Privacy Policy

WorthIt judges whether a specific product is worth buying at its current price. The input is a product link from an online shop, a screenshot from a shop, or a product name.

The link, screenshot or name you enter is sent through Anthropic's API (Claude), which acts as a processor: it identifies the product, loads the shop page and looks up comparison prices at other shops. The data sent is not used to train models.

A screenshot contains whatever it captured — before uploading one, check that it does not include your personal data (name, address, cart contents, an email inbox in the background).

The finished result of a link or name analysis is kept for 6 hours in a shared cache in Firestore (Google/Firebase, europe-west3 region), with no user identifier of any kind. Anyone who enters the same link or name in the same language and currency during that time gets the same result without a new request to the model. Screenshots are never cached. Expired entries are deleted once a day. The only part of a result stored permanently is the price record described in section 4.

The sources card in a result loads the logos of the sites the analysis drew on directly from those sites — just as if you opened the page in a browser.

None of this is synced to the cloud. Deleting the app deletes it.

When you submit a product link, the price the server read from the shop's page is stored in Firestore (Google/Firebase, europe-west3 region), together with the product name, the shop name and the date. This lets another user who looks at the same product later see how the price actually moved.

No user identifier is stored with this record — not even an anonymous one. The archive therefore cannot reveal who viewed a product, nor link several searches together.

Only prices from a link the server fetched itself are stored. A price entered as text or taken from a screenshot never enters the archive. Alongside this, a count of how often each product was searched is kept so the app can show the most-searched products — again with no link to any user.

When you turn on price watching, the following is written to Firestore (Google/Firebase, europe-west3 region): an anonymous user identifier from RevenueCat, the product link, its name and image, the current and target price, currency, language and the date of the last check. No name, no email, no account — the identifier is random and not linked to your identity.

Once a day the server re-checks the product's price, and when it drops to your target price it sends a push notification via Firebase Cloud Messaging and the Apple Push Notification service. A device registration token for notifications is stored for that purpose.

You can turn watching off in the app at any time — the product record is deleted from the server immediately. Once you are no longer watching anything, the notification registration token is deleted too.

WorthIt Pro includes up to 20 analyses a week and the app’s interface is public, so the server counts analyses that actually call the model (a cached result does not count). Only a count and the time of the last change are written to Firestore (Google/Firebase, europe-west3 region): one counter per calendar week for the anonymous RevenueCat identifier (the same one used for price watching), and one per day for a fingerprint of the IP address. The fingerprint is an HMAC-SHA-256 with a key known only to the server — the IP address itself is not stored and cannot be recovered from the fingerprint.

An analysis pack (a one-time purchase) adds a balance of unused analyses to the same anonymous identifier, together with Apple’s transaction id so that one purchase cannot be credited twice. Unlike the counters, that balance is not deleted after 14 days — it stays until you use it up.

No analysis content is stored with the counters. The records are deleted automatically after 14 days.

WorthIt Pro subscriptions are handled by Apple via In-App Purchase and managed through RevenueCat. The app never sees or stores payment details — they stay with Apple. RevenueCat works with an anonymous identifier, not a name or email.

An analysis pack is a one-time Apple In-App Purchase. The server verifies it directly from Apple’s signature on the transaction; no payment details ever reach us.

The app offers a short survey (3 multiple-choice questions plus a free-text field), shown automatically after a few completed analyses, or any time from Settings → Feedback.

Answers are stored in Firestore (Google/Firebase, region europe-west3) along with the app version, language, and an anonymous RevenueCat identifier — the same one the app uses for price watching. No name or e-mail is stored; don't put anything in the free-text field you wouldn't want to share.

At the end of the survey the app asks whether you'd like us to get back to you. If you choose "Yes" and enter your e-mail, the support chat opens with your answers pre-filled. The message and e-mail are stored as a support ticket under section 4 of the general part; the survey answers themselves stay without an e-mail address.

To improve the app, it sends daily counts: how often each screen was shown, on which screen the app went to the background, and how often the WorthIt Pro offer was shown and for which feature it ended in a purchase. Only these totals are sent, without any user or device identifier and without links, screenshots or any content of your analyses. They are stored in Firestore (Google/Firebase, region europe-west3) and cannot be linked to a specific person.

Reach out via Support.