A clear plan for your next ad dollar.

A confident recommendation starts with comparable data. Confirm the marketplace, currency, date range and attribution window before drawing conclusions.

• Use matching purchase and sales windows. Separate duplicate rows, totals and overlapping exports.
• Check that each report contains the detail the decision needs. A Search Term Report does not contain campaign placement totals.
• Review stock, listing eligibility and recent promotions separately. A delivery problem needs a different response from weak conversion.

Use amazon-report-preflight to validate these Amazon reports before analysis. Confirm marketplace, currency, date range, attribution windows, report grain and duplicate rows. Preserve the originals. Separate usable data from blocked inputs and list the missing context. Do not infer exact organic sales from differences between reports.

Record: the usable files, settled date ranges and anything still blocking a decision.

An account average can hide products with very different economics. Review comparable products together and make the cost assumptions explicit.

• Confirm selling price, fees, landed cost and other relevant costs. Recheck after a price change or promotion.
• Calculate profit before ads and break-even ACoS. Choose a target below break-even that reflects the contribution you want to retain.
• For search-term decisions, use contribution per attributed purchase. Adjust a one-unit estimate for multiple units or other products purchased.

A simple example: $12 contribution before ads on a $40 sale gives a 30% break-even ACoS. Keeping $4 after ads leaves $8 for advertising, or a 20% target ACoS. These are teaching numbers, not recommended targets.

Record: the cost assumptions, break-even ACoS and target for each product group. Missing costs remain an open question.

A higher ACoS can reflect more expensive clicks, fewer purchases per click, or less revenue per purchase. Each points to a different investigation.

• Compare the same product group over comparable, settled periods. Note promotions, price changes and shifts in traffic mix.
• Separate the arithmetic contribution of CPC, purchase rate and sales per purchase. Then check the underlying terms, placements and listing context.
• If Ads and Business Report sales differ, check their definitions and dates before treating the gap as a business result.

Record: the main numerical driver, your working explanation and the evidence still needed to test it.

Start with spend that has produced few or no purchases. Treat it as a review queue. The right response depends on relevance, settled evidence and the product’s margin.

• Keep campaign, ad group and product context with every term. Separate search queries from product targets.
• Check relevance first. For relevant terms, compare clicks and purchases with the conversion rate needed to cover their click cost.
• Review exclusion scope and match type before adding a negative. Protect terms that need a strategic review; avoid a universal click cutoff.

A statistical flag is a reason to review. Repeated checks, multiple tests and changing conversion conditions affect how you interpret it.

Record: the term, product context, evidence, proposed action and reason. Keep unproven “savings” out of the result.

A first sale is a useful signal. An exact-match test still needs a relevant query, plausible economics and enough evidence to justify its budget.

• Review converting terms in their original campaign and ad group. Use a comparable campaign baseline and product contribution.
• Check the fresh bulk export for an existing exact keyword, its state and the products it advertises. Missing from one report does not prove missing from the account.
• Plan the destination, starting bid and test budget. Confirm the destination is eligible and delivering before adding negatives that route traffic away from discovery.

Record: the candidate, existing coverage, intended destination, test limits and what would justify the next change.

A budget increase should have a reason beyond yesterday’s sales. Check click value, campaign purpose and stock before funding more traffic.

• Use actual placement data. Compare attributed sales per click with the product’s target ACoS and note sparse samples.
• Review the base bid, placement adjustments and dynamic bidding strategy together. A placement percentage is only one part of the setting.
• Check budget constraints, inventory cover and listing readiness. Prioritize a bounded test where evidence and capacity support it; document strategic exceptions.

Record: the current setting, proposed setting, economic reason, spending limit and conditions for reviewing the test.

Turn the review into a short, controlled change list. Each action should have a reason, a limit and a way to check what happened.

• Compare proposed changes with a fresh, complete configuration export. Check IDs, parent entities, duplicates, negative conflicts, names and bid or budget limits.
• Review the checked file before uploading through Amazon. Then inspect processing results and a fresh read-back; a prepared file does not prove a change was applied.
• Confirm delivery before routing traffic. Set a review date that allows new performance to settle, and compare with the expectations in your change log.

Use amazon-ppc-bulk-qa to validate these proposed Sponsored Products changes against this fresh, complete bulk export. Preserve names and IDs. Check parent relationships, duplicate exacts, negative conflicts and bid/budget limits. Default to no more than 25 new exacts per campaign unless I specify another cap. Prepare the checked upload and before/after list only if the batch passes. Hold routing negatives until destination read-back and delivery are confirmed. Do not make live account changes.

Record: the approved changes, held items, upload outcome, read-back, owner and next review date. Next week, begin here.