The ROI of Cardiology Practice KPIs, Location by Location

Most cardiology practice KPIs answer a question nobody asked. The packet shows call volume, speed to answer, and one group-wide no-show rate. Meanwhile the manager of the office that just lost two weeks of new-patient access sits through the whole review. That office never shows up in the numbers.

A cardiology group with several sites does not buy automation to answer the phone. It buys automation to move access and capacity. Then it has to prove to a board that both moved. That second job is where most reporting falls apart.

The easy numbers come from the phone system. Calls offered. Calls answered. Average handle time. The numbers that say whether a site is healthy live on the schedule side, and athenaOne does not roll those up the way an operator needs. Slot use by visit type. Provider-hour density across new, follow-up and procedure visits. Booking and fill rate by site. Wait time to a visit by type and by provider. Referral conversion. Open slots against pending outreach.

Those are practice reports. Most groups rebuild them by hand in a spreadsheet each quarter, if at all. So the real question, what happened at each site and for each provider, comes back answered with a group average. A group average is the one number shape that cannot answer it.

A group average is not a measure

A July 14, 2026 MGMA Stat poll asked medical groups about new-patient wait times. 46% said waits held flat year to date. 28% said waits got longer. 22% said shorter. The poll had 197 applicable responses.

The detail under the topline is the useful part. Groups that hired physicians and advanced practice providers landed in all three outcome groups. Some practices changed nothing and ended up where the hiring groups ended up. The actions were close to identical. The results were not.

That is what a group average does to a multi-site cardiology practice. Two sites move in opposite directions. The mean holds. The report says nothing happened. MGMA’s own advice is to segment the measure, because one average wait hides variation by specialty, site, provider, payer, referral source and visit reason.

For a group running several offices under one athenaOne account, that split is the whole exercise. The unit is a department and a provider. It is never the practice.

Slot use by visit type is the number athenaOne will not hand you

Ask how well your imaging slots get used. The answer will be quietly wrong. Most schedules are not built out of specific visit types at all. They are built out of generic template slots, the Any 15 and the Any 30. Search for a specific type and the EHR hands back a generic one. An Any 15 may need to be a full hour for a new consult. Which specific types each generic slot can hold, per provider and per department, is the map under every slot-use report you have ever run.

MGMA describes the same failure from the other side. A template carved for one plan and one visit reason cannot be filled by a patient who does not match. The slot goes unused while patients wait. In a poll with 823 applicable responses, 23% of leaders rated their group’s patient access process as low.

Resolving that map is the first real piece of work. Our AI team reads the visit type, provider, department and site mapping in athenaOne. It infers from booking history which types each generic slot actually absorbs. Then it reports slot use against the real type, not the template label. Software should not settle that map alone. It goes back to the practice for sign-off, and a scheduling lead owns the exceptions from then on.

Wait time belongs to a provider, not a group

Time to third next available visit is the access measure that survives a real schedule. The first and second openings are often cancellations, so they overstate open time. The third is a steadier read on backlog. AHRQ advises checking appointment availability daily on exactly that basis.

It is a provider-level number, and it tracks what patients say. A study of a large primary care group, published in JAMA Network Open, looked at that link. As third next available rose by one week, the share of patients rating their wait as good or excellent fell by 7.35%.

That study was primary care, not cardiology. The number is not a cardiology benchmark. The mechanism is what carries over. A patient feels the open time on one provider’s schedule, not the group’s. Average it across sites and the signal is gone. Run it per provider, per visit type, per department, and the two or three schedules driving your wait time stop being a mystery.

Your no-show rate on linked visits is wrong before you read it

Cardiology runs on paired bookings. An imaging visit and a provider visit, booked together with set spacing, is an ordinary shape. Nothing in the scheduling template enforces the pair. It exists as convention. Patients just call and ask to see the doctor.

Two measurement problems follow. First, the second leg often never gets booked, and a report counting completed visits cannot see a booking nobody made. Second, and worse, native EHR reminders fire on the first appointment of the day only. Give a patient a 9:30 imaging slot and a 10:00 provider visit and one of them gets a reminder. When the patient misses the other, your no-show rate reads it as patient behavior. What happened was a reminder that never went out.

No-shows were the top patient access focus for 2026 in a December 9, 2025 MGMA Stat poll. 27% of 236 respondents chose it, ahead of online scheduling at 24% and phone access at 22%. A no-show number that quietly includes unreminded second legs sends all of that effort the wrong way.

The fix is administrative and specific. The automation spots the pair at booking, creates both visits, replaces the native reminder with one confirmation per leg, and re-links the pair when either side moves. It does not decide which visits belong together. That rule gets proposed from 90 days of booking history and confirmed by the practice. When a pair cannot be completed, the call goes to a scheduler instead of being closed out.

Open slots against pending outreach, by site

The most useful view in a multi-site review is not a KPI. It is a heat map. Open slots at each site, set against the patients waiting to be scheduled there. One office shows a hundred open slots and a dozen pending patients. Another shows three open slots and forty pending. Those two sites need opposite work. A group average recommends neither.

Cardiology stacks a referral funnel on top. Referrals arrive by fax, through a provider portal, and as self-referrals. None of them arrive as structured orders. Conversion from referral received to first visit completed, read per referring source and per site, is where the growth question lives. A group that cannot produce that number is guessing when it decides where to spend.

Turn outbound scheduling up where slots are open and the queue is thin. Turn it down where the queue already runs longer than the capacity. Let the same view decide where new-patient demand is worth buying.

Per-call scoring, and what a real 30-day read looks like

Access metrics give you the outcome. Per-call scoring gives you the reason. Score each call on four things. Was the patient matched to the right chart. Was the right visit type selected. Was eligibility checked. Did the handoff to staff happen at the right moment. That turns a queue statistic into something a supervisor can coach against. It also catches errors that never reach a report, like a caller told the practice does not take their plan when it does.

Set the baseline before anything is deployed. Third next available by provider and visit type. Fill rate and booking rate by site. No-show rate by visit type, with linked visits pulled out. Referral conversion by source. Abandoned calls by hour. All of it sits in athenaOne data now. None of it can be rebuilt after the fact.

Customers report more than 50% of calls auto-resolved in roughly 30 days, calls answered in under five seconds, hold times under five minutes, and 13% of visits booked after hours. Your numbers will vary by workflow, staffing, seasonality and call mix. The after-hours figure is worth pausing on. It stays invisible to any practice that is not already counting bookings by hour.

Key takeaways

Report access by department and provider. A group average can hold flat while one site loses two weeks of new-patient access. That is the case the review exists to catch.
Resolve the generic Any 15 and Any 30 slots to the visit types they can actually hold before you trust a slot-use number. Have the practice sign off on that map.
Pull linked visits out of the no-show rate and check that a reminder fired on each leg. A native reminder that covers only the first visit of the day turns a notification gap into what looks like patient behavior.
Baseline first: third next available by provider and type, fill and booking rate by site, no-show by visit type, referral conversion by source, abandoned calls by hour.
Run open slots against pending outreach per site every week. The site with a hundred open slots and the site with forty pending patients need opposite work.
Score calls on chart match, visit type, eligibility check and handoff timing. Then every move in an access number has a reason attached to it.

The phones getting answered is not the result. Ask whether the third site’s schedule filled, whose panel absorbed it, and what getting there cost. Those numbers sit in athenaOne today, one report away from an operator who has never had time to build it.

Sources: