Patient Intake Software: Custom Build or Off the Shelf

What the off-the-shelf products actually do well

Groups at this size have almost always bought something already, which is worth remembering before anyone proposes a build. Phreesia tablets at the flagship sites, IntakeQ or Jotform links emailed ahead, Klara or NexHealth handling reminders. Those products are not the problem, and several of them are very good.

What you get without a project:

• A phone friendly packet that patients complete before arrival, with reminder cadences already tuned by vendors who have watched millions of patients abandon halfway.
• A signed business associate agreement and a hosting posture you did not have to design, which is a genuine cost you would otherwise carry.
• Demographics and coverage synchronised into the chart, usually one way and usually nightly, which covers the fields most single specialty practices care about.
• Eligibility checking through the vendor's clearinghouse, priced per transaction, which beats not checking at all by a wide margin.
• Kiosk or tablet flows for the walk in who never saw the email, and support staff who will answer on a Monday morning.

If you run one to four sites on one record system, this is the answer. A custom build there is a suit tailored for a body that is still growing, and off the shelf also wins whenever there is no internal owner, because software without a product owner decays inside eighteen months. We say this on first calls and it costs us work.

Where they stop: the integration that turns out to be a PDF

The specific failure is not the form. It is what happens to the answers.

Six weeks after go live, an operations director learns what the sales deck meant by integrated. A document lands in the chart's files tab. Demographics sync one way overnight. Medication history, surgical history, social history and screening scores arrive as flat images, and a medical assistant confirms every field by hand because nobody trusts a sync they cannot see. So the packet gets typed twice: once by the patient, once by your staff.

This is structural rather than lazy. A vendor supporting forty record systems builds to the lowest common denominator across all of them, and the marketplaces add their own friction with listing fees, certification queues and rate limits negotiated for the average customer rather than for you. Nobody is going to solve your athenahealth ceiling on your behalf.

A build works against what your record system will actually accept. FHIR R4 resources such as Patient, Coverage, AllergyIntolerance and QuestionnaireResponse where they are exposed, HL7 version 2 ADT and SIU feeds for registration and schedule, native interfaces where they exist, and an explicit exception queue for the fields the system refuses. A date of birth mismatch or an unrecognised plan name becomes a side by side comparison a coordinator resolves in under a minute rather than a full packet retyped.

Two more places the seams show. Insurance, because a card photographed on a kitchen table with glare over the member identifier becomes a denial forty-five days later, and running a real time 270 and 271 eligibility check two days before the visit is the only moment when a correction is still cheap. And form variants, because a group running general clinic, surgery and cosmetics across two states needs different questionnaires per visit type, different consent language per state, Spanish at three sites and a financial policy the lawyers revised in March. In a form builder that becomes forty maintained variants, and when legal changes one paragraph somebody edits forty forms and hopes. Then a payer audit asks which version this patient signed in February and nobody can answer.

The arithmetic: per provider fees and the transcription hours

Two numbers decide this, and the second is bigger.

Start with licensing. Published list pricing at the form tool end sits near fifty dollars per practitioner each month, and the enterprise intake platforms quote privately, structured per provider and per location, with add ons for messaging and a margin taken on every payment. At sixty dollars per provider per month, a forty provider group pays about $28,800 a year. A first release at $95,000 with 18 percent annual upkeep is roughly $129,000 across three years, near $43,000 a year averaged. On licence alone the crossover sits around sixty providers, and higher if you are on published list pricing rather than an enterprise contract.

Now the number that actually decides it. In Digital Heroes delivery engagements we have timed the re-keying loop at 12 to 22 minutes of staff work per new patient packet. Run your own version: stopwatch ten packets end to end, from the moment the response arrives to the moment a medical assistant trusts the chart.

At 25 new patients per location per week across twelve locations, fifteen minutes each is 75 staff hours a week. That is two full time positions doing transcription, before counting reworked claims caused by a mistyped member identifier at twenty-four dollars an hour and twenty minutes each.

So the honest crossover is about ten locations or forty providers, whichever arrives first, and it arrives immediately if you are already paying a subscription and still employing people to transcribe. That is paying for the same work twice.

What a custom build actually costs

These bands come from Digital Heroes delivery across more than 2,000 projects. A focused first release covering phone first pre-visit intake, integration with one record system, real time eligibility, versioned electronic signature consents, the staff exception queue and form logic for one specialty runs $60,000 to $130,000 and ships in 12 to 16 weeks. A full platform adding payments with card on file, kiosk mode for walk ins, a second record system after an acquisition and operational analytics runs $150,000 to $400,000 phased over 6 to 12 months.

Two costs nobody quotes:

• Data migration is 10 to 25 percent of the build. Live patients are easy. Historical signed consents are not, because they have to remain retrievable for your state's full retention period after the old vendor is gone, with the exact version and rendering the patient saw. Get that export before you give notice, not after.
• Year two is 15 to 20 percent of build cost annually. Record vendors version their interfaces, payers rename plans, consent language changes, and each new specialty you add brings its own packet logic.

What pushes the range up: the interface itself, since a documented modern interface sits at the cheap end, an HL7 feed through an engine in the middle and a marketplace certification process at the expensive end; how many specialties ship at launch; and whether payments are in scope, because taking the copay on your own merchant account is where the margin stays with you.

The four situations where building wins

• Regulatory fit. Every subprocessor touching protected health information needs a business associate agreement, and several popular character recognition services used to read insurance cards are not eligible for one. Consents must be versioned documents storing the exact wording, timestamp and rendering signed, because that is precisely what a records request or payer audit asks for. Those are ownership questions, not features.
• Scale economics. Past roughly ten locations, per provider and per location pricing grows every time you open a site or hire a physician assistant, while a build's run cost is infrastructure plus maintenance regardless of count. Opening location thirteen becomes a configuration change rather than a licence negotiation.
• A workflow that is your competitive advantage. Multi specialty packet logic is the case. One form library with rules keyed to appointment type, provider, location, payer class and patient language assembles the right packet per visit, and returning patients confirm what changed rather than re-entering their history every January.
• Integration sprawl across three or more systems. Record system, clearinghouse, merchant processor, reminder tool and a call centre dashboard all holding part of one arrival is where staff become the integration layer. Owning the join is what lets intake data feed your own operational reporting.

How to decide in a week

Time ten packets. Pick one busy location, take the next ten new patient arrivals, and record two clocks per packet: minutes the patient spent, and minutes your staff spent after the patient finished. Include the scanner, the exception the coordinator chased, and the medical assistant confirming the medication list in the room.

Multiply the staff figure by your weekly new patient volume across all sites and annualise it at a loaded hourly rate. If the total sits under $40,000 a year, keep your current tool and spend the effort on reminder timing, which lifts completion rates for nothing. Above $120,000 and no configuration change reaches it, because the cause is that the vendor cannot write the fields you need.

While that runs, ask your billing lead for last quarter's denials caused by eligibility or member identifier errors. That number belongs in the same business case and is usually missing from it.

Then turn the findings into a specification. Digital Heroes runs a paid discovery phase that ends in a signed product requirements document covering the data model with Patient, Coverage, Encounter, QuestionnaireResponse and versioned Consent as first class entities, the interface ceiling of your record system, and acceptance criteria. You keep that document either way.

Who we are wrong for: practices under five locations, groups with no internal product owner, and anyone wanting developers without a written specification. We fit multi site groups already paying a subscription and still paying people to retype. Over fifty specialists, more than 2,000 projects, and a named team you meet before signing. India LLP, US LLC and UK LTD entities so intellectual property assigns under your own law. Verifiable on Clutch, Trustpilot, Fiverr Vetted Pro and D-U-N-S.

Book a 30-minute call with Digital Heroes and get a written plan and a fixed quote within 48 hours.