An audit is not a report. A report says what happened; an audit says what to change. The difference is that each step of an audit ends with a finding: a specific problem, an estimate of what it costs, and the fix. This process is seven steps, ordered so the ones that find the most money come first.
Use 60 to 90 days of data. Pull event days (Prime Day, Black Friday week) out of the main analysis and look at them separately, because their numbers distort everything else.
Step 1: wasted search terms
Report: search term report, all campaigns.
Sort by spend. Mark every term with meaningful clicks and no orders. Then sort those into two piles: terms that do not fit the product at all, and terms that fit but have not converted yet.
Finding: total spend on terms that do not fit. That is money that can be removed with negative keywords without touching sales. In most accounts that have not done this work recently, it is the largest single finding.
Also note: converting terms that are not yet exact match keywords anywhere. These are harvesting opportunities, terms the account found and never moved into a place where their bids can be controlled.
Step 2: dead hours
Report: hourly performance data.
Compute conversion rate and ACoS by hour of day and day of week. Mark the hours where conversion is far below the daily average.
Finding: total spend in those hours, and what it produced. This is the second-largest finding in most accounts and the easiest to fix, because dayparting removes it without changing a single keyword. The heatmap guide shows how to build the view.
Step 3: placements
Report: placement report by campaign.
For each campaign, compare spend, conversion rate and ACoS across top of search, rest of search and product pages.
Finding: campaigns where most spend goes to a placement that converts poorly, and campaigns where top of search converts far better but the modifier is not set to favor it. Both are fixed with placement modifiers and base bid changes.
Step 4: structure
Report: bulk file download of the whole account.
Look for: the same keyword in several campaigns or ad groups, competing with itself. Products in shared campaigns where one product's budget is starving another's. Automatic and broad campaigns with no negatives, so they keep buying terms the exact campaigns already cover. Campaign names that do not say what is inside them.
Finding: a list of structural problems and a target structure. This step finds less immediate money than steps 1 to 3, but it determines whether every later fix holds. The structure guide describes the target shape.
Step 5: bids
Report: targeting report.
For each keyword with enough clicks to trust, compare ACoS to your target. Mark keywords well above target, which are candidates for bid decreases, and keywords well below target with low impression share, which are candidates for increases because they are profitable and under-delivered.
Finding: two lists, and an estimate of the spend saved from decreases and the sales gained from increases. Note also whether the campaign bidding strategy (dynamic up and down, down only, fixed) suits each campaign's purpose. The bid strategy guide covers the options.
Step 6: budgets
Report: campaign performance by day, plus budget status.
Mark campaigns that regularly run out of budget, and note when in the day they run out. A converting campaign that runs out by noon is losing the evening's sales. Mark campaigns whose budgets are far above what they spend, which usually means they are not competitive, not that the budget is generous.
Also check for event budgets still in place after the event. Elevated post-event budgets are a common and easy finding.
Finding: campaigns underfunded during their best hours, and campaigns overfunded for no reason.
Step 7: listings
Report: business reports by product (sessions, conversion rate), alongside ad conversion rate.
For each advertised product, check conversion rate against the account's norm. Products with low conversion rate are capping what every ad change above can achieve. Look at the main image, price relative to the search results, review count and rating, and title.
Finding: products where the listing, not the ads, is the constraint, with the specific problem named. This is the step most audits skip and the one that most often explains why an account's ACoS will not come down.
Writing it up
One page. Each finding as a line: what the problem is, what it costs over the audit period, the fix, and how confident you are in the estimate. Ordered by cost. The top three findings are the first month's work. The rest wait.
Then decide which fixes are one-time (restructuring, the first negative sweep) and which recur (negatives every two weeks, schedule reviews each quarter, budget restores after every event). The recurring ones belong in rules or a calendar, or the next audit will find them again. The pre-automation checklist is the shorter version of this process for an account about to adopt rules.
Frequently asked questions
How often should I audit my Amazon advertising account?
A full audit once or twice a year, and before any major change such as adopting automation, hiring an agency, or going into Q4. Lighter monthly reviews of search terms and placements cover the ongoing work between audits.
How long does an Amazon PPC audit take?
For a small account with a handful of products, an afternoon. For an account with dozens of campaigns across several products, a few days. Most of the time goes into the search term report and the structural review.
What date range should an audit use?
At least 60 days, and 90 if the account has low-volume keywords. Exclude major event days like Prime Day or Black Friday from the main analysis, because their conversion rates and CPCs distort everything else, and review them separately.
Off Hours turns the recurring findings into rules: dayparting for step 2, bid adjustments for step 5, and budget and event rules for step 6, all logged. Start a free 14-day trial.