Compliance first: We never take HIPAA/PHI data or personal customer data. Everything is de-identified before it leaves your control.
Home / Sell my data / Industry software / Industry software · Admin side only, no patient records
Industry software · Admin side only, no patient records

Practice Management Data for AI: Admin-Side Scheduling and Billing Only, Never Patient Records

Practice Management Data for AI: Admin-Side Scheduling and Billing Only, Never Patient Records

Practices can license practice management data for AI, admin side only: scheduling flow, billing operations and front-office workflow. Patient records and PHI are never included, ever. Here is what qualifies, what is excluded, and how to request a bid.

Read this first: this page is about the administrative side of a practice system only. Patient records, clinical notes, diagnoses, treatment details and any protected health information (PHI) are never part of what we accept. What remains is the operational machinery of a practice: how the calendar fills, how claims move and how the front desk keeps the day on track.

Admin side only: the line we never cross

Patient records and PHI are never included. Not partially, not de-identified, not as a sample. This is a hard boundary, not a preference.

We never take HIPAA/PHI data or personal customer data. Everything is de-identified before it leaves your control.

What that excludes, specifically:

  • Patient names, dates of birth, addresses, phone numbers, emails, insurance member IDs and any other identifiers
  • Medical, dental or veterinary-owner records, clinical notes, charts, images and lab results
  • Diagnosis codes, procedure codes tied to visits, treatment plans and prescriptions
  • Any field that describes a patient's health, condition or care

What can qualify is administrative operations data with no patient attached. If a field could tell anyone anything about a patient, it stays home. HIPAA and state privacy rules are complex, so talk to your attorney and compliance officer before you share anything. See PHI in the glossary for the definition we use.

What administrative practice data looks like

Once patient information is fully excluded, a practice system still holds a detailed picture of how a busy office operates:

  • Scheduling operations: slots offered, slots filled, appointment-type category (new, follow-up, hygiene, procedure block), lead time from booking to visit, same-day adds, reschedules, cancellations and no-show rates in aggregate.
  • Provider and room utilization: booked versus available hours by anonymous provider ID and operatory or room, template changes over time.
  • Front-office workflow: check-in to check-out durations, phone and portal booking volume by channel, waitlist fills and confirmation cadence.
  • Billing operations: claim submission lag, payer category, denial reason categories, resubmission cycles, days in accounts receivable and write-off categories, all without visit-level clinical detail.
  • Staffing: front-desk and billing headcount by role, shift coverage against schedule load.

This is the business side of care delivery, and none of it requires knowing who any patient is.

Why AI builders want front-office and billing workflows

Healthcare administration is heavy on manual work, and automation builders need to see how real offices handle it.

De-identified, admin-only data helps train:

  • Scheduling agents that fill templates, manage waitlists and reduce gaps using real utilization patterns.
  • No-show and cancellation forecasting at the schedule level, so offices can overbook or confirm intelligently.
  • Revenue-cycle automation that learns claim timing, denial categories and resubmission paths.

Founder Jeff Cline puts it simply: "Every industry is a geek away from being uberized." Practice operations are no exception, and the builders working on them need real workflow data, never patient data. The healthcare practices industry page and dental and veterinary page explain the admin-only scope in more depth.

How de-identification works for an admin-only dataset

Because patient data is excluded at the source, de-identification here focuses on everything else that could point to a person or practice:

  1. Exclude before export. Only administrative tables and fields are selected. Patient demographic, clinical and insurance-member tables are never queried.
  2. Remove the link to any person. Appointment and claim rows carry no patient key at all; where an event-level row exists, its ID is a random token with no path back to a chart.
  3. Anonymize staff. Providers, hygienists, front-desk and billing staff become role labels with stable anonymous IDs.
  4. Generalize the practice. Name and address become specialty, size band and region.
  5. Drop free text. Scheduling notes and billing comments are excluded; free text is where patient details hide.
  6. Review before release. You and your compliance contact approve the final field list before anything leaves your control.

Again: patient records are never included. If in doubt about a field, it is left out.

Finding your numbers, admin side only

Where to find it: Admin side only (scheduling, billing). Patient records are never included.

Ask your office manager or billing lead for administrative counts only: total appointments, earliest date, total claims, providers, staff and locations. Never pull patient lists or charts. Those administrative totals are enough for the valuation calculator and for a buyer to understand scale.

An admin-only header sample (no patient fields)

A header sample is column names only, with no rows. For a practice system, it must also contain no patient fields of any kind. A compliant example:

  • practice_id_anon, specialty_category, practice_size_band, region
  • slot_date, slot_start, slot_length_min, provider_id_anon, provider_role, room_id_anon
  • appointment_type_category, booked_at, lead_time_days, booking_channel
  • slot_status (filled, open, cancelled, no-show, rescheduled)
  • checkin_to_checkout_min, confirmation_attempts
  • claim_batch_date, payer_category, submission_lag_days
  • denial_reason_category, resubmission_count, days_in_ar_bucket
  • staff_role, shift_start, shift_end

You will not see patient name, date of birth, address, member ID, diagnosis, procedure or any clinical field, because those are never part of a dataset. Our header sample how-to has more on preparing one.

What makes admin-side practice data valuable

  • Years of schedule history. Long runs show seasonality, template changes and how no-show patterns evolve.
  • Locations and providers. Multi-location groups give buyers comparative workflow data.
  • Billing depth. Clean claim-status and denial-category history is especially sought after for revenue-cycle tools.
  • Specialty. Dental, veterinary, therapy and specialty clinics each run their front office differently.

According to founder Jeff Cline, small companies have gotten over $100,000 for their data. Results vary; not a guarantee.

The process, and why timing matters

  1. Describe the practice. On the bid request page, share the system you use, years on it, number of providers and staff, and approximate appointment and claim volume. Admin numbers only.
  2. Send an admin-only header sample. Column names from scheduling and billing exports, with no patient fields.
  3. Review competing bids. Several buyers can bid so you see the highest. According to Jeff Cline, bids come back within 72 hours if the data is correct and a header sample is provided. Results vary; not a guarantee.
  4. De-identify, license, deliver, get paid. The admin-only dataset is cleaned before it leaves your control, licensed under a written agreement, and delivered. Jeff Cline says deals typically close in about 90 days. Results vary; not a guarantee.

Admin-automation models are being trained now; once the patterns are learned, later datasets are worth less. A free valuation is the simple next step. One more time, because it matters: patient records are never included.

How to request a bid for your Practice / patient system (admin side only) data

  1. Confirm admin-only scope. Agree internally, with your compliance contact, that only scheduling, billing-operations and front-office data will be considered. Patient records are never included.
  2. Gather admin totals. Have your office manager or billing lead note total appointments, total claims, the earliest date and provider and staff counts.
  3. Request a bid. Submit those administrative figures at /get-a-bid.
  4. Send an admin-only header sample. Share column names from scheduling and billing exports, with no patient fields and no records.
  5. Approve and close. Compare competing bids, review the de-identified field list with your attorney, then license and deliver.
Our compliance promise: We never take HIPAA/PHI data or personal customer data. Everything is de-identified before it leaves your control.

Frequently asked questions

Do you take patient records or PHI?

No, never. Patient records, clinical notes, diagnoses, procedures, insurance member IDs and any other PHI are excluded entirely, not even in de-identified form. We never take HIPAA/PHI data or personal customer data. Everything is de-identified before it leaves your control.

What practice data can be sold, then?

Only administrative operations data: schedule utilization, appointment-type categories, booking lead times, aggregate no-show and cancellation patterns, front-office timing, staffing and claim workflow timing. Nothing that identifies or describes a patient.

Is selling admin-side practice data legal?

It can be, when the data contains no PHI and your agreements allow it, but the rules are complex and vary by state and by contract. Talk to your attorney and compliance officer before sharing anything, including a header sample.

What goes in a practice header sample?

Only administrative column names, such as slot date, appointment type category, provider anonymous ID, slot status, payer category and denial reason category. No patient fields appear in the header sample, and no records are sent.

How quickly could I get an offer?

In Jeff Cline's experience, bids come back within 72 hours if the data is correct and a header sample is provided, and deals typically close in about 90 days. Results vary; not a guarantee.

Updated November 19, 2026.

OpenDataPolicies.org connects businesses with buyers of de-identified operational data. We are not a law firm, broker-dealer or investment adviser, and nothing on this site is legal, tax or financial advice. Statements about amounts, timing and outcomes are Jeff Cline's statements based on past transactions. Results vary; nothing here is a guarantee of any bid, price, buyer or closing. Every deal depends on the data, its quality, its rights and buyer demand at the time.