Judge a Front-Office Vendor by the Calendar, Not the Features

The most useful question to ask a front-office vendor is not what it does. It is when it will be doing it, who does the work between signature and go-live, and what the practice has to supply along the way. Feature lists are comparable across vendors and largely interchangeable. Calendars are not, and the calendar is what determines whether the thing you bought in March is answering your phones in April or in October.

Urgent care feels this more sharply than most practices because the after-hours volume does not wait for a rollout. The clinic closes at eight and the phone keeps ringing: patients who cannot get through in the morning, people asking whether to come in tomorrow, callers who need an appointment at another site and will otherwise go somewhere else entirely.

Every month of implementation is a month of that volume going to voicemail. The cost of a slow rollout is not the project management overhead. It is the calls, and unlike most implementation costs it does not appear on any invoice.

Practices have also been trained to expect the slow version. Long timelines, a discovery phase, staff pulled into workshops, a training program, and an integration bill. That expectation is worth interrogating, because it is a habit from a different category of software rather than a law of nature.

Time to live is the number, and it should be quoted in weeks

Ask for a date, then ask what has to be true for that date to hold.

A credible answer breaks the work into phases with owners. Setup and configuration first, then a pilot period where the automation runs against real calls and gets corrected, then go-live and tuning. Each phase should have a length and a named owner, and the owner should mostly not be you.

Pretty Good AI works in that three-phase shape, with most practices reaching measurable results in about thirty days and deployment measured in days or weeks rather than quarters. The reason it can be that short is that there is no infrastructure to stand up and no workflow rebuild: the automation operates inside athenaOne as it is already configured.

A vendor quoting quarters is telling you something real about their architecture. Ask what specifically takes that long. If the answer is integration work, ask why, because the platform is the same one everybody else is connecting to.

Ask who does the configuration, and be specific about it

The hidden cost in most rollouts is not money. It is the hours of your own staff that get consumed, usually by the one person who understands how your schedule really works.

The questions to ask: who writes down the booking rules, who maps the appointment types, who reconciles the provider roster, and how many hours of your team’s time does the vendor’s plan assume. A vendor that has done this before will answer in specifics because they have a checklist. A vendor that has not will say the process is collaborative.

Expect to supply a small number of things regardless: your appointment types and what each one actually means, which providers are bookable for what, your after-hours routing rules, and a handful of call recordings so the system learns how your practice talks. That is a genuinely short list and it is reasonable to ask for it in writing before you sign.

What should not be on the list is a training program for your front desk. If the plan requires retraining staff on a new process, the automation is being layered on top of your work rather than absorbing part of it.

The rule that trips up every fast after-hours rollout

Here is the complication that makes an urgent care launch harder than it looks, and it is worth raising in the first configuration conversation.

Most practices run two different minimum-lead-time rules at once. New patients cannot book inside three or four business days, because that window exists so registration and forms get completed. Established patients can often book the next business day. At 11pm those two rules collide with a caller who wants to be seen tomorrow and does not know which category they are in.

Getting that wrong in either direction costs you. Offer the new patient tomorrow and they arrive with nothing completed, which turns a fifteen-minute visit into a forty-minute one. Refuse the established patient tomorrow because the stricter rule was applied to everyone, and they go to the clinic down the road.

So the automation has to resolve which rule applies before it offers a time, using the patient record rather than what the caller says about themselves. When it cannot resolve it, or when the caller’s situation does not fit either rule, the call routes to on-call staff with what has been gathered so far. Anything about whether the patient should come in tonight belongs to a clinician, so those calls go straight to on-call staff, every time.

A vendor who can talk fluently about that distinction in a sales conversation has implemented urgent care before. One who cannot will discover it in week three of your rollout.

Pick the launch scope by where the calls actually pile up

Fast rollouts stay fast because they start narrow, and the way to choose the starting point is to look at what the sector is aiming at and then check your own queue.

A February 10, 2026, MGMA Stat poll of 177 applicable responses ranked scheduling at 31%, calls at 27%, registration and eligibility at 23%, and prior authorization at 16% as the top front-office targets for automation. Those are close enough together that the ranking is not the instruction. Your own after-hours mix is.

For most urgent care operations the highest-value first scope is the call that arrives after close and would otherwise become a voicemail: capture the intent, book into tomorrow’s correct appointment type at the correct site, and route anything clinical to on-call. That is a narrow enough surface to configure in days and it covers the volume that hurts most.

Expand after it is stable. Eligibility and registration next, then the daytime overflow. A phased scope is also how you keep the go-live date honest, because each phase is small enough to actually finish.

Judge the contract on the same axis as the calendar

Implementation speed and contract length are the same question asked twice, and a vendor’s answers should agree.

A vendor confident about time to live does not need a long lock-in, because the results arrive before the first renewal decision. Pretty Good AI is month to month, cancellable at the end of any month with no cancellation fee, and there is no upfront infrastructure investment to amortize. Those terms only work if the deployment is genuinely fast.

The inverse tells you as much. A multi-year term with a long implementation is a structure that gets paid before it performs, and the length of the term is doing the work the product is supposed to do.

This matters for access more than for procurement. Asked about their top patient access focus for 2026, 27% of 236 respondents in a December 9, 2025, MGMA Stat poll chose no-shows, ahead of online scheduling at 24% and phone access at 22%, with wait times at 21%. Whichever of those you are trying to move, you are trying to move it this year. A vendor whose calendar runs past your problem is not solving it.

Key Takeaways

Ask for a go-live date broken into phases with named owners, and check that the owner of each phase is mostly not your staff.
Get the list of what the practice must supply in writing before signing: appointment types, provider bookability, routing rules, and sample calls.
Treat a required staff retraining program as a signal that the automation sits on top of your work instead of absorbing part of it.
Resolve which minimum-lead-time rule applies from the patient record before offering a time, since new and established patients often run on different rules.
Launch on the after-hours call that would otherwise become a voicemail, then expand to eligibility and daytime overflow once it is stable.
Read a long contract term with a long implementation as the same claim twice, and prefer terms that only work if deployment is fast.

Time to live is the honest comparison between front-office vendors because it is the one number that cannot be written into a feature matrix. Ask for the calendar, ask who does the work, ask what you have to supply, and ask what happens to the after-hours call that arrives on day one of the rollout. The answers will sort the market faster than any demo.

Sources