Amazon's account linking detection is more sophisticated than most sellers realize. While most guides focus on IP addresses, Amazon's system uses a multi-signal approach where browser fingerprinting plays a significant role.
What Amazon Actually Tracks
Amazon uses browser fingerprinting as part of their fraud detection system. When you log into Seller Central, your browser sends dozens of signals:
- Canvas fingerprint — unique rendering signature of your GPU + browser
- WebGL renderer — reveals your graphics card model
- AudioContext fingerprint — unique audio processing signature
- Installed fonts — OS + application-specific font list
- Screen resolution and color depth
- Timezone and language settings
- Browser plugin list
The Correlation Problem
The issue isn't any single signal — it's how they're combined.
If Account A and Account B both access Seller Central with:
- Different IPs ✓
- Different cookies ✓
- Same Canvas fingerprint ✗
- Same font list ✗
- Same AudioContext signature ✗
Amazon's system flags them as likely operated by the same person. The fingerprint signals match even when IP and cookies don't.
How Detection Actually Works
Amazon doesn't necessarily ban accounts immediately. Instead, they build a correlation graph:
- Browser fingerprint A → associated with Account 1
- Browser fingerprint A → also associated with Account 2
- Account 1 and Account 2 are linked
Once linked, a policy violation on one account can cascade to all linked accounts.
Testing Your Own Fingerprint Overlap
You can verify this yourself using free tools:
BrowserLeaks.com shows your current fingerprint signals. If you open this tool in two different browser windows and get identical Canvas hash + font hash, those windows would appear linked to any fingerprinting system.
CreepJS (GitHub) provides a more comprehensive analysis including the trust score that fingerprinting systems assign.
Effective Isolation Approaches
What Doesn't Work
- Different Chrome profiles — share the same Canvas/WebGL fingerprint
- Incognito mode — same fingerprint as normal mode
- VPN alone — changes IP but not fingerprint
- Browser extensions that claim to "block" fingerprinting — usually detectable
What Does Work
True isolation requires changing the fingerprint at the browser engine level, not just blocking access to fingerprinting APIs (which itself is a fingerprint signal).
The browser needs to:
- Return different Canvas pixel data per profile
- Report different WebGL renderer strings
- Generate different AudioContext signatures
- Present a different (but plausible) font list
This has to happen consistently across all APIs, not just some of them — inconsistency is itself a fingerprint.
Practical Implications
If you're running multiple Amazon seller accounts, the key question is: does each account have a genuinely different browser fingerprint?
You can test this with BrowserLeaks in each browser/profile you use. The Canvas fingerprint hash should be different for each account's browser environment.
I built FireKey specifically to address this — it assigns a completely isolated fingerprint identity to each browser profile at the engine level. Currently in free open beta.
Top comments (0)