Privacy Policy

Last updated: October 20, 2026

1. Introduction

VeriFined is committed to protecting your privacy. This Privacy Policy explains how we collect, use, disclose, and safeguard your information when you use our tracking and verification platform, including our website, APIs, and associated applications (collectively, the "Service").

We are based in the European Union and comply with the General Data Protection Regulation (GDPR) and other applicable data protection laws.

2. Information We Collect

2.1 Information You Provide

  • Account Information: Name, email address, password, and company details when you register. We also store which of our sites you signed up on (for example verifined.com or rivetchain.com), how you signed up (website form, Google, Shopify install or team invitation) and, if you accepted cookies, the campaign (UTM) parameters, referring website and first page visited described in section 2.2, with your account. If you turn on two-step sign-in, we also store the authenticator secret (encrypted) and your one-time recovery codes (only as one-way hashes, never in readable form). If your workspace was created by installing our Shopify app, we fill in its name, description and website once from your store's details in Shopify (the store name, its store description and its main web address); you can change or clear them in Settings, and we do not fill them in again
  • Product Data: Information about products you register, including descriptions, metadata, images, and serial numbers
  • Ownership Claims: When someone claims or receives a product, we store the first and last name, email address and optional phone number they enter. Public verification pages show only an abbreviated name and a masked email address and phone number.
  • Data About Your Customers (Shopify app): When a merchant connects a Shopify store, we receive order data for the products we certify — the buyer's name, email address, phone number and shipping address, and the products ordered. We process this on the merchant's behalf to issue and email certificates of authenticity; the merchant is the controller for this data.
  • Newsletter: Your email address, if you subscribe to our newsletter and confirm the subscription
  • Custody Events: Chain of custody information including timestamps, actors, and notes
  • Support Communications: Messages and attachments sent through our support system
  • Payment Information: Plans bought on our website are paid through Stripe, which processes your card details — we never see or store your full card number. Plans bought through the Shopify App Store are billed by Shopify.

2.2 Information Collected Automatically

  • Usage Data: API calls, verification requests, and feature usage patterns
  • Device Information: Browser type, operating system, and device identifiers
  • Log Data: IP addresses, access times, and pages viewed
  • Cookies and Similar Storage: We store your login session and your cookie choice in your browser's local storage. Only if you click "Accept" on the cookie banner do we load Google Analytics 4 and the Google Ads tag, which set cookies to measure site visits and ad conversions; if you click "Reject", they are not loaded. You can change your choice at any time with the "Cookie settings" link in the footer of our website pages, or by clearing this site's data in your browser. Only if you click "Accept", we also keep in your browser the campaign (UTM) parameters in the link that brought you here, the website that referred you and the first page you visited, to know how people find us; if you reject or have not chosen, we neither keep nor use them.
  • Email Records: For each email we send, we keep a record of its type and delivery status (delivered, opened, clicked or bounced, as reported by our email provider) with a masked recipient address and a keyed hash of the address — never the email's content

2.3 Brand Guard

Brand Guard is an optional VeriFined feature that lists a merchant's official websites, detects copies of their store and prepares reports to get copies removed, and can check the merchant's own website every day for changes they did not make (Site Watch). When a merchant uses it, we process:

  • Official websites: the website addresses the merchant lists; how we confirmed they control them (their Shopify store's domain list; for the store's primary domain, also its public DNS records, looked up through the public DNS-over-HTTPS resolvers below, and the store's public shop details that the domain itself serves, such as its Shopify shop ID; or a DNS TXT record we look up through public DNS-over-HTTPS resolvers); the results of our daily re-checks; and the public registration details of those domains (registrar, creation date, name servers), so we notice if a domain changes hands. Our team reviews every new website before it is published, and we keep the review decision and notes.
  • Business check: for a website that is not confirmed through Shopify, the merchant's owner can ask us to verify the business behind it ("Verify your business"). We store what they send: the business's legal name, its registration number, its country and, if they add them, the address of a page on their website that shows these details and a note; who sent it and when; and our decision, who on our team made it and, if we could not verify the business, our reason. A person on our team compares these details with the public company register of that country; we ask for no documents. These details are shown only to the merchant's account and our team, never on the public brand page, and we email the merchant's owners the outcome. We keep them with the account's Brand Guard websites and delete them together (section 5.4). For a business run by one person, its name and number can be personal data; the legal basis is the performance of our contract with the merchant.
  • Other addresses for the daily email: the merchant's owner can add up to 5 email addresses, for example a colleague's or a shared inbox, to get the Brand Guard daily email (below). We store each address, who added it and when, and when its holder confirmed it. We send it one email with a confirmation link, and no Brand Guard email until its holder confirms; if they don't, the link stops working after 7 days. Each daily email to it has a link to stop it. An address is deleted when its holder unsubscribes, when the owner removes it, and with the account's Brand Guard data (section 5.4). These addresses don't get the weekly summary. The legal basis is the performance of our contract with the merchant and, for the holder, their confirmation.
  • Public brand page: once approved, the brand name, its page address (verifined.com/v/<alias>), its official websites and the full history of changes to that list are public on verifined.com by design, and the website check at verifined.com/check tells anyone whether an address is on a brand's list.
  • Change history on the blockchain: each change to a public list is recorded as a salted commitment (a keyed hash of the website address and of the brand's page name) with its type, a sequence number, the time and a fixed reason code, and anchored to the Solana blockchain in our hourly batch (see section 5.1). Each entry has its own random salt, stored encrypted on our servers; for entries shown on a public brand page, the downloadable proof includes the salt so that anyone can check the entry. See section 5.4 for what happens when this data is erased.
  • Visitors to a merchant's store (badge and hidden files): if a merchant switches on the Brand Guard badge in their Shopify theme, their store's pages load a small badge frame and three small hidden files (an image, a style sheet and a font) from verifined.com. Each request reaches us with the visitor's IP address and the address of the website that loaded it. We keep only that website's host name, never the page path or query. We shorten the IP address to its network (the first 24 bits of an IPv4 address, the first 48 bits of an IPv6 address) and store it only as a keyed hash; the key is random, made for one day and deleted after 2 days, after which the stored hashes can no longer be linked to any network. The badge sets no cookies and uses no local storage or fingerprinting. We use this only to detect copies of the merchant's store running on other websites; the legal basis is our and the merchant's legitimate interest in preventing fraud. The merchant must mention this in their own store's privacy policy.
  • Lookalike websites: for a website that may copy a merchant's store (found through the badge and hidden files, through a report, through our misspelling sweep, through our new-domain watch, through our photo search, through a link in a message that a visitor checked on verifined.com/check, or added by the merchant), we look up public technical data: its DNS records, its domain registration data and server network from RDAP (the registries' public registration data service), and the network's AS number. We keep only the domain and server details, the name of the registrar and of the network, and their abuse role contacts (such as an abuse@ mailbox and phone number), never the name, address or contact details of the person who registered the domain. Our team may also set keywords for a brand (distinctive words such as its brand name, never generic words); we match them against lookalike website addresses to score them. These lookups never contact the lookalike website itself; the only visits to it are the captures described below.
  • Misspelling sweep: from the merchant's listed official domain names and their company alias, we generate names a copier could register (misspellings, look-alike characters, the name with a word such as "official" or "shop" added, and the name on other domain endings). We check whether each name is registered by asking public DNS-over-HTTPS resolvers (Cloudflare and Google); the most likely names are checked every day, the others weekly or monthly. The first time a name appears registered, we confirm it once with RDAP. Only the domain name is sent. We do not visit these names: a registered name with a live address becomes a lookalike website on the merchant's list and is handled as above. For each name we keep only its last check result and dates, while the account uses Brand Guard.
  • Captures of lookalike websites: when a lookalike website is live, is not a parked domain and its signals are strong enough (or the merchant added it, or the badge or hidden files saw it), we open its public home page in an automated browser that runs in an isolated environment on AWS (AWS Lambda, us-east-1). We do this once when we first find it, again every week while it stays open with a meaningful score, and every 3 days while a removal case against it is open; a capture that fails or times out is retried, at most 3 times within 7 days. We store: screenshots (the first screen and the full page), the page as it was rendered (its HTML), its title, an excerpt of its text (up to 8,000 characters), fingerprints of its images (perceptual hashes) and small copies of up to 4 images that match the merchant's own photos, and technical data (the final address after redirects, the response code, timings, the shop platform, analytics and advertising IDs and Brand Guard badge IDs found in the page, and whether the page has password, email or payment-card fields). A capture shows the public page as any visitor would see it, so it can contain whatever the site's operator published there, including personal data. The browser does not log in, click, type, fill in or submit forms, or buy anything; it starts without cookies and does not load analytics, advertising or chat services. To refresh a website we already captured for this merchant, if another Brand Guard customer captured the same public address in the last few hours, we may reuse that capture's screenshots and page facts instead of opening the site again; a merchant's first capture of a website is always taken fresh, and nothing about either customer is shared with the other. We compare the capture with the merchant's own website (below) to compute automatic signals such as matching photos or the store's name in the page title, which raise the website's private risk score by fixed amounts. Captures are shown only to the merchant's account (and to our support team when it signs in to help the merchant; each such session is logged).
  • The merchant's own website (reference): once a week we capture the home page of the merchant's main official website in the same way. From it, and from up to 30 product names already in the account when it is connected to a Shopify store, we keep a reference: the store name and its variants, fingerprints of up to 200 images, the logo and icon fingerprints, analytics and advertising IDs, Shopify file paths, and a fingerprint of the page text.
  • Phone-size screenshot and evidence image: each capture also includes a screenshot of the same loaded page at a phone-screen size; no further request is made to the website for it, and it is kept with the capture. When a merchant opens a removal case, and again after each new capture while the case is open, our isolated function on AWS Lambda composes one image from captures we already hold: the capture of the merchant's own website next to the capture of the reported website, with up to 4 pairs of matching photos below. Its captions state only the two website addresses, the capture times and a number for how alike two photos look. It is shown to the merchant, who can download it, and on the case's evidence page (below); it contains nothing that is not already in the captures, and a newer image replaces the older one.
  • Reference photos and photo search: the merchant can choose up to 20 of their own product photos and one logo as reference photos. While a merchant has Brand Guard on, their Shopify store is connected and they have never had a reference photo (when they turn Brand Guard on, and once a day after that), we pick up to 8 for them by default: the main photos of their best-selling products of the last 60 days of orders (we read only which products were ordered and how many, not who ordered them), then of their most recently updated active products; the merchant can remove them or add others, and we do not pick again once they have added or removed a photo. Reference photos can be photos of their Shopify products (we keep the photo's address on Shopify's servers, not a copy), files they upload (stored in our file storage) or product pictures from the weekly capture of their own website (a copy is stored). For each photo we keep its size and fingerprints (perceptual hashes and a checksum), computed by our isolated function on AWS Lambda, which downloads the photo for that purpose. We use the reference photos in two ways. First, we compare them with the pictures in captures of lookalike websites (above). Second, unless the merchant switches photo search off, about once a week we send each reference photo, or a link from which to fetch it (the photo's address on Shopify, or a short-lived link to our copy), to Google Cloud Vision's web detection (see section 6), which answers with the addresses of public web pages that show the same or a similar picture. For each page we keep its address (without any query string), its host name, its title (up to 300 characters), how it matched, when we first and last saw it, and whether it is on a marketplace, a social network, one of the merchant's own websites or a website the merchant allowlisted; image-hosting addresses are dropped. Marketplace and social-network pages are only listed to the merchant; the website of any other page becomes a lookalike website on the merchant's list (at most 20 new websites per search) and is handled as above. We never open a page that Cloud Vision names: the only visit is a capture, if the website qualifies for one. We keep Cloud Vision's answer for a photo for 6 days, so that the same photo is not searched twice, also when another merchant uses the same file. A photo the merchant removes is no longer searched: its file is deleted when it is removed, and only its fingerprints are kept until the account's Brand Guard data is erased (section 5.4), so that adding it again does not repeat the search. An upload that is never completed is deleted after one day. A page that shows a merchant's photo may be a reseller, a review or a licensed use: photo search gives indications, not findings, and the merchant decides. The legal basis is our and the merchant's legitimate interest in detecting impersonation and fraud against the merchant and their customers (GDPR Art. 6(1)(f)); for the merchant's own photos it is also the performance of our contract with the merchant.
  • Site Watch (the merchant's own website): a merchant can switch on Site Watch for an official website that they verified as their own through Brand Guard; no other website can be watched. We then open the pages they choose (when the website is served by their Shopify store connected to our app, we suggest pages from it: its collections, with their names, addresses and numbers of products, and its best sellers, which we work out from the products and quantities in up to 250 orders of the last 60 days; we read no customer, address or payment details from those orders, and we keep only the pages the merchant chooses). We open each chosen page on a daily schedule, at about the same time between 04:00 and 09:00 UTC, and again when the merchant clicks "Check again". Each check opens the page twice in our isolated automated browser on AWS Lambda (us-east-1): once as a desktop browser, whose same page load is then also resized to a phone-screen width (a resized view, not a separate visit from a phone), and once with a browser identification (user agent) that presents itself as Google's smartphone crawler, Googlebot, to see whether the page shows search engines something different. These visits come from our servers, not from Google. The browser identification of each of them, and of the request below, also names our service ("VeriFined-SiteWatch", with a link to a page that explains it), so the website's operator can recognise them. Alongside the desktop visit we send one plain request for the page that says the visitor came from Google search, to see whether search visitors are sent elsewhere. The browser starts without cookies; it does not log in, click, type, fill in or submit forms, add to cart or check out, and it does not load analytics, advertising, chat and several marketing, newsletter and review widgets; checkout, account and cart item pages cannot be watched. It does not read the website's robots.txt file, because it visits only pages the website's own operator chose. The website's server logs and analytics may record these visits, including, once per page and check, one that appears to come from Google search. For each page and check we store: screenshots (the full page and the first screen as a desktop browser, the full page at phone width, the screen when a pop-up window appeared, the page's main picture, and the full page as seen with the crawler's user agent), the page as it was rendered (its HTML), and what we extract from it: the text of its parts (up to 400 characters each), its links with their text and where they lead, the addresses of its scripts and fingerprints of the scripts written into the page, embedded frames, fingerprints of up to 40 of its images, the prices and products it publishes, its contact links (phone numbers, email addresses, WhatsApp and social-media links), its payment and checkout links, its pop-up windows and banners, and any redirects. From these we keep a compact summary of each page and a record of what is normal on it, which we compare with later checks to find changes (such as hidden links, a new script, a different payment or contact link, or a changed price, picture, pop-up window or text) and to check the merchant's checklist. A capture shows the public page as any visitor would see it, so it can contain personal data published there, such as the merchant's own contact details or the names in customer reviews; we process it on the merchant's instructions (they choose the website and its pages, and they decide what their website publishes) and use it only to provide Site Watch to that merchant: it is never used for anything else, never compared across merchants and never shared, except with the processors in section 6. Captures and findings are shown only to the merchant's account (and to our support team when it signs in to help the merchant; each such session is logged). If your name or other data appears on a merchant's website watched by Site Watch, the merchant runs that website and published it there; you can ask the merchant, and you can also write to us (section 7) to learn what we hold and have our copies of that page deleted. We record which team member marked a change as made by them, as not made by them, or as a part to ignore, and the team members the owner chose to receive Site Watch emails. The legal basis is the performance of our contract with the merchant; for personal data of other people that appears on the merchant's pages, it is our and the merchant's legitimate interest in keeping the merchant's website secure (GDPR Art. 6(1)(f)).
  • Site Watch and the merchant's Shopify store: when the watched website is served by a Shopify store connected to our app, we read from Shopify, with the access the app already has, about once a day and before a check: the list of files of the store's published theme (file names, checksums, sizes and when each changed) and the text of its theme files (up to 256 KB each), so we can tell the merchant that a change on the website followed a theme edit and show them the edited lines; the products updated recently (their title, address, status, when they were created and updated, and their variants' SKUs and prices, with the store's currency), so we can recognise new products, hidden products and price changes the merchant made; and the store's main domain. Theme files contain whatever the merchant put in their theme, which can include their own contact details. We keep copies of the theme files in our file storage (Amazon S3) and a record of each theme or product change. We cannot see, and do not read, who made an edit, the store's installed apps or its script tags.
  • Site Watch AI notes and checklist: after a check that found new changes, we may send Google's Gemini (see section 6) up to 25 of them, to get a short note on each: whether it looks expected, unexpected, like a security problem or like noise, with one sentence. For each change we send its kind, the names of its pages, the part of the page, the text before and after (up to 300 characters each), the website addresses involved, which of our fixed rules matched, the matching Shopify activity and, for a changed main picture, the old and new picture as small thumbnails (which the model reads at a reduced resolution). For checklist items (rules the merchant writes in their own words, such as "the newsletter pop-up shows on the home page") that our fixed rules cannot check, we send up to 15 items at a time with the visible text of their pages (up to 6,000 characters per page), their pop-up windows and up to 4 screenshots of their first screen or pop-up window (which the model reads at a reduced resolution), and the AI answers whether each item holds, or that it cannot tell; when the merchant adds an item, we send its text with the names and addresses of the watched pages, so the AI can pick the pages and the kind of check. The page text can contain personal data published on the merchant's pages (above); we send nothing from the merchant's account about their team or their customers. We keep each note (shown only when it contains no contact details, no percentage and no website address other than the merchant's own), the model and when it was made, and each checklist item's latest result and reason. A note never decides whether a change is a possible hack, never changes a change's status and is never put in an email. A checklist result given by the AI does decide whether that item shows as failed, including in the daily email. This is information for the merchant about their own website, which the merchant reviews and acts on themselves; we take no action on it, and it is not a decision with legal or similarly significant effects on the merchant or on anyone else (GDPR Art. 22). The number of AI calls per merchant is limited per day and per month; a check without an AI call still works, and an item that could not be checked shows as not checked. The legal basis is the performance of our contract with the merchant; for other people's data on the merchant's pages, it is our and the merchant's legitimate interest in keeping the merchant's website secure (GDPR Art. 6(1)(f)).
  • If your website or your data appears in a capture: we capture public pages of websites that look like a Brand Guard merchant's store, and we have no way to tell the people behind those pages. We may also send a capture to Google's Gemini for an AI opinion (see "AI opinion on lookalike websites" below). Photo search can also record the address and title of a public page that shows a merchant's photo (above). If you operate a website we captured, or your name or other data appears on one, write to contact@verifined.com. We will tell you what we hold about the address and delete our captures of it and the AI opinions made from them, and the photo-search records of its pages, unless a merchant's open removal case still needs them as evidence; in every case a capture is deleted 90 days after it was taken. You can object to this processing at any time (section 7).
  • New-domain watch: certificate authorities publish every HTTPS certificate they issue in public Certificate Transparency logs. Every hour we read new entries of the logs listed in Google Chrome's public log list that publish them as downloadable files (most new certificates, not all), and compare the website names in each certificate with the merchant's listed official domain names, their company alias, the keywords our team set for the brand, and our own brand names, plus one fixed generic test word whose matches we only count (we keep no website name for it), to check that the watch works. We download the logs; we send them nothing about the merchant or anyone else. When a name matches, we keep the website name and when its first certificate was issued; a matching website becomes a lookalike website on the merchant's list and is handled as above. Names that look like our own brands go to a list only our team sees. We keep nothing else from the logs. For lookalike websites found in other ways, we may look up when their first certificate was issued on crt.sh, a public certificate search service, by sending it the website name. The legal basis is our and the merchant's legitimate interest in detecting impersonation and fraud against the merchant and their customers (GDPR Art. 6(1)(f)).
  • Why we sweep and capture: to find and document copies of the merchant's store, so the merchant can decide whether a website is a copy and support their removal reports with evidence. For the misspelling sweep and the lookups of lookalike websites, the legal basis is our and the merchant's legitimate interest in detecting impersonation and fraud against the merchant and their customers (GDPR Art. 6(1)(f)); for the sweep, the data sent is the domain name only. For captures of lookalike websites it is the same legitimate interest; for the merchant's own website it is also the performance of our contract with the merchant.
  • AI opinion on lookalike websites: after a capture of a lookalike website, unless our own checks already found strong matches or the page is parked or empty, we may ask an AI model, Google's Gemini (see section 6), for an opinion: whether the website is likely a copy of the merchant's store, a reseller or review page, an unrelated business, or a parked or empty page, and how likely a copy is. We send it the capture's screenshot (which the model reads at a reduced resolution), the page's title and text excerpt, the website's address, the age of its domain, its registrar, its hosting network and country, whether it sits behind Cloudflare, its shop platform and any checkout or payment-card fields, and the automatic signals we measured; and, from the merchant's reference, the store name and its variants, the official website addresses, up to 30 product names and up to 3 of the store's own product photos. We do not send the names or contact details of the merchant's team or of their customers. We keep the opinion (a label, a likelihood, a short summary and up to 6 points it relied on, which can quote the page), the model used and when it was made and, once the merchant decides about the website, whether that decision agreed with the opinion, so we can measure how reliable the opinions are. When our own checks already found strong matches, we record a rule-based opinion instead and send nothing to the AI. The account owner can switch AI opinions off for their brand; existing opinions are then hidden and no new ones are made, and they are deleted on the schedule in section 5.4.
  • Automated suggestions, not decisions: an AI opinion helps the merchant sort and read the list, and decides one thing on its own: which lookalike websites are named in the merchant's daily alert email. A website is emailed once an opinion (the AI's, or our own rules') rates it a likely copy, or once the merchant's own store files were seen loading on it and no opinion says the page is unrelated or parked; every other website stays in the merchant's list, unemailed, until the merchant looks at it. The opinion itself adds no points to a website's risk score, changes no status, and never opens a removal case or sends anything to anyone but the merchant. Whether a website is a copy, and whether to report it, is always the merchant's decision, and we take no action against any website. Neither the opinion nor the alert email is a decision with legal or similarly significant effects on the merchant or on the website's operator (GDPR Art. 22): it is information to the merchant, who reviews each website before acting. The legal basis for sending this data to the AI is our and the merchant's legitimate interest in detecting and stopping impersonation and fraud against the merchant and their customers (GDPR Art. 6(1)(f)): it helps the merchant assess and report likely copies of their store. For the merchant's own store data and case facts it is also the performance of our contract with the merchant.
  • Weekly summary email: we may email the addresses that get Brand Guard alerts (the account's owners and the contact address of its connected Shopify store) a weekly summary of what Brand Guard found in the last 7 days, such as new lookalike website addresses and why they were flagged, captures that matched the store's photos or name, changes to the brand's official websites and removal-report updates. It contains no risk scores and no AI text. When someone unsubscribes from it, we keep only a keyed hash of their address with the account, not the address itself, so we can leave them out (see section 3).
  • Removal reports: when a merchant prepares removal reports for a confirmed copy, we store the statement they confirm (its version and time), the account that confirmed it, the full name and role they type, a keyed hash of their IP address, the prepared report texts, and which reports they marked as sent (with any reference number). The merchant sends every report themselves; we send none. Once a registrar has not answered a report for 14 days, we also prepare a complaint text for ICANN Compliance (the body that oversees registrars) from the same case facts and the dates and outcomes of the earlier reports; the merchant sends it through ICANN's own form, and we keep the draft and, if they mark it as sent, the record with the case like any other report. A case's evidence page, which the merchant can share with the recipients for 14 days, shows the reported website's technical details, which automatic signals matched, the evidence image and the brand's official websites, but not the signatory.
  • "Report a fake site" form: the reported website address, the brand named (optional) and a keyed hash of the sender's IP address. We ask for no name, email address or message.
  • Website check (verifined.com/check): for a link you check, we do not store the address, and our request logs leave it out; we keep only a count of checks. For a message you paste ("Check a message"): we read it once on our server to find the web addresses in it (up to 10) and compare them with the websites that brands have verified. The text, and the shop name you may type next to it, are used for that request only: they are not stored, not written to our logs or error reports, and not sent back to you. We never open, visit or expand a link in a message. If a link's host name closely resembles a Brand Guard brand's official website, we may record that host name (only the host name, with the date; never the message or anything about you) as a lookalike website on that brand's list, at most 10 per brand per day; it is then handled as above, which can include a capture of that website, and the brand sees it as "Shopper check". The check is protected by Cloudflare Turnstile (section 6). To limit use, we count checks per IP address in our server's memory for up to one day, and, for the explanation below, keep a keyed hash of the day and your network (the first 24 bits of an IPv4 address, the first 48 bits of an IPv6 address) for 36 hours; it cannot be turned back into your address.
  • Explanation of a pasted message: when it is available, we may also send the text of your message (up to 3,000 characters, together with the result for each link and the shop name you typed, never your network address or anything else about you) to an AI model, Google's Gemini (see section 6), for fixed warning signs and hints on what to do and, when the check is in English, a short written summary. The summary is text written by the AI; it is shown as a guide, can be wrong, and is never allowed to call a message or a website safe. We keep neither the request nor the answer. This happens at most 10 times a day per network and within daily limits; otherwise the page answers without an explanation. The legal basis is our legitimate interest in helping shoppers avoid fraud (GDPR Art. 6(1)(f)); you choose whether to paste a message.

3. How We Use Your Information

We use the collected information for:

  • Service Delivery: Providing product authentication, verification, and chain of custody tracking
  • Ownership Management: Processing ownership claims, transfers, and OTP generation
  • Account Management: Managing your account, subscriptions, and preferences
  • Communication: Sending service updates, security alerts, and support responses
  • Setup Reminders: If you install our Shopify app but haven't used it yet, we may email the store's address (as listed in Shopify) a short reminder — at most three times. Every reminder has a one-click unsubscribe link, and reminders stop once you use the app or uninstall it.
  • Newsletter: If you subscribe and confirm your subscription, we send you our newsletter. Every newsletter has a one-click unsubscribe link.
  • Brand Guard Weekly Summary: If your account uses Brand Guard, we may send its alert addresses a weekly summary (section 2.3); a quiet week usually sends nothing. Each summary has a link that stops only the summary, and the summary can also be switched off in Brand Guard settings. If you instead use our email provider's own unsubscribe link in a summary, we stop the summary (and any newsletter or setup reminders to that address) and keep sending Brand Guard alerts and removal-report reminders, which are part of the service.
  • Brand Guard alerts: If your account uses Brand Guard, we email its alert addresses (the account's owners, the contact address of its connected Shopify store, and any other addresses the owner added that confirmed, see section 2.3) at most once a day about lookalike websites rated a likely copy of the store (section 2.3), and reminders about open removal reports. For a website watched by Site Watch, the same daily email also lists possible hacks (first), other changes and failed checklist items, unless the owner chooses to see them only in the app; there is no separate Site Watch email, and the Site Watch part of that email also goes to the team members the owner chooses. These are part of the service and have no unsubscribe link, except in the email to an address the owner added, which can stop it; they stop when Brand Guard stops for the account or, for Site Watch, when the website is no longer watched.
  • Improvement: Analyzing usage to improve our Service and develop new features
  • Published statistics: We may publish anonymous, aggregated statistics about how the Service is used, for example in our blog articles. They cannot be used to identify a company, an account, a store, a product owner or a shopper, and each published figure combines the data of at least five companies.
  • Security: Detecting fraud, preventing abuse, and ensuring platform integrity
  • Legal Compliance: Meeting regulatory requirements and responding to legal requests

4. Legal Basis for Processing (GDPR)

Under GDPR, we process your data based on:

  • Contract Performance: Processing necessary to provide the Service you requested
  • Legitimate Interests: Improving our Service, producing anonymous statistics (section 3), preventing fraud, and ensuring security
  • Legal Obligations: Compliance with applicable laws and regulations
  • Consent: Where you have given explicit consent for specific processing activities

5. Data Storage and Security

5.1 Where Your Data Is Stored

  • Hosting: The Service runs on Amazon Web Services (AWS) in the United States (us-east-1, N. Virginia) — application servers, the PostgreSQL database, cache and file storage (Amazon S3). Data is encrypted at rest.
  • Public Blockchain and IPFS: Only for products where blockchain anchoring is turned on (it is off by default): once an hour we publish a Merkle root to the Solana blockchain and an IPFS copy of the batch record (event hashes, event types, timestamps and internal ids — never event contents, names, buyer email addresses or other personal data). Batch records published before we made this change in October 2026 also contain the merchant's company name, which for a sole trader can be their own name. These public records cannot be changed or deleted. Brand Guard is the exception to "off by default": every change to a brand's public list of official websites is always anchored, as a salted commitment that contains no website address, name or other readable personal data (see section 2.3).
  • International Transfers: Our hosting and several of our providers are in the United States, so personal data leaves the EU/EEA. With each provider we have a contract with, we rely on the European Commission's Standard Contractual Clauses or, where that provider has certified under the EU-U.S. Data Privacy Framework, on that certification. The public lookup services in section 6 receive only domain names and server addresses, without a contract (see section 6). What Brand Guard sends to Google's Gemini (section 2.3) is processed by Google in the United States and elsewhere. For that transfer we rely on the terms Google attaches to the Gemini API (see section 6), which for the paid service include Google's data processing terms with the European Commission's Standard Contractual Clauses. What Brand Guard sends to Google Cloud Vision (section 2.3) is processed by Google in the United States and elsewhere under Google Cloud's standard terms (we have no separately negotiated agreement); Google states that the Vision API complies with its Cloud Data Processing Addendum, which provides for the European Commission's Standard Contractual Clauses for such transfers unless Google adopts another approved transfer mechanism.

5.2 Security Measures

  • Encryption in transit between your browser and our servers (HTTPS with TLS 1.2 or higher; plain HTTP is redirected to HTTPS) and at rest (AES-256) for the database, cache, backups and file storage
  • Reasonable security practices appropriate for the nature of the Service
  • Access controls and authentication mechanisms
  • Rate limiting and error monitoring
  • Encrypted automated backups (see section 5.4 for how long they are kept)

5.3 Order Records Used for Authenticity Verification

When a merchant uses VeriFined with their Shopify store, the order records we use to verify a product's authenticity are kept in a tamper-evident ledger. New records store only a one-way fingerprint of the buyer's email address, never the address itself. When we receive a deletion request for that buyer, the key needed to link the fingerprint to the email address is deleted, so the fingerprint can no longer be linked to that person. Older records may retain the buyer's email address, or the name an owner gave when claiming a product, where this is required to preserve the integrity of the ledger (GDPR Art. 17(3)).

5.4 How Long We Keep Data

  • Account, company and product records: for as long as your account exists, subject to the ledger-integrity exception in section 5.3
  • Shopify order data: kept with the products and certificates created from those orders, which belong to the merchant's account. When a customer asks the merchant to delete their data through Shopify, we remove that customer's details from those records (subject to section 5.3). When a merchant uninstalls the app, Shopify asks us 48 hours later to delete the store's data: we then delete the store connection and its email records, and remove the buyers' names, email addresses, phone numbers and shipping addresses from the products created from that store's orders, together with the recipient names and PDF files of their certificates (unless the account still has another connected store). If the workspace was created automatically for your store when you first opened the app (you never connected a separate web account), the workspace itself is deactivated as well, and permanently deleted a further 30 days later, together with its products and certificates, except for billing records and any records kept in the tamper-evident ledger under section 5.3 (those entries, which contain no readable personal data, and the workspace record they belong to, are kept). Otherwise the products stay in your account — email us to have them deleted.
  • Email records: 180 days, or sooner when Shopify asks us to delete a customer's or store's data (entries logged before October 10, 2026 cannot be linked to a store, so those expire after 180 days)
  • Support requests: the request, our replies, attachments and the emails we received are kept for 24 months after their last activity and then deleted; when Shopify asks us to delete a customer's data, we also delete the support requests and emails from that customer's address, and you can email us to have yours deleted sooner
  • Newsletter: until you unsubscribe
  • Backups and previous file versions: automated backups of the database and of the cache are kept for 7 days and then deleted automatically. A file we delete from our file storage (such as a certificate PDF, an image or an upload) is kept as a previous version for up to 90 days and then removed. Until these copies expire, they can still contain data we have deleted, including the keys described in section 5.3; we use them only to restore the Service after a failure.
  • Brand Guard websites and change history: for as long as the account uses Brand Guard. The anchored commitments can never be deleted. When a workspace that was created automatically for a Shopify store is deleted after the store uninstalls the app (see above), or a workspace is permanently deleted, we delete the salts and the readable copies of the websites and page name, so the commitments can no longer be linked to anyone. A web account keeps its change history; email us to have it erased.
  • Brand Guard badge and hidden-file signals: unconfirmed signals are discarded within 8 days; IP address hashes are deleted after 30 days (and cannot be linked to a network after 2 days); records of the merchant's own websites after 30 days and of other websites after 90 days
  • Lookalike websites: while they are on the merchant's list, and deleted with the workspace; a website known only from a "Report a fake site" submission that the merchant never acted on is deleted after 30 days
  • Removal report cases: one year after the case's last activity, together with the statement, the signatory's name and role, the IP address hash and the report texts
  • "Report a fake site" submissions: 90 days
  • Brand Guard captures (of lookalike websites and the weekly reference capture of the merchant's own website; Site Watch captures have their own line below): 90 days after each capture is taken; the screenshots, page copy and image copies are then deleted together with the record
  • Brand Guard reference of the merchant's own website: replaced at each weekly capture; deleted 30 days after Brand Guard stops for the account, or sooner when the workspace's Brand Guard data is erased (see below)
  • Brand Guard AI opinions: the summary and the points the opinion relied on are deleted 90 days after it was made; the label and likelihood (and whether the merchant's decision agreed) are kept for one year, then deleted with the record. They are also deleted when the website's captures are erased at the request of its operator (section 2.3), and when the workspace's Brand Guard data is erased (see below).
  • Brand Guard weekly summary unsubscribes: the keyed hash of the address is kept while the account exists and deleted with the workspace's Brand Guard data; each summary sent is in our email records (above)
  • Brand Guard misspelling sweep: the list of checked names and their results while the account uses Brand Guard, deleted 30 days after Brand Guard stops for the account
  • Brand Guard new-domain watch: a matching website, with its certificate date, is kept as a lookalike website on the merchant's list, like any other; the hourly files of matching names are deleted after 30 days; names that look like our own brands are deleted 180 days after we last saw a new certificate for them
  • Brand Guard reference photos: while the merchant uses them; the file of a removed photo is deleted when it is removed, and its fingerprints are kept until the workspace's Brand Guard data is erased (see below); an upload that is never completed is deleted after one day
  • Brand Guard photo search: each page found is deleted 180 days after we last saw it in a search result; Cloud Vision's answer for a photo is kept for 6 days
  • Brand Guard evidence image: 90 days after it was composed, or sooner when a newer one replaces it or the case is deleted; the phone-size screenshot is deleted with its capture
  • Message check (verifined.com/check): nothing from the message is kept; a host name recorded as a lookalike website is kept like any other (above); the per-network counts in our server's memory last at most one day, and the keyed hash for the explanation 36 hours
  • Brand Guard Site Watch: the screenshots, page copies and extracts of each check are deleted 30 days after the check, except the latest successful capture of each page and view, which is kept while the page is watched and the website stays on Site Watch (also while it is paused); when the merchant takes a page out of the watch, its captures are deleted 30 days after their check like any other; the summary of each check (including its checklist results) after 90 days (a check whose latest capture is still kept waits until that capture is replaced); changes while they are open or marked as not made by the merchant, and 180 days after they were resolved; the record of what is normal on each page, the checklist (with removed items) and the choices of pages, email recipients and parts to ignore while the website is watched; AI notes go with their change
  • Stopping Site Watch: when the merchant stops watching a website, or the website is removed from the merchant's official websites, everything Site Watch kept for it is deleted 30 days later (while a website is only waiting to be verified again, Site Watch is paused and keeps its data); "Stop and delete" in the app deletes it straight away. All of it is also deleted 30 days after Brand Guard stops for the account, and straight away when the workspace's Brand Guard data is erased (see below).
  • Brand Guard Site Watch Shopify activity: each record of a theme or product change is deleted 30 days after the change; the copy of a theme file is kept while it is part of the published theme of a watched website and otherwise deleted once no record from the last 30 days refers to it; the copies go with the website's Site Watch data when that is deleted (above)
  • Erasing Brand Guard data: when a workspace that was created automatically for a Shopify store is deleted after the store uninstalls the app (see above), or a workspace is permanently deleted, we also delete its captures and their files, its evidence images, its reference, its reference photos and photo-search records, its AI opinions, its weekly-summary unsubscribes, its Site Watch data and its misspelling-sweep names straight away, and remove the capture-based signals from its lookalike websites. A web account keeps them for the periods above, also after the account is closed, until you email us to have them erased.

6. Data Sharing and Disclosure

We use these service providers, under contract, to run the Service:

  • Amazon Web Services (USA): hosting, database and file storage, and the isolated automated browser (AWS Lambda, us-east-1) that takes Brand Guard captures, and the hourly function (AWS Lambda, us-east-1) that reads Certificate Transparency logs for Brand Guard, and the isolated function (AWS Lambda, us-east-1) that downloads and fingerprints Brand Guard reference photos, and the isolated function (AWS Lambda, us-east-1) that composes Brand Guard evidence images from our captures, and the isolated automated browser (AWS Lambda, us-east-1) that opens the pages of a merchant's own website for Brand Guard Site Watch
  • Brevo (Sendinblue SAS, France): sending our emails and receiving mail sent to our contact and support addresses; it reports delivery, opens and clicks
  • Stripe: payments for plans bought on our website
  • Google (Sign in with Google, USA): if you choose to sign up or log in with Google, Google confirms who you are and sends us your name, email address and Google account id, and Google learns that you use the Service. We never receive your Google password.
  • Shopify: order and store data when you use our Shopify app (for Brand Guard Site Watch, also the store's collections and best sellers, theme files and product changes, see section 2.3); Shopify bills plans bought through the Shopify App Store
  • Cloudflare Turnstile (Cloudflare, Inc., USA): blocks automated signups on the newsletter form and automated use of the Brand Guard website check (its link and message modes) and "Report a fake site" form; it processes your IP address and browser signals, without cookies or tracking
  • Google Cloud Vision (Google Cloud, USA): the image search behind Brand Guard's photo search; it receives a merchant's reference photos (or a link from which to fetch them) and returns the addresses of web pages that show them, under Google Cloud's standard terms (we have no separately negotiated agreement). Google states that it does not use the content sent to the Vision API to train or improve its Vision features, and that images sent in online requests are processed in memory and not persisted to disk.
  • Google (Gemini API, USA): the AI model behind Brand Guard's AI opinions, explanations of messages pasted into the website check and Site Watch notes and checklist checks on a merchant's own website; it receives only the data listed in section 2.3, under Google's Gemini API terms (we accept Google's standard API terms; there is no separately negotiated agreement). We use Google's paid Gemini API service. Google states that, on that service, it does not use what we send to improve its products and keeps prompts and responses only for a limited period to detect abuse.
  • Sentry (USA): error monitoring. Our setup does not send Sentry IP addresses, cookies or request bodies; we remove the data of background jobs and the query part of web addresses in your browser before a report is sent, and identify accounts only by internal ids. An error report can still contain a piece of the data being processed when the error happened.
  • Google Analytics 4 and Google Ads: site measurement, only if you accept cookies
  • Tatum: access to the Solana blockchain and IPFS, for products with anchoring turned on
  • Solana and IPFS: public networks where anchored batch records are published (see section 5.1)

Public services we query, without a contract: for Brand Guard we send domain names (a merchant's websites, lookalike websites, and the look-alike names the misspelling sweep generates) and lookalike websites' server addresses to public lookup services. They receive the name or address we look up and our server's address, nothing about you or the merchant's account. They act under their own policies, not under a contract with us.

  • Cloudflare (1.1.1.1) and Google Public DNS: DNS-over-HTTPS lookups of the domain names above
  • RDAP servers of domain registries, registrars and regional internet registries (found through IANA's public directory): the domain name or server IP address of a merchant's website or a lookalike website we look up, or a look-alike name the misspelling sweep found registered
  • Team Cymru: network (AS number) lookups of a lookalike website's server IP address, through the DNS resolvers above
  • Certificate Transparency logs (run by certificate authorities and other operators, such as Google, Let's Encrypt, Geomys, IPng Networks and GoDaddy, as listed in Google Chrome's public log list): we download their public entries; we send them nothing
  • crt.sh (Sectigo): the website name of a lookalike website, to find when its first certificate was issued

We may also share information with:

  • Verification Requesters: When someone verifies a product, they see the product's public verification status and the current owner's abbreviated name and masked email address
  • The public, for Brand Guard: a brand's public page shows its official websites and their change history (see section 2.3)
  • Recipients of removal reports: merchants send their removal reports themselves to registrars, hosting companies, Cloudflare, Google, Microsoft, Netcraft, Shopify or ICANN Compliance, and may share the case's evidence page with them; we do not send reports
  • Legal Requirements: When required by law, subpoena, or to protect our rights
  • Business Transfers: In connection with a merger, acquisition, or sale of assets

We do not sell your personal data to third parties.

7. Your Rights (GDPR)

Under GDPR, you have the right to:

  • Access: Request a copy of your personal data
  • Rectification: Correct inaccurate or incomplete data
  • Erasure: Request deletion of your data ("right to be forgotten"), subject to the ledger-integrity exception described in section 5.3
  • Restriction: Request limitation of processing
  • Portability: Receive your data in a structured, machine-readable format
  • Object: Object to processing based on legitimate interests
  • Withdraw Consent: Withdraw consent at any time for consent-based processing

These rights also apply if you are not our customer but appear in a Brand Guard capture, AI opinion, photo-search result, Site Watch capture of a merchant's own website or lookup (section 2.3).

If you bought from a store that uses our Shopify app, that store is the controller of your order data (section 2.1). When you ask the store for a copy of your data and Shopify forwards the request to us, we compile by hand what we hold about you across our apps and send it to the store, within 30 days, so that the store can answer you. Deletion requests forwarded by Shopify are handled automatically (section 5.4).

To exercise these rights, contact us at contact@verifined.com.

8. Contact Us

For privacy-related questions or to exercise your rights, contact us at:

You also have the right to lodge a complaint with a supervisory authority if you believe your data protection rights have been violated.