An Amazon PPC audit you can act on
From report checks to a reviewable change list
Audit Amazon Sponsored Products with a practical checklist for data quality, bids, search terms, budgets, placements and upload verification.
An Amazon PPC audit should finish with three lists: changes worth reviewing, changes that need more evidence, and things to leave alone. A report that labels all high-ACOS spend as recoverable waste is not enough. Start by confirming what the data can support.
This checklist covers Sponsored Products. You can use it manually, with an analyst or while evaluating software. AutoPPC follows a rules-based review process, but the checklist is not an instruction to upload changes without checking your own account.
1. Establish the scope and the data
Record the marketplace, currency, report dates, ad format and business goal. Use matching time windows when comparing performance. A campaign report, targeting report and search-term report show different grains of the same activity; adding their spend together would count it more than once.
Use targeting data for keyword and product-target performance, search-term data for the actual queries, placement data for where traffic appeared, and current campaign state for what exists now. Amazon's reporting guide describes these report roles.
Check for missing reports, recent changes, partial periods and unsettled attribution. Today is useful for monitoring delivery, but it is not a settled measure of conversion performance. Flag insufficient evidence instead of converting it into a confident recommendation.
2. Separate ad efficiency from profitability
Calculate ACOS as spend divided by ad-attributed sales. With zero sales, report spend and zero orders directly rather than manufacturing a finite ACOS. Calculate TACOS only with matching total sales; do not relabel advertising revenue as total revenue.
Set a target from your business goal and usable unit economics. Break-even ACOS depends on contribution before advertising: product revenue less the relevant non-ad variable costs. A target below break-even leaves room for contribution; a launch may deliberately prioritize something else. State that choice.
Use product groups when materially different margins would make one account target misleading. Keep manual targets and margin-derived targets distinguishable so you can explain where each number came from.
3. Review bids with converting traffic
For a settled, comparable sample, revenue per click multiplied by target ACOS gives a reference target CPC. Compare that reference with observed CPC and the current bid. It is not a promise that Amazon will deliver clicks at that price.
Check sample size, prior changes, bid limits, stock and placement mix before acting. A keyword can be above target yet need no further cut if a recent change is still settling. Likewise, one inexpensive conversion is weak evidence for a large raise. See the bidding guide.
4. Investigate zero-order search terms
Sort zero-order terms by spend to find where a review could matter. Then inspect their literal queries, click counts, match types, product relevance and recent history. A high-priced product can require more evidence than a low-priced one.
Before adding a negative, check whether its scope could block converting queries. A negative phrase can reach more than the one search term you first noticed. Inspect related converting families and existing negatives before choosing the match type and scope.
Keep observed zero-order spend separate from projected avoidable spend. The first is historical data. The second assumes future traffic and conversion behavior and is not guaranteed savings.
5. Check harvesting and campaign structure
Look for converting search terms worth controlling more explicitly. Confirm the destination campaign and ad group exist or can be created through your supported workflow, in the same marketplace and with the right product.
Check for existing targets before adding duplicates. If a harvest includes an accompanying negative, preserve the order and eligibility of both parts. Creating the destination and blocking the source are separate actions with different consequences. See harvesting and negation.
6. Review budgets, placements and inventory together
A campaign spending its budget is not automatically underfunded. Check its target, conversion quality, budget history and business priorities. Decide whether additional spend is justified before calling a higher budget an opportunity.
A placement modifier changes an auction bid, not the price necessarily paid for a click. Examine realized placement performance and the bidding strategy. A top-of-search placement is not always your best-performing placement. Amazon explains placement and bidding controls.
Check stock and the reliability of your inventory inputs before buying more traffic. A recommendation held for missing evidence can be as useful as one that passes.
7. Produce a review packet and check delivery
For each proposed change, record the entity, marketplace, previous value, proposed value, evidence dates, reason, relevant safeguards and delivery method. Keep unresolved questions visible. See a fictional worked packet.
After approval, inspect the upload or write result. An exported CSV is not proof of application. Verify the expected account state, preserve the prior values and wait for attribution before judging performance. Document other changes that could explain the result.
What an AutoPPC audit includes
AutoPPC's free audit covers one eligible advertiser profile and one full engine run, using up to 95 days of available account history. It reveals a limited portion of findings in each section and includes capped read-and-explain copilot access. Closed-beta access is subject to availability.
The free audit does not include approval, bulk export, drafting or scheduled optimization. Paid plans unlock the ongoing process. Advertising delivery remains seller-approved manual upload, followed by a later read-only verification. Check current access and plan details before starting.
The value of the audit is a better next decision, including when the right decision is to wait. It is not a guaranteed percentage reduction in ACOS or a forecast that all flagged spend can be eliminated.