Local scenario storage
The calculator stores your working scenario on your device. Clearing browser storage, changing devices, or using another browser can remove or hide that local copy.
Privacy
FERSCalc is built around local scenario storage. This page explains what stays on your device, what the public site measures, and what happens when you export or share a scenario.
Privacy Model
Anonymous Analytics
Public pages use a narrow first-party analytics layer so the maintainer can tell whether people are finding the site and whether visitors are new or returning.
The calculator also sends a fixed allowlist of anonymous event names — for example, that the calculator opened, a local scenario resumed, or results were viewed — together with anonymous visit metadata. It never sends names, dates, balances, scenario inputs, or calculated results.
The public-page and calculator-usage channels use the same first-party anonymous visitor cookie. They do not use third-party ad tracking or try to follow you across unrelated websites.
Public pages also use Microsoft Clarity, a third-party tool from Microsoft that records anonymous session replays and heatmaps, so the maintainer can see which parts of these pages people actually read. It runs on the public pages only and is deliberately never loaded on the calculator, so the salary, TSP, and date-of-birth values you enter there are not part of any recording.
Local-Only Details
The calculator stores your working scenario on your device. Clearing browser storage, changing devices, or using another browser can remove or hide that local copy.
Exported files are created in your browser and saved only if you choose to save them. FERSCalc does not keep a server-side file cabinet of your exported reports.
A share link encodes a point-in-time scenario snapshot in the URL you choose to copy. Opening one prompts before it replaces your current local scenario, and later edits do not update links that were already copied.
Need Model Scope?
Privacy is only one part of trust. For modeled components, known gaps, and when to treat output as directional, use the canonical assumptions and coverage page.