Plain-language data boundary
Measure the product.
Not your recipes.
PrepMargin is in early access. These notes explain what information is collected, why it is needed, and what is deliberately excluded.
Usage events
We record a randomly generated session ID that lasts for the current browser tab, the PrepMargin page visited, broad device size, section views, and named actions such as starting a calculation, saving a dish, opening an offer, or completing an import review. Campaign tags and the referring website's hostname may also be recorded so we can evaluate ads and search traffic.
The usage events generated by the product exclude dish names, menu text, ingredient names, recipe quantities, supplier values, uploaded filenames, or document contents. Browser “Do Not Track” is honored.
Feedback and mobile-app interest
Feedback type, message, page, mobile-platform preference, and session ID may be stored for product review. Email is optional and is stored only when you explicitly allow follow-up or request a future app invitation. Do not include confidential recipes, personal data, or payment information in feedback.
Payments
When checkout is enabled, PrepMargin sends Stripe your signed-in account identifier, contact email, and checkout selection. Stripe also receives the contact and payment information you enter on its hosted checkout. PrepMargin stores the signed-in account identifier, Stripe customer, checkout, and subscription identifiers, plan, status, and renewal date needed to grant and remove access. PrepMargin does not receive or store full card numbers.
A PrepMargin session ID may be included in Stripe metadata so a completed purchase can be attributed to the campaign that led to it. This can link usage events to your purchase and customer record, so these events are not fully anonymous. It does not contain menu, recipe, supplier, filename, or document content.
Smart import
CSV and text files are parsed in your browser and are not sent to OpenAI. For signed-in users, each selected PDF or image—up to three per bundle—is sent to OpenAI for processing under a monthly document allowance to extract a combined, reviewable draft. PrepMargin records account-level attempt and success counts plus estimated processing cost, but does not save the source files in your PrepMargin workspace or application database. This does not describe OpenAI’s separate processing or retention. Only the menu, recipe, supplier, and unmatched-row data you choose to save enters your workspace.
Service providers
Cloudflare hosts the application and its database. Sign-in supplies the account identity used to separate workspaces. Stripe handles checkout and billing, and OpenAI processes assisted PDF and image imports. These services process information needed to provide their functions; their own policies and applicable agreements govern their separate handling of data.
Your workspace
When you are signed in, PrepMargin stores your menus, recipes, ingredient purchase costs, and unmatched import rows in an account-isolated workspace so you can return in another session. A browser copy is kept for responsiveness and offline-safe editing. Without a signed-in identity, the workspace remains only on that device.
Retention and access
Workspace records are separated by PrepMargin user identity. From the workspace, a signed-in user can permanently delete synced menus, recipes, ingredient price history, unmatched rows, and the associated AI-usage record. Workspace deletion does not delete billing records, usage events, separately submitted feedback, or outreach opt-out records, and does not cancel a subscription. It also does not delete records held separately by service providers or automatically clear copies in other browsers. Use Feedback and choose “Other” for additional privacy requests.
Contact
PrepMargin is operated by Cairnfield Works LLC. For privacy, deletion, or access questions, use the in-product Feedback control or email support@cairnfieldworks.com. Do not email confidential recipes, supplier documents, or payment-card information.
Beta terms
Imported values and cost calculations remain operator-reviewed estimates. Read the early-access terms for acceptable use and current service limitations.