Standardizing Intake Across Practices That Do It Differently

Standardizing intake is the first project every MSO tries and the one most likely to stall. The reason is that the word standardize gets read as make every practice do it the same way, and the practices have reasons for their differences that survive the meeting where you tell them to stop.

Intake looks like the most standardizable process in the group and turns out to be one of the least.

The required fields at check-in are configured per department. What counts as a new patient is a different rule at each practice: six months at one, two years at another, three years at a third, with a separate billing definition sitting underneath. Some sites collect a referral source and some never have. Consent forms attach to appointment types that do not exist at every location.

None of these are mistakes. They came from real payer mixes, real specialty requirements, and real staffing. Rolling one script across all of them produces a call that is wrong at most sites.

Meanwhile the cost of not standardizing is a group that cannot report on its own new-patient funnel, because no two practices define the terms the same way.

It helps to size this correctly before deciding how much to invest.

A March 10, 2026, MGMA Stat poll of 294 applicable responses asking practice leaders which phone tasks consume the most staff time put eligibility and prior authorization at 45%, scheduling at 31%, intake at 9%, and prescription refills at 6%. Intake is not where the minutes are.

It is where the errors are. A registration captured wrong produces a duplicate chart, a failed eligibility check, a claim with the wrong subscriber, a reminder sent to a phone number that belongs to a relative, and a patient who arrives without completed forms. Each of those consumes time in a different department, weeks later, attributed to something else.

That is why intake standardization pays even though intake is a small slice of the call volume. You are not optimizing nine percent of phone time. You are removing the input errors that generate work everywhere downstream.

Read the required fields, do not maintain a copy of them

The mechanical part of standardization is simpler than the political part, and most groups do it the hard way.

Each department already carries its own definition of what has to be captured before a patient can check in. Reading GET /departments/{departmentid}/checkinrequired gives the automation that list per site, and creating the chart through POST /patients writes back against it. The rules stay where the practice manages them, and the automation adapts rather than holding a second copy.

The second copy is what kills these projects. A spreadsheet of required fields per site is accurate the week it is written and wrong within a quarter, because a practice adds a field and tells nobody. An automation reading the live configuration cannot drift.

What gets standardized instead is the vocabulary. Define one set of intake outcomes for the whole group, chart created, chart matched to existing patient, insufficient information, needs human review, and map each site’s field requirements underneath. Every practice keeps its rules. The group gets one report.

The complication: intake is data completion, not a form

The largest quality gain in intake comes from asking fewer questions rather than asking them more consistently.

When a referral has already loaded demographics into the chart, walking the patient through the full intake is a needless call that patients abandon partway through. The right behavior is to ask only for what is missing. Where nothing exists at all, several practices prefer the intake call to happen the day before the appointment, so most of the paperwork is already complete when the patient arrives.

Identity is where this gets genuinely difficult. Family members share a phone number, and calls get documented against whoever owns the number rather than whoever the call is about. A parent calling about a child, or one spouse about another, both patients at the practice, produces a message filed in the wrong chart that staff then move by hand. Two fixes cover most of it: ask whether the call is for the caller or someone else at the top of every call, and detect the mid-call handoff when the phone changes hands.

Third parties are an explicit exception rather than a feature of identity verification. Practices routinely need authorized relatives, and sometimes service companies calling on a patient’s behalf, to be able to book. That permission has to be configured per patient, not inferred from the fact that the caller knows a date of birth.

The handoff is any identity ambiguity the automation cannot close. Two charts with the same name and date of birth, a caller who cannot confirm a second identifier, or a partial match against an existing patient goes to staff with both candidate records attached. An automation that creates a chart to avoid an awkward pause has just created the duplicate the group is trying to eliminate.

New patient means four different things, and all of them are right

This is the definition that most often blocks group-level reporting, and it cannot be resolved by picking one.

One practice treats a patient as new again after six months, partly because a full re-intake is required for their patient population. An obstetrics practice uses two years, mapped to different appointment types and different documentation time. Another uses three years, after which patients are booked as a follow-up rather than as a new patient. A podiatry group flagged the billing edge directly: the Medicare new-patient rule runs on a three-year boundary, but the practice still wants fresh paperwork after two.

So there are two definitions in play at every site, a scheduling definition and a billing definition, and they do not have to agree. Trying to collapse them creates either wrong appointment lengths or wrong claims.

The workable model keeps both, per site, as configuration the automation reads. Scheduling uses the practice’s own boundary to pick the appointment type and duration. Registration uses the paperwork boundary to decide which forms to send. Group reporting normalizes afterward, at the report layer, where changing a definition costs nothing.

Getting this wrong is expensive in a specific way. A returning patient booked as established into a short slot when the site considers them new produces a visit that runs over, a provider who is behind for the rest of the day, and paperwork completed in the room instead of before it.

Language and confirmations are part of intake, not extras

Two details get treated as polish and are actually where patients decide whether the practice is competent.

Language handling is harder than a menu option. Automatic detection works, but bilingual callers mix languages mid-sentence and a system that switches on every cue flips back and forth. Practices also want certainty before switching, because guessing wrong about a patient’s language in a healthcare setting is worse than asking. Some sites want a Spanish path on for certain queues and off for others. Beyond the most common languages, quality drops and the right answer is a human language line rather than a confident machine.

This is also a compliance surface. Federal rules require covered entities to take reasonable steps to provide meaningful access to individuals with limited English proficiency, which makes the language path part of the intake design rather than a feature request.

Confirmation language has to be conditional on the booking itself. One health center asked that the standard please give us 48 hours notice to cancel line be dropped from same-day bookings, where it reads as nonsense. Small, and exactly the kind of thing that makes an automated confirmation feel machine-generated. Confirmations should be assembled from the booking, not printed from a template.

Asked where practices are aiming AI first, a Feb. 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%. Registration sits third on that list, which is roughly where its difficulty belongs.

Key Takeaways

Standardize the intake vocabulary across the group, not the field requirements at each site. One set of outcomes, mapped to per-site rules.
Read each department’s required check-in fields from live configuration rather than maintaining a spreadsheet copy that drifts within a quarter.
Ask only for the fields a referral did not already supply. Full intake walkthroughs on pre-loaded charts are the calls patients abandon.
Ask whether the call is about the caller or someone else at the top of every call, and detect the mid-call handoff when the phone changes hands.
Configure authorized third-party callers per patient. Knowing a date of birth is not authorization to book.
Keep both new-patient definitions per site, the scheduling one and the billing one, and normalize at the reporting layer instead of forcing one.
Route identity ambiguity to staff with both candidate records attached. Creating a chart to avoid an awkward pause is how duplicates happen.
Assemble confirmations from the booking. A cancellation-notice line on a same-day appointment tells the patient a machine wrote it.

The intake project that works across forty practices is not the one that makes them identical. It is the one that reads what each site actually requires, asks the patient only for what is missing, keeps both new-patient definitions instead of arguing about them, and produces one vocabulary the group can report on. An AI team working inside athenaOne can hold that per-site configuration without a spreadsheet, run the identity checks that stop duplicate charts before they are created, handle the language path with the certainty a healthcare conversation needs, and hand the ambiguous records to staff with the candidates already assembled.

Sources

Intake is a smaller share of phone time than you think, and a larger share of the damage

Read the required fields, do not maintain a copy of them

The complication: intake is data completion, not a form

New patient means four different things, and all of them are right

Language and confirmations are part of intake, not extras

Key Takeaways

Related reading

Ready to See It in Action?