What we keep, what we don't, and where the data comes from

The short version

DocBeat holds no patient records and asks for none. There are no accounts, no logins and no passwords. Using the site means typing a place, a specialty or a name into a search box and reading public information about physicians. That is the whole exchange.

What we keep is small and listed below. What we do not keep is your identity.

What the site stores about a visit

Three things, and only when they happen:

Search text. The words typed into the search box, what they resolved to (a town, a specialty, a name), how many results came back, and whether a spelling correction was applied. This is how we find searches that return nothing and fix them.

Result opens. When a result is opened, we record the search text, the physician's public NPI and the position in the list. This tells us whether a search gave people what they were looking for.

Profile views and taps. When a profile is opened, when the Call or Share control is tapped, when a filter or the map is used, we record that it happened, with the physician's public NPI where one applies.

Reports. If you use "Report inaccurate information" on a profile, we store the NPI, the reasons you chose, any note you write, and an email address only if you choose to give one so we can reply.

Email sign-ups. If you ask to be told when we cover your state, or when a physician near your search is accepting new patients, we store the address you typed, which of those you asked for, and the search it was about. It is used for that one message and nothing else, and it is deleted once the message has gone or the moment you ask (write to hello@docbeat.ai).

A visit id. To count whether a search led to a profile and then to a call, each browser tab gets a random id that lasts only as long as the tab is open; the browser discards it when the tab closes. It is not a cookie, it is not shared with anyone, and it is never stored next to anything that identifies you. Each record also carries the country and state the request came from, at that level only.

None of these records carries an IP address, a cookie, a device identifier, a name or an account. We do not build profiles of visitors, and we do not sell or share visit-level data with anyone.

What the site does not do

No cookies for tracking. The site sets no advertising or tracking cookies. Traffic counts come from a privacy-preserving analytics service that does not use cookies or identify individuals.

No third-party advertising scripts. Nothing on any page loads from an ad network.

No sale of visitor data. There is nothing to sell.

No health information. We never ask why you are looking for a physician, and the search box is not a place to put medical details. If you type them anyway, they land in the search log described above; we do not read individual searches, and the log exists to improve results, not to study people.

Where the physician information comes from

Every physician listed is drawn from public records: the federal National Provider Identifier registry, state medical licensing boards, Medicare enrollment and opt-out files, and the practices' or their networks' own published pages for facts such as a membership care model. Each fact on a profile traces to one of these sources. Where we are not sure, we leave the physician out rather than publish with a caveat.

Street addresses are shown only where the record indicates a clinical location. Where an address may be a residence, the profile shows the town only, and the map shows an approximate position rather than a point.

Physicians who believe their listing is wrong can use the report button on their profile or write to us; corrections are checked against the primary source and applied.

How the site is built and hosted

The site runs on Vercel and reads physician data from a database hosted on Supabase. All traffic between your browser and the site, and between the site and its database, is encrypted in transit. The browser is served read-only views of the data; the credentials that can write to the database never reach the browser and are held as server-side secrets.

The hosting provider keeps standard, short-lived request logs for operating the service, as any host does. Those logs are the provider's, are not exported into our database, and are not used to identify visitors.

Search endpoints are rate-limited, so the directory cannot be copied wholesale by an automated client.

Reporting a security concern

If you find a vulnerability or something that looks wrong, write to hello@docbeat.ai with what you saw and how to reproduce it. We read every message, we will not take action against good-faith research, and we will tell you what we did about it.