FeedMender

FeedMender/Privacy

Privacy

What happens to your file

This describes what the software actually does with the data you give it, taken from the code rather than from intentions. It is written to be checkable, not reassuring.

This draft must be reviewed by a qualified lawyer before launch. It is a factual description of the software's behaviour, not legal advice, and it deliberately asserts no company name, address, or registration. Those, and the formal data-controller language a privacy policy needs, have to be added and checked by counsel before this page goes live.

What we receive

When you run a scan, we receive the product file you upload and three settings you choose next to it: the currency, the selling country, and the language the feed is written in. That is all. There is no account to create, no name or email required to scan, and the free scan collects no payment details. If you buy a repair, Stripe collects the billing details, tax identifier if supplied, and payment details needed for Checkout. FeedMender receives the resulting order reference, billing country, email address, amounts, tax amount, and payment or refund status, but never your full card details.

Where it is stored, and for how long

The uploaded file is written to temporary storage on the server that runs the scan, under a random job identifier. It is kept only long enough for you to read the report and decide, because producing the fix needs the file a second time, and it is then deleted automatically after six hours. You do not have to wait for that: the report carries a control that deletes your file from the server immediately. Storage left behind by an earlier run of the server is swept on the same six-hour basis, so nothing lingers past it. An upload may be up to 60 MB and contain up to 5,000 products.

What leaves our server, and what does not

The free scan runs entirely on our own server and sends your data to no language model at all; it is deterministic. The only outbound requests it makes are to your own domain, and only to validate the feed: it fetches your product images, and it reads your robots.txt over HTTPS. Those requests come from our server, identified as feedmender/0.1, which is why your image host sees our requests rather than a visitor's. The scan does not crawl the rest of your site.

The paid repair uses Stripe to process payment. It is also the one step that may involve a language model service. To rewrite a thin or non-compliant title, selected fields from the affected products, not the whole file, are sent to a language model provider, Anthropic or OpenAI, with descriptions truncated to what the task needs. Every value the model proposes is checked against your own row before it is used. If you never buy a repair, none of this happens.

The corrections

The fix is delivered to you as a file. When you install it, it goes into a spreadsheet in your own Google account, and Merchant Center reads it from there through your account rather than through anything we host. The output remains on the same temporary job until its original six-hour deadline, so a failed browser download can be retried. The immediate delete control removes the whole job.

The pages of this site do not require a login and the scanner sets no cookies of its own. The current scan and order references are kept in the browser's session storage so the same tab can return from Stripe Checkout without putting a private access token in a URL. Closing that browser session removes this local recovery information. The site uses Google Tag Manager to serve tags for Google Analytics, for visitor statistics, and for Google Ads, for measuring advertising. These run in Google's Consent Mode: until you give consent through the cookie banner, every category is denied, no advertising or analytics cookie is set, and nothing is stored on your device except the entry that remembers your choice, kept for 180 days. Without consent Google may still receive basic, cookieless signals that carry no stored identifier, which Google uses for aggregate modelling. You can change or withdraw the choice at any time through Cookie preferences in the footer, and withdrawing removes Google's cookies. What the tags collect once allowed is governed by Google's privacy policy. The demo shown on the home page runs entirely in your browser and uploads nothing.

Getting your data removed

Use the delete control on the report to remove an upload straight away, or simply let the six-hour window pass. For anything else, write to hello@feedmender.com.