Privacy

Meridian is built to need as little of your data as possible. You can browse, search, and query every dataset without an account, and your queries run in your browser, not on our servers.

What we collect

Anonymous usage events. We record product events (a page was viewed, a search happened, a CSV was exported, a dataset was requested) with random identifiers for the visit and the tab, your browser's user-agent string, which tells us the browser, operating system and type of device and lets us separate people from crawlers, and the screen and window size, browser language and time zone your browser reports. If you submit a dataset request, the request text you type is included; what you type in the search box is not. When you run a query in the dataset explorer on this site, we record its shape: the columns, filters and functions it uses, with every value you typed replaced by ?, plus how many rows came back and how long it took. The values themselves are never recorded. PostHog's library keeps the random identifiers for the visit and the tab, the full address of the page you landed on (its query string included) and the address of the page that sent you, in your tab's session storage, which your browser clears when the tab closes; where session storage is unavailable it keeps them in memory instead, lost when you leave the page. Only the site and path of any address are sent to PostHog. From a link's query string we send just its five utm_ campaign tags (source, medium, campaign, content, term), never an advertising platform's click identifier. No cookies, no cross-site tracking, no advertising identifiers, no IP-based profiles. Events are processed in the EU by PostHog.

Request logs. Like any website's host, Cloudflare logs each request to meridian.online, to the install site at install.meridian.online and to the open lake at openlake.meridian.online: your IP address, the user-agent string of your browser or tool, the page or file requested, the page that sent the request and your country. We keep a copy without the page or site that sent the request, and with nothing from a link's query string but its five utm_ campaign tags, to count people, AI assistants, crawlers, DuckDB clients and installs, and to tell them apart, and we remove the IP address from it once it is 90 days old. A query you run from your own DuckDB against the open lake reaches us only as requests for parts of files; we never see its SQL.

Email addresses you give us. If you join a waitlist or request a dataset with your email, we store that address with the list you joined and use it only for that purpose. Lists are held by our email provider Loops in the United States, with EU/UK transfers under standard safeguards (Data Privacy Framework, standard contractual clauses). Every email includes an unsubscribe link. The sign-up form on the early-access page, the Windows and Linux waitlist and the roadmap sends a little more with the address: how you heard of us, if you say; where you signed up (which page, what you came to it for, and on the waitlist whether your browser reports Windows, Linux or something else); the visit's campaign tags; the site that sent you here; and the page you landed on, so we can tell which launch link brought each person.

Payments. If you buy something, checkout is handled by Polar (our merchant of record). Card details never touch Meridian.

What we don't do

No selling or sharing of personal data. No ads. No consent-banner dark patterns — we avoid needing one. The data you query, and every value in the SQL you write, stay in your browser.

Contact

Questions or a data request: email hello@meridian.online.