Learn → Privacy & IT approval
This page exists to be forwarded. If someone at your firm has to approve new software before you can use it on client data, send them this — it answers the questions they will ask, and shows them how to check the answers rather than take our word for them.
Print-friendly — this page prints as a plain document| Data | Where it lives | Who can reach it |
|---|---|---|
| Your spreadsheet | Browser memory, for as long as the tab is open | Only you, on that machine |
| Autosaved workspace | IndexedDB, in your browser profile on your device | Only you. Clearing site data deletes it |
| Saved recipes and view settings | IndexedDB and localStorage on your device | Only you |
| Your licence key | localStorage on your device | Only you. It is verified by local cryptography, never checked against a server |
| Exported files and reports | Wherever you choose to save them | Only you |
There is no row in that table pointing at us, and that is the whole design. We cannot produce a customer's spreadsheet on request, because we never receive one.
Do not take the claims above on trust. Every one of them is checkable from the browser your team already has.
Open DevTools (F12) → Network tab → load a spreadsheet into Prism and let it analyse. You will see requests for the page's own scripts and fonts, all to prismreview.app, and no request to any other host. No upload appears, because none happens.
In the same Network tab, click the document request and look at Response Headers. The policy includes connect-src 'self', which instructs the browser itself to block any attempt to send data to another origin. This is not a promise in our documentation — it is a rule the browser enforces against our own code. If Prism tried to upload your file, the browser would refuse and log a violation.
Load Prism once, then turn off Wi-Fi and reload the page. Upload a file, run the quality checks, browse the dashboard, export a report. Everything still works. Software that needed to send your data somewhere could not do this.
There is a fuller technical walkthrough, including the exact policy we ship and the large-file benchmark, on the verify it yourself page.
Prism is a web page, and a web page can change. Everything above is true of the version we serve today, and you can confirm it in the moment using the steps above — but a future deployment could, in principle, ship different code. Any vendor telling you otherwise about a hosted web application is overselling.
Two things make that risk small and checkable. The Content-Security-Policy is served as a real HTTP response header on every request, so a change would be visible immediately in DevTools without reading a line of our code. And we set Cache-Control: no-transform, which prevents our own hosting provider from rewriting pages in transit — a protection we added after discovering the host silently injecting its own analytics beacon into pages we served. It was blocked by our policy, and it is now prevented outright.
No, and not as a policy choice — as an architectural one. There is no server-side component that could receive it.
Customer data is not hosted. It stays on the device that opened the file, in the jurisdiction that device is in. The static site itself is served from a CDN.
None for the application. If you choose to buy a licence, the payment is taken by Stripe on Stripe's own hosted checkout page — that is a separate page, involving your billing details and no spreadsheet data whatsoever. Stripe acts as merchant of record. Using Prism for free involves no third party at all.
They aren't — there is nothing to authenticate to. A paid licence is a cryptographically signed key that your browser verifies locally using WebCrypto. Activating it contacts nothing.
Clear the site data in your browser and every trace is gone from the device. There is no account to close and no deletion request to file, because there is no record of you to delete.
No. That is a genuine trade-off worth stating plainly: because nothing is transmitted, there is also no central log of who opened which file. If your firm requires centralised audit trails for document access, Prism will not provide them.
Because everything runs in browser memory, file size is bounded by the browser rather than by a server. We have tested to 500,000 rows / 38 MB with all views working and no interface freeze; the practical ceiling is around 640,000 rows. Locally stored work can also be evicted by the browser under storage pressure, so exported files remain the durable copy of any work you care about.
The usual reason a tool like this gets refused is that it uploads client files to a service the firm has not assessed. That objection does not apply here, and the three checks above let your IT reviewer confirm it themselves in less time than it takes to fill in a software request form.
Nothing to install, no account to create, and the free tier analyses files of any size.
Open Prism →