See what each insurance company really costs your practice, from your own payment files and your contracted rates. Free to use.
Open the payer check in any web browser. There’s nothing to install and no sign-up.
Your files never leave your computer. The page works like a calculator inside your browser. Nothing is uploaded, not to us and not to anyone else, and it works with your internet turned off. Your IT team can verify that in a few minutes (steps below).
You’ll need two things your billing team already has: the insurance payment files (835s, also called ERAs) from the last 90 days or so, and a spreadsheet of your contracted rates. The page has a template for the rates.
The rest of this page is for whoever checks the tool on your behalf.
For each payer: claims, dollars paid, denied claims (valued at your contract rate), visits paid at a lower code than billed, lines paid short of contract, reversals, days to pay, the staff time all of that takes, and one number that pulls it together, the cost for every $100 the payer paid you.
The check runs on your machine, in a browser tab or in a Python script. Nothing is uploaded, and neither one has any code that talks to the internet.
How a practice’s IT person can verify it:
payer-check.html, turn off Wi-Fi (or unplug the network cable), open the file and run it. It works the same.default-src 'none'; connect-src 'none') that tells the browser to block every fetch, upload and outside file the page could ask for. The policy doesn’t cover the page sending the tab somewhere else, so the page has no links or code that open another site, and the test below watches for that too.tests/test_page.py opens the page, runs the sample remits through it, saves every output, waits a few seconds, and records every request the browser makes and every window it opens. Any request or new window fails the build, and it runs on every push.The patient’s name in the 835 (the NM1*QC segment) is never read.
payer-check.html (double-click it, no install).Python 3.9 or newer, nothing to install.
python3 payer_check.py path/to/835-folder path/to/fee-schedule.csv --out payer-check-output
Options: --hourly 30.00 (staff cost an hour), --writeoff-pct 46.0 (share of denied dollars never recovered), --slow-days 30.
The page and the script share one logic, and the tests require them to write byte-identical files from the same input.
.835, .txt, .edi, .x12, .era or .rmt extension that starts with an ISA segment.payer, cpt and contracted_allowed, plus an optional effective date (YYYY-MM-DD) for rates that change partway through. Extra columns are ignored. The payer name has to match the payer name in the 835 (the N1*PR segment), case aside. A payer with no rows is named in the results and left unpriced.| File | What’s in it | Patient-level fields |
|---|---|---|
summary-by-payer.csv |
One row per payer | None |
summary-by-payer-and-code.csv |
One row per payer and billed code | None |
chart-pull-list.csv |
Every flagged line (paid at a lower code, paid short, denied) with the patient account number, so your team can pull the chart | Yes. Keep it on your computer. |
Every row starts with a source column. Files built from the sample remits say “Sample data” on every row.
Cost per $100 paid adds the dollars paid short (lower codes and short pays), the share of denied dollars never recovered, and staff time, then sets it against what the payer paid. The inputs are assumptions you can change:
What a remit can’t show stays out: prior authorization hours, records requests, portal and phone time, credentialing.
sample-data/ holds 90 days of made-up remits for a made-up practice, plus its fee schedule. See sample-data/README.md. It is Sample data, labeled that way on every surface, and the numbers it produces are not results.
python3 payer_check.py sample-data/835 sample-data/fee-schedule.csv
python3 -m venv .venv && .venv/bin/pip install -r requirements-dev.txt && .venv/bin/python -m playwright install chromium
.venv/bin/python -m pytest -v
tests/reference/).Code under the Apache License 2.0. Documentation under Creative Commons Attribution 4.0. From Serious AI Partners.