← Back to blog
Field notes

Amazon Voice of the Customer and NCX Rate Explained

January 30, 2026·EcomSanity Team·6 min read

Quick answer: Amazon's Negative Customer Experience rate, or NCX rate, is generally orders with a reported product or listing problem divided by total orders in the measured period. Voice of the Customer compares that against similar offers and assigns CX Health from Excellent to Very Poor. A high rate can lead to warnings or listing deactivation. Inspect the underlying reasons, sample size, and physical product before choosing a fix.

The product had sold twice. One buyer kept it. The other returned it and selected "not as described." The next morning, the ASIN showed a 50% negative customer experience rate and a frightening red status. For a used-book seller, the number felt absurd, one return had transformed a normal listing into what looked like a product crisis. Mathematically, the dashboard wasn't wrong. Operationally, the sample was too small to interpret like a mature ASIN.

What counts as a negative customer experience

Amazon uses several customer signals to understand whether a product or listing caused a poor experience, including returns and return reasons, refunds, customer service contacts, and complaints about condition, quality, compatibility, or damage. Voice of the Customer is not merely a review monitor. A product can have a high star rating and still have poor CX Health because recent orders generated returns or complaints.

The basic NCX formula

NCX rate equals orders with a reported negative customer experience divided by total orders. If one of two orders creates an NCX event, that's 50%. If ten of 1,000 orders create NCX events, that's 1%. The first percentage looks catastrophic, but the evidence is one event. The second looks small but represents ten actual customers and may be meaningful in a category with very low expected problems. Always read rate and count together.

CX Health is comparative

Amazon compares the offer's customer experience with similar offers, with status labels from Excellent through Good, Fair, Poor, and Very Poor. A 4% NCX rate might be normal in one product type and severe in another, and the exact comparison logic isn't fully public. There's no reliable universal "safe" NCX rate to aim for.

Why low-volume products are volatile

Low denominators create violent percentages. For one-off inventory, books, collectibles, and newly launched products, a single return can produce 20%, 50%, or even 100% depending on the sample size. That doesn't mean the signal should be ignored, it means the seller should distinguish a true product or listing problem from a one-off customer mismatch, a condition-specific issue, or plain statistical volatility. The right action may be a listing correction rather than a product recall.

The five root-cause families

Product quality: breakage, defect, leakage, a weak component, or manufacturing inconsistency. Listing accuracy: wrong dimensions, misleading color, unclear quantity, or unstated compatibility requirements. Fulfillment and packaging: transit damage, a product that moves inside the package, or labels that damage the retail box. Variation structure: reviews or images that create expectations for a different child, or ambiguous size names. Customer fit: a product that's technically correct but unsuitable for common use, or a capacity that's hard to visualize.

The jars that were "too small"

A seller offered 105 ml glass jars. The title stated 105 ml, but the main image showed a close crop with no hand, ruler, or familiar object for scale. Returns said "smaller than expected." The capacity was correct. The customer expectation was not. Adding a dimension image, a hand-held scale photo, and a comparison with the larger version left the product unchanged while the buying decision became more informed. This is why Voice of the Customer should influence content, not only quality control.

A 45-minute Voice of the Customer investigation

Identify material ASINs by revenue, NCX count, NCX rate, and recent trend. Read every available reason and group comments into themes, don't treat "defective" as a complete diagnosis. Compare return reasons with listing claims: if customers say "does not fit," locate every compatibility claim. Inspect physical inventory across different production batches. Compare children and fulfillment paths to see whether the problem belongs to one size, color, or warehouse route. Then choose containment (pausing ads, removing a defective batch) and permanent action (supplier correction, packaging redesign, or content fix).

The NCX action matrix

PatternLikely first action
High rate, low count, unclear reasonReview order detail and listing, monitor next events
High count, repeated identical reasonImmediate root-cause action
Damage concentrated in one fulfillment routePackaging and fulfillment investigation
One child much worse than siblingsChild listing, batch, or variation investigation
Returns say size mismatchDimension and expectation content
Good reviews but high returnsCheck fit, sizing, condition, and silent dissatisfaction

Frequently returned item badge

Amazon can display customer-facing warnings on products with return patterns that stand out. These badges can hurt conversion even before a listing is deactivated. Treat a badge as both a customer-trust problem and a profit problem, don't try to hide the return reason with vague copy, make the buying decision more accurate instead.

When a listing is deactivated

A useful response includes the root cause, immediate corrective action, evidence of correction, and preventive action. Avoid generic statements like "we checked everything and it's fine." If the issue is dimensions, show the corrected attribute and image. If the issue is a defective batch, identify the batch and removal action.

Use profit to prioritize fixes

An ASIN with a 6% return rate and $80 contribution per order may still be profitable, but the customer-experience risk can justify urgent work. An ASIN with a 3% return rate and $2 contribution may already be economically broken, covered further in how to calculate Amazon profit per SKU. Estimate expected return cost per order as return rate times average total loss per return, including refund leakage, the returns processing fee where applicable, and unsellable inventory.

Prevention routine

Weekly: review Poor and Very Poor products, read new return reasons, and compare low-volume percentages against raw counts. Monthly: rank ASINs by expected return cost, compare supplier batches, and review dimension and compatibility content.


EcomSanity cross-references return counts with units sold in the same window and provides reason and disposition context, making it easier to see whether a red Voice of the Customer signal is isolated or part of a growing commercial pattern. If returns rise and conversion falls before sales do, customer trust may already be weakening, and ads continuing to scale a Very Poor product is a signal to stop buying more exposure until the cause is understood.

Frequently asked questions

What is a good NCX rate on Amazon?

Amazon compares products against similar offers, so there's no reliable universal threshold. Use the CX Health status, event count, category context, and trend together rather than chasing a specific percentage.

Can one return make CX Health show Very Poor?

Yes, on very low sales volume, one return can create a very high rate mathematically. Investigate the actual event and the sample size together before treating the label as a verdict on the product.

Does Voice of the Customer affect Account Health Rating?

Product CX Health and Account Health Rating are related operational concerns but are separate systems. A product can be deactivated on Voice of the Customer grounds while overall Account Health remains healthy.

Cleared for takeoff

See this on your own catalog.

Storage-fee radar, Buy Box tracking, and return-rate analytics, connected to your Amazon account in under a minute.

Get started