Platform Operations · 2026-04-13
Amazon Account Association: Every Detection Signal Ranked by Severity
A complete ranking of every signal Amazon uses to detect linked accounts — from CRITICAL triggers like shared addresses and identical payment methods that cause instant action, to LOW signals like timezone overlap that only matter when combined with stronger indicators.
How Association Detection Actually Works
Amazon does not rely on any single signal to determine that two seller accounts are linked. It uses a weighted scoring system where each signal contributes to an overall association confidence score. When that score crosses a threshold, the system takes action — ranging from enhanced monitoring to immediate suspension of all linked accounts.
Understanding which signals carry the most weight lets you make informed decisions about your business infrastructure. This is not about evading detection. It is about understanding which legitimate business choices create unintended association signals and which choices produce clean, independent profiles.
The signals below are ranked based on observed enforcement patterns across thousands of suspension cases documented between 2024 and 2026. Each signal is categorized by the severity of enforcement action it triggers.
CRITICAL Signals — Instant or Near-Instant Action
These signals individually carry enough weight to trigger immediate investigation or automated enforcement. Any single CRITICAL signal can result in account suspension without additional corroborating evidence.
Same Physical Address
Severity: CRITICAL | Detection: Automated | Action: Immediate investigation
When two seller accounts share the same business address, the association is flagged instantly. Amazon maintains a comprehensive address database that normalizes formatting, resolves abbreviations, and matches addresses even when written differently. "Suite 100" and "Ste 100" and "#100" all resolve to the same location.
This is the most common CRITICAL signal for legitimate sellers who use shared address services. Virtual offices, coworking spaces, and registered agent addresses create this signal by definition. The address does not need to be identical character-by-character — Amazon uses fuzzy matching that catches variations.
The severity increases with the number of seller accounts at the same address. Two accounts at one address triggers investigation. Five or more triggers automated enforcement with minimal human review.
Same IP Address or ASN Block
Severity: CRITICAL | Detection: Automated | Action: Enhanced monitoring, escalation with corroboration
When two seller accounts consistently log in from the same IP address, the association is flagged. But Amazon goes beyond individual IPs. It analyzes the ASN (Autonomous System Number) — the network block that the IP belongs to.
If two accounts share the same small ASN block, the system flags them even if they never use the exact same IP. This is particularly relevant for datacenter environments where all IPs in a facility share the same ASN. Two sellers using different VPS instances at the same provider are on the same ASN and therefore associated at the network level.
Commercial ISPs at different physical addresses produce different ASN associations. This is why separate physical internet connections at separate locations provide genuine network isolation — they are on different ASNs controlled by different infrastructure.
Same Payment Method
Severity: CRITICAL | Detection: Automated | Action: Immediate suspension of all linked accounts
Sharing a bank account, credit card, or any payment instrument between seller accounts is the strongest single association signal. The action is immediate and automated. Both accounts are suspended, funds are frozen, and the appeal process requires proving that the accounts are operated by genuinely separate businesses.
This includes not just the primary disbursement account but also credit cards used for advertising, FBA fees, or subscription payments. Even a one-time use of the same card across two accounts creates a permanent association record.
HIGH Signals — Triggers Manual Review and Escalation
HIGH signals do not cause instant automated action on their own, but they significantly increase scrutiny. When combined with other HIGH or CRITICAL signals, they can trigger enforcement.
Same Device Fingerprint
Severity: HIGH | Detection: Automated | Action: Manual review queue
Amazon collects detailed device fingerprints through Seller Central browser sessions and the Seller App. This includes hardware identifiers, browser configuration, canvas fingerprint, WebGL renderer, screen properties, and dozens of other attributes.
When two accounts produce the same or highly similar device fingerprint, the association is flagged for manual review. The fingerprint does not need to be identical — the system uses similarity scoring that accounts for minor variations.
Virtual machines and anti-detect browsers can alter some fingerprint components but often introduce detectable inconsistencies. A fingerprint that claims to be from a specific hardware model but has virtual GPU identifiers is more suspicious than a simple fingerprint match.
Same WiFi Network or Local Network
Severity: HIGH | Detection: Automated | Action: Enhanced monitoring
Amazon can detect when two accounts access Seller Central from the same local network, even if they use different devices. This is detected through network-level signals that go beyond IP addresses — including network latency patterns, MTU values, and other TCP/IP stack characteristics.
This signal is particularly relevant for sellers who operate multiple accounts from the same physical location, even on different devices with different internet connections that route through the same local network equipment.
Same Email Domain
Severity: HIGH | Detection: Automated | Action: Flagged for correlation
Using email addresses from the same custom domain (e.g., seller1@mybusiness.com and seller2@mybusiness.com) creates an immediate association flag. Even different domains hosted on the same server or registered with the same registrant information can be correlated.
Free email providers (Gmail, Outlook) do not trigger this signal between unrelated accounts, but using the same free email provider with similar naming patterns (johnsmith.store1@gmail.com and johnsmith.store2@gmail.com) can still be flagged through pattern analysis.
MEDIUM Signals — Contribute to Compound Score
MEDIUM signals individually have low enforcement weight. They become significant when multiple MEDIUM signals align or when they corroborate a HIGH or CRITICAL signal.
Similar Listing Style and Templates
Severity: MEDIUM | Detection: Algorithmic | Action: Score contribution
Amazon analyzes listing content for stylistic similarity. This includes writing patterns, HTML template structure, image composition style, A+ Content layouts, and keyword usage patterns. Two accounts with listings that share unusual stylistic elements are associated through content analysis.
This is not about selling similar products. It is about the way listings are constructed — the same distinctive bullet point format, the same unusual image cropping, the same branded template that was used across accounts.
Same Supplier or Fulfillment Source
Severity: MEDIUM | Detection: Algorithmic | Action: Score contribution
When two accounts source from the same supplier and this is detectable through invoice audits, brand registry applications, or product origin documentation, the association is recorded. Amazon can also detect shared fulfillment origins when FBA shipments from two accounts originate from the same preparation center or freight forwarder.
Overlapping Inventory or Product Catalog
Severity: MEDIUM | Detection: Automated | Action: Score contribution
Two accounts selling identical or nearly identical product selections — especially niche products that are not widely sold — creates an association signal. The weight increases when the overlap is in specific ASINs that have very few sellers, not broad categories with thousands of sellers.
Same Business Entity Type and Formation Pattern
Severity: MEDIUM | Detection: Algorithmic | Action: Score contribution
Multiple seller accounts registered to LLCs formed in the same state, at the same time, through the same registered agent, with similar naming conventions creates a pattern-based association. Each individual factor is MEDIUM, but the combination can escalate to HIGH.
Similar Pricing Strategies
Severity: MEDIUM | Detection: Algorithmic | Action: Score contribution
Accounts that consistently price products within the same margin bands, adjust prices at the same time, or follow identical repricing patterns are flagged for potential coordination. This is particularly weighted when the pricing behavior is unusual or non-competitive.
LOW Signals — Background Context Only
LOW signals do not trigger any enforcement action and contribute minimally to the compound score. They serve as contextual background that can tip a borderline case in either direction.
Same Timezone
Severity: LOW | Detection: Passive | Action: Contextual only
Accounts operated from the same timezone have a marginal association weight. This matters only when combined with multiple other signals — it can be the factor that pushes a compound score over the threshold, but it never contributes meaningfully on its own.
Similar Photography Style
Severity: LOW | Detection: Algorithmic | Action: Contextual only
Image analysis can detect similar photography setups — same lighting, same background, same camera angle patterns. This is a very weak signal because many sellers use the same photography services or follow the same Amazon image guidelines.
Similar Customer Communication Patterns
Severity: LOW | Detection: Algorithmic | Action: Contextual only
Templates used in buyer-seller messages, response timing patterns, and communication style are analyzed but carry minimal weight. Many sellers use the same customer service templates from popular course materials or software tools.
Same Advertising Keyword Targets
Severity: LOW | Detection: Automated | Action: Contextual only
Accounts targeting the same advertising keywords is expected for sellers in the same product category and carries very low association weight. It only becomes relevant when combined with other signals and when the keyword selection is unusually specific.
How Signals Compound
The danger of association detection is not any single signal — it is how signals multiply each other. Amazon uses a compound scoring model where the combination of signals produces a score greater than their individual sum.
Example 1: Low compound risk
Same timezone (LOW) + Similar product category (LOW) = No action
These signals are too common and too weak to indicate coordination
Example 2: Moderate compound risk
Same ASN block (CRITICAL) + Similar listing template (MEDIUM) = Manual review
The CRITICAL signal triggers review; the MEDIUM signal provides corroboration
Example 3: High compound risk
Same address (CRITICAL) + Same device fingerprint (HIGH) + Same supplier (MEDIUM) = Suspension
Multiple layers of association evidence create overwhelming confidence
Example 4: Maximum compound risk
Same address (CRITICAL) + Same payment method (CRITICAL) + Same IP (CRITICAL) = Immediate suspension of all accounts, possible permanent ban
Multiple CRITICAL signals indicate deliberate coordination
The compounding model means that eliminating even one CRITICAL signal can dramatically reduce your overall risk. A seller with a unique physical address, unique payment methods, and unique network but some MEDIUM signal overlap is at far lower risk than a seller with a single shared address.
What This Means for Multi-Account Sellers
Sellers who operate multiple accounts for legitimate business reasons — separate brands, separate product lines, separate geographic markets — must ensure that each account has genuinely independent infrastructure:
Per-account requirements for minimal association risk:
Unique physical address with a real lease document
Unique bank account at a different financial institution
Unique credit/debit card for all platform charges
Unique internet connection (different ISP, different ASN)
Unique physical device for platform access
Unique email address on an unrelated domain
Unique business entity with independent formation history
When every data point for each account points to a separate, real business location, there is nothing for the association system to detect. Each account appears as what it actually is — an independent business operating from its own infrastructure.
For a deeper technical explanation of how Amazon builds its detection model, see How Amazon Detects Linked Accounts: Fingerprint Model. For the foundational concept behind account linking, read What Is Amazon Account Association: Multi-Account Detection.
Building Association-Free Infrastructure
The most effective approach to association risk is not to manage signals individually but to build infrastructure where association signals cannot exist. This means each seller account operates from a genuinely independent business environment:
A dedicated physical address where your business is one of very few tenants, not hundreds
A commercial ISP connection at that address that produces a unique, location-appropriate ASN
Physical hardware dedicated to that account and stored at that location
Financial accounts opened for that specific business entity
When your infrastructure is real and independent, association detection works in your favor. Every scan confirms that each account is a separate, legitimate business — because it is.
The alternative — trying to make shared infrastructure appear independent through technical workarounds — is a fundamentally losing strategy. Detection systems improve continuously. Real infrastructure does not need to keep up with detection because there is nothing to detect.