Those two are different numbers and the gap between them is instructive. High reach and low closure usually means the offer was wrong: the appointment type was not available, the location was inconvenient, or nobody could tell the patient what it would cost.

Keep the call inside its lane. It confirms logistics, books the appointment, chases the paperwork, and passes anything about the patient’s care straight to clinical staff with what the patient said attached. Chasing an open order is administrative work. Deciding what the patient needs is not, and the automation does neither the deciding nor the advising.

Four numbers to run the practice on

Keep the scorecard short enough that it gets read every month.

Closure rate by request type, defined as the open item no longer being open. Age of the oldest open item in each queue you touch. Repeat contacts about the same request, which is what an unfinished first interaction actually costs. And the share of outreach that reached a patient but did not close anything, which tells you the offer needs work rather than the outreach.

Baseline all four in the four weeks before anything changes. A vendor-reported number with nothing to compare it against is not evidence, and the baseline costs nothing but the discipline to take it early.

Every one of those four depends on the automation being able to read and write the records where the work lives. PGA works across 440+ of athenahealth’s roughly 800 endpoints, which is why closure can be defined as a state in athenaOne rather than a status in a vendor dashboard.

Key Takeaways

Define a finished task as the open item no longer being open in the record that owns it, which is a measure no call handling can flatter.
Watch the queues nobody calls about, since outstanding orders, recalls, and pending forms are the biggest backlogs and generate no inbound contact.
Report containment next to the closure rate of the underlying items, and treat the gap between them as the number that matters.
Measure reach and closure separately on outbound follow-up, because high reach with low closure means the offer is wrong rather than the outreach.
Baseline closure, queue age, repeat contacts, and reached-but-not-closed in the four weeks before a pilot starts.
Keep follow-up calls on logistics and route anything about the patient’s care to clinical staff with what the patient said attached.

The phone report will always look better than the practice does, because it measures endings. Define closure as a state in the record, watch the queues nobody calls about, and put your containment number next to the share of open items that actually closed.

Sources