Give a negative keyword a second review when the reason for excluding it may have changed. Do not remove a clear product mismatch merely because it has been inactive for a long time. A second chance is a controlled test of an uncertain performance decision.

Separate permanent mismatch from uncertain economics

Original reason	Reopen?	Evidence needed
Product does not fit the requested size or use	Usually no	A genuine product or catalog change
Query used immature zero-order data	Review	Re-export the original period after attribution matures
High spend after a stock or delivery problem	Consider	Current availability and delivery restored
Margin was too low at the old price	Consider	Updated contribution and an affordable CPC
Query routed to another campaign	Check routing first	Confirm where the intended destination now serves

Amazon supports negative phrase and negative exact matching; match behaviour includes the applicable close-variation rules. Do not assume a negative applies only to identical characters or that negative broad is an available Amazon keyword type. Amazon match-type guidance.

Recover the decision context

Find the original term, match type, campaign or ad group, date, mature clicks, spend, orders and reason. If that history is unavailable, mark the original rationale unknown rather than claiming the exclusion was wrong. Check other campaign and ad-group negatives before testing; removing one entry may not remove every blocking condition.

Re-export the original data using the same attribution definition. New sales credited to earlier clicks can change the original assessment. The history check does not prove the term will perform well now. Amazon conversion metrics.

Build a bounded retest

Choose a relevant term, an eligible product and an affordable starting bid. Set an explicit maximum test spend and a review date that accounts for conversion delay. Keep the retest separate enough to identify its traffic, but verify that retained negatives do not block it. Give one operator or tool ownership of the changes.

Hypothetical example: an order provides £12 contribution before ads. At an assumed 5% CVR, break-even expected CPC is £0.60. A £0.40 CPC would imply £8 advertising per expected order under that assumption. An approved £20 test limit bounds exposure; it does not establish that 50 clicks are statistically sufficient or guarantee an order.

Judge the retest on comparable data

After the period matures, inspect the actual search terms, spend, orders and contribution. A negative exact removal can expose close variations, so confirm what received traffic. If the test produced no useful evidence before its loss limit, record it as inconclusive or uneconomic under the test conditions rather than continuing indefinitely.

Restore an exclusion when its specific reason remains valid. Continue only when relevance and economics justify the exposure. Do not interpret a higher sales count without considering additional spend or seasonality.

Keep a useful review record

Term	Original_reason	Changed_condition	Scope	Spend_limit	Review_date	Decision
example_query	immature_data	mature_export_reviewed	assign_scope	set_limit	set_date	pending

This is a manual record for a spreadsheet, not an Amazon upload template. Save the old setting so a reviewer can see exactly what was removed and why.

Should every negative expire automatically?

No. Product incompatibility and deliberate traffic routing need different treatment from uncertain performance. Review by reason and changed conditions. See negative keyword design for scope selection and bid calculation for economic limits.

Related Arctavia resources

Use these supporting pages to compare Amazon PPC operating models and implementation choices.

• Amazon Ad Rejections: Check Claims, Images and Destinations
• Amazon Ads ACoS Suddenly Increased: What to Check First
• Amazon Ads ML Bid Review: Guardrails Before Aggression
• Amazon PPC Automation: Data, Decisions and Operating Controls
• Amazon PPC Automation Migration Checklist
• Amazon PPC Automation vs Agency: Which Operating Model Fits Better?