The Amazon PPC Audit Guide (2026)

An Amazon PPC audit is a structured check of whether your ad spend is making money, where it is leaking, and what it is missing. You compare each product's ACoS (advertising cost of sale) with its break-even point, watch the TACoS (total advertising cost of sale) trend, then work through search terms, targeting, bids, placements, campaign structure and budgets in that order. The last step, and the one most audits skip, is checking whether the real problem is the product page rather than the ads.

This guide walks through that process for Sponsored Products, with shorter notes on Sponsored Brands and Sponsored Display. Every step has a formula, a worked example with numbers, and the Ads console or Seller Central path you need.

Key takeaways

• Set your break-even and target ACoS before you open a single report. Without them, "good" and "bad" have no meaning.
• Judge the account on TACoS trend and organic sales, not ACoS alone. ACoS can hold steady while organic sales fall.
• Most recoverable money sits in three places: search terms with clicks and no sales, bids that ignore conversion rate, and placements that cost more than they return.
• Leave out the most recent days. Seller Sponsored Products sales are credited up to 7 days after a click, and Sponsored Brands and Sponsored Display up to 14 days.
• Fix product problems (price, reviews, stock, listing) before you cut bids. Ads amplify a product; they cannot repair it.

1. What an audit should answer

An audit is a set of questions; the changes follow from the answers. Write these four at the top of your notes:

1. Is advertising profitable, product by product? Not the account average. Each product.
2. Where is money leaking? Search terms, targets and placements that spend without returning enough sales.
3. Where is money missing? Converting search terms you are not bidding on directly, profitable campaigns that run out of budget, and placements where you are underbidding.
4. Is this actually an advertising problem? A drop in conversion rate, a lost Buy Box or a stock-out looks like "PPC stopped working" in the ads data.

Work through the sections in order. There is no point fine-tuning placements on a campaign whose search terms are half irrelevant, or raising budgets before removing waste.

2. Prepare: the data you need before you start

The reports to pull

You need data from two places: the Amazon Ads console, which knows about ad clicks, and Seller Central, which knows about all sales and traffic.

If you prefer one file, a bulk operations spreadsheet (Campaign Manager > Bulk operations) can include campaign structure plus search term data. It is the easiest file to filter for structure problems, and you can upload changes back through it.

Choose your date ranges

Different checks need different windows:

• Search terms and targets: 30 to 60 days. The console only keeps roughly the last two months of search term data, so save a copy every month.
• Account health trend: the last three full months, plus the same months last year if you have them, to allow for seasonality.
• Placements and bids: 30 days for busy campaigns, 60 for quieter ones.
• Weekly checks: the last 7 to 14 days, read with caution.

Allow for attribution lag

Ad sales are tied to the date of the click, not the date of the order. For sellers, Sponsored Products credits a sale to a click for up to 7 days. Sponsored Brands and Sponsored Display use 14 days. Business Reports count sales by order date.

That means the last few days of any ads report always look worse than they will end up. When your report period ends today, leave out at least the last 3 days for Sponsored Products decisions, and up to a week for Sponsored Brands and Display. It also means ad sales and Business Reports sales will not line up exactly over short periods. Over 30 days or more, the difference mostly evens out.

Know your numbers first: break-even and target ACoS

Break-even ACoS is the ACoS at which an ad sale makes zero profit. It equals your profit margin before advertising. Every check in this guide compares against it, so calculate it per product before you look at a campaign. Our guide to break-even ACoS covers it in depth.

Your target ACoS is lower than break-even, because you want ads to earn a profit. How much lower depends on the product's stage, which section 12 covers. In this example, a 30% target leaves about $3.68 of profit per ad sale (10.5% of $34.99).

Two timing notes for October 2026. Holiday peak fulfillment fees apply from 15 October 2026 to 14 January 2027, so the FBA fee is higher and break-even ACoS is a little lower during that window; see Amazon holiday peak fees. And if you run a coupon or deal, use the price after the discount.

Free toolBreak-even ACoS & Max BidWork out break-even ACoS, a target ACoS and the matching max bid for each product before you audit its campaigns.Open tool →

3. Account health in five numbers

Before you look at a single search term, get a one-page view of how advertising is affecting the business. Five numbers do most of the work.

For more on the first two, see ACoS vs TACoS.

Worked example: the trend that ACoS hides

Say a brand's last three months look like this:

Looked at through ACoS alone, nothing is wrong. It moved from 28.1% to 30.0%, well inside a typical target. But total sales did not grow. Ad spend rose by $1,500 a month while organic sales fell by $4,000. The business is paying more to stand still.

Likely causes: ads now winning searches the product used to win organically (such as your own brand name), slipping organic rank on key searches, a drop in conversion rate, or a planned launch inside the totals. The sections below narrow it down.

Free toolTACoS & ROAS CalculatorTrack TACoS and organic sales month by month for each product, so you can see whether ads are growing the business or just buying back existing sales.Open tool →

4. Search term audit: find the leaks

The search term report is where most audits find their quickest savings. It lists what shoppers actually typed (or the ASIN of the product page they were on) before clicking your ad. If you are new to it, read our search term report guide first.

You are looking for four things: waste, gaps, conflicts and patterns.

4.1 Set a click threshold

You cannot call a search term a failure after three clicks. Set a minimum number of clicks before you judge a term with zero orders. The usual starting point is based on your conversion rate.

The 1.5 multiplier gives each term a fair chance. Obviously irrelevant terms (wrong size, wrong product, wrong use) can be negated at any click count.

4.2 Measure wasted spend

Wasted spend is the money spent on search terms that passed the threshold and produced nothing. It is the clearest single number for how well negatives are being maintained.

Keep two other buckets separate:

• Unproven: zero orders but below the click threshold. In the same example this might be another $860. It is not waste yet; it is the cost of testing.
• Inefficient: terms with orders but an ACoS well above target. These need lower bids, not negatives, unless the term is irrelevant.

Some waste is normal. Discovery costs money. What you want to know is whether the same kind of waste returns every month, which means negatives are not being added, or whether new waste appears and is cleaned up quickly.

4.3 Check negatives for conflicts

Negatives are easy to add and easy to forget. Over time they can start blocking traffic you want. Pull your negatives (the bulk file shows them all) and check:

• Negatives that block your own targets. A negative phrase "foam" in an exact campaign that also bids on "memory foam pillow" stops that keyword serving. The console does not always warn you.
• Negative phrase on a word that also appears in winners. Before you keep a single-word negative phrase, search the converting terms for that word.
• Missing source negatives. When a term is harvested into exact, it should be added as a negative exact in the auto, broad or phrase campaign it came from. Otherwise your campaigns bid against each other and the data splits.
• Negative ASINs. In auto and product targeting campaigns, competitor ASINs that took many clicks without sales should be negative product targets.

Our guide to negative keywords covers how negative exact and negative phrase behave.

4.4 Look for harvesting gaps

The reverse of waste: converting search terms that only run through auto, broad or phrase campaigns, sharing one bid with hundreds of other searches. A simple rule many sellers use is to move a term into exact match when it has two or more orders at or below target ACoS. The full routine is in keyword harvesting.

In the audit, count how many search terms meet your harvesting rule but are not yet exact keywords anywhere in the account. If that list is long, harvesting has stopped happening, and your best searches are being bid on with a discovery-level bid.

The same applies to ASINs. A competitor's product page that converts well for you belongs in a product targeting campaign with its own bid.

4.5 Word-level (n-gram) analysis

Single rows hide patterns. Twenty search terms that each waste $6 do not stand out, but if they all contain the word "kids", that word cost you $120. Word-level analysis (often called n-gram analysis) splits every search term into single words (and sometimes two- or three-word phrases), then adds up clicks, spend and orders for each word.

Here is a worked example. Say you sell an adult cooling pillow for $39.99. A 60-day search term report, split into single words, shows:

How to read it:

• "kids" and "travel": you do not sell a kids pillow or a travel pillow. Together they cost $126.00 + $81.60 = $207.60 over 60 days for one sale. Both are strong negative phrase candidates.
• "cheap": one sale, but at 150% ACoS. Shoppers who type "cheap" are price-led. Check whether your price is competitive for them before negating; often a negative phrase is still the right call.
• "cooling" and "queen": both well under a 30% target. If any search terms containing them are not yet exact keywords, they are harvesting candidates. They are also words to make sure no negative accidentally blocks.

One caution: the rows overlap. "cooling pillow queen" counts under both "cooling" and "queen", so you cannot add the rows up to get a total.

4.6 Odd search terms to expect in 2026

Two recent changes can put unfamiliar rows in your report:

• Longer, conversational searches. Since 13 May 2026, Amazon's Rufus assistant has been folded into Alexa for Shopping, which answers shopper questions from the search bar. Expect more question-style phrases. Judge them by results like any other term. See Amazon Rufus and Alexa for Shopping for sellers.
• Sponsored Products prompts. AI-written prompt ads for Sponsored Products and Sponsored Brands have billed on CPC under your existing bids since 25 March 2026 in the US. They have their own Prompts report type. Include it in monthly audits.

Sponsored Products ads can also serve off Amazon. Where there is no shopper search, Amazon may show a search term it has inferred. Treat those rows with the same click-threshold logic.

Free toolSearch Term Report AnalyzerUpload a downloaded search term report to sort terms into winners, bleeders and negative phrase candidates, and see word-level totals without building pivot tables.Open tool →

5. Targeting and match-type mix

Once search terms are clean, step back and look at where the money goes by targeting type. This tells you whether the account is still mostly "discovering" or is spending on what it already knows works.

Pull the targeting report (or a campaign report, if your campaigns are split by targeting type) and group spend and sales by type.

Worked example: a mature product still paying to explore

Say one mature product's Sponsored Products campaigns show this over 60 days:

Exact match produces the best ACoS and a third of all ad sales ($9,000 of $28,125), yet it gets only 20% of the spend. Auto and broad together take 55% of spend at 37.5% to 40% ACoS. For a launch product, heavy discovery spend is normal. For a mature product, it usually means harvesting has fallen behind (section 4.4), or the exact campaign is held back by low bids or a small budget.

Next, check whether the exact campaign runs out of budget (section 9), whether exact bids sit below what their revenue per click supports (section 6), and whether the auto campaign's four targeting groups have separate bids, since close match usually behaves very differently from complements (see auto campaign targeting groups). There is no single correct split; what you want is a reason for the split you have.

Branded vs non-branded

If shoppers search your brand name, branded campaigns will usually show a very low ACoS. Mixed into other campaigns, that low ACoS hides weak generic performance. In the audit, separate spend and sales on searches containing your brand name from everything else. Then ask two questions: is the non-branded ACoS acceptable on its own, and is branded spend protecting sales from competitors or just paying for sales you would have made organically? The Search Query Performance report helps answer the second question; see our Search Query Performance guide.

6. Bids: are they based on the numbers?

Many accounts run on suggested bids, round numbers or bids set at launch and never changed. The audit question is simple: for each target with enough data, is the bid what its results can afford?

Sales per click: the number behind every bid

Every click is worth, on average, the ad sales it brings in. That is revenue per click, also called sales per click. It combines order value and conversion rate in one number.

Actual revenue per click beats list price as an input, because it picks up multi-unit orders, coupons and sales of other products in your brand.

Worked example: four keywords, four different answers

Same cooling pillow, $39.99, target ACoS 30%, 60 days of exact match data (most recent week excluded):

Now work out revenue per click and the max CPC at 30%:

Notice that "cooling pillow" has a bid ($1.40) above its max CPC ($1.32) but still hits target, because the average CPC paid ($1.30) is below the bid. Amazon's auction often charges less than your maximum. Bid on the numbers, then judge on actual CPC and ACoS.

Check the bidding strategy and stacking

Each Sponsored Products campaign has one bidding strategy: dynamic bids – down only, dynamic bids – up and down, or fixed bids. Under up and down, Amazon's dynamic bidding guide says it can raise bids by up to 100% for all placements. Placement adjustments are applied first, and dynamic bidding works on top. Our bidding strategies guide compares the three.

In the audit, list every campaign that uses up and down together with a large top of search adjustment. Those are the campaigns where actual CPC most often drifts far above the bid you think you set.

7. Placements: where your ads show and what each costs

Sponsored Products ads show in three main placements: top of search (the first row on page one), rest of search (other search result slots) and product pages (detail pages and other non-search pages). Each can be raised by 0% to 900% at campaign level. Adjustments can only increase bids, so the base bid has to be right for the weakest placement. The full method is in our placement bid adjustments guide.

In the audit, open each important campaign's Placements tab in Campaign Manager (or download the placement report) and calculate conversion rate and ACoS per placement.

Worked example: one healthy placement hidden in a 35% average

Say an exact match campaign sells a $49.99 product, target ACoS 30%. Current settings: base bid $1.40, top of search +30%, rest of search 0%, product pages 0%, fixed bids. Thirty days of placement data:

The blended 35.3% suggests a campaign slightly over target. In fact top of search is on target, and the problem is mostly product pages.

The new setup is a $0.60 base bid with +200% top of search and +75% rest of search. Two cautions before you apply it. First, $1.40 to $0.60 is a 57% cut in the base bid; make it in two or three steps and watch impressions. Second, keep this campaign on fixed or down-only bidding, because +200% with up and down could take top of search to $3.60 per click.

Off-Amazon placements

Sponsored Products ads can also run on third-party sites and apps. Campaigns have a setting for ads served off Amazon that lets you lean toward reach or limit that spend, and the placement report can show an off-Amazon row. Trade press also reported that from August 2026 Amazon began showing Sponsored Products through creators in its influencer programme, with existing campaigns enrolled by default. Top of search and product page adjustments do not apply off Amazon. In the audit, check that row's ACoS and decide, campaign by campaign, whether to keep the default. The exact setting names have changed during rollout, so check the campaign's settings in your console.

8. Campaign structure and naming

Structure problems show up as an account you cannot read, budgets going to the wrong traffic, and bid changes hitting targets you did not mean to touch. Our campaign structure guide covers how to build a clean setup. Here is how to audit an existing one.

What to check

• One job per campaign. Budget, bidding strategy and placement adjustments are all set at campaign level. A campaign that mixes exact keywords, broad keywords and competitor ASINs cannot have the right settings for all three.
• Related products grouped sensibly. Variations that share keywords (sizes, colours) can share campaigns. Unrelated products in the same ad group share bids that suit neither.
• Portfolios. Grouping campaigns by product or brand line makes the advertised product report and Campaign Manager far easier to review.
• Naming. Can you tell from the name alone which product, ad type, targeting type and match type a campaign covers? If not, filtering and bulk edits become guesswork.
• Dead weight. Paused or archived campaigns with old negatives, enabled ad groups with no active ads, and ads for ASINs you no longer sell.
• Duplicates. The same keyword in the same match type in two enabled campaigns for the same product.

A simple naming pattern many sellers use is Brand | Product | Ad type | Targeting | Match, for example BrandX | Cooling Pillow Queen | SP | KW | Exact. Pick one pattern, keep the field order fixed, and leave dates out of names.

9. Budgets and out-of-budget checks

A campaign that runs out of budget stops showing ads until the budget resets at midnight. If that campaign is profitable, you are leaving sales on the table every afternoon and evening. If it is not profitable, more budget just buys more waste.

How Amazon applies daily budgets

The daily budget is averaged over the calendar month. Under Amazon's budgeting policy, a campaign can spend more than its daily budget on a given day, using budget unspent on earlier days; Amazon's default allows up to 100% more, and you can set that to 25% in the account settings. Over the month, total spend stays within your daily budget multiplied by the number of days. Check the current rule in your console's settings.

What to check

1. Which campaigns run out of budget, and when? Campaign Manager flags out-of-budget campaigns, and Amazon's budget tools can show time in budget, estimated missed impressions, clicks and sales, and a recommended budget. Amazon's Sponsored Products budget guide explains these.
2. Are those campaigns at or below target ACoS? Only those deserve more budget.
3. Has waste been removed first? Run the search term checks from section 4 on any campaign before raising its budget.
4. Are budget rules in place for known peaks? Schedule-based rules can raise budgets for events and busy days; performance-based rules can raise them when a campaign hits a metric you choose.
5. Is budget spread to match results? Proven exact and branded campaigns usually deserve more of the budget than discovery campaigns.

Amazon's missed-click figures are estimates, so treat them as a guide. If the campaign's ACoS is under target, raise the budget in steps and check that ACoS holds as spend increases; the extra clicks later in the day are not always as good as the morning ones.

10. When it is not a PPC problem

Ads send shoppers to your product page. What happens next depends on the page, the price, the reviews and whether you can actually sell the item. If those break, every ad metric gets worse at once, and cutting bids makes it worse again.

Signs the problem is the product, not the campaigns

• Conversion rate fell across all campaigns and targets at once, not just one.
• Organic sales fell at the same time as ad sales.
• CPC is stable but ACoS rose. If clicks cost the same but return fewer sales, something changed after the click.

When you see these, open Reports > Business Reports > Detail Page Sales and Traffic by Child Item and compare Unit Session Percentage (units ÷ sessions) and Featured Offer (Buy Box) Percentage with the previous period.

The product checklist

Our conversion rate guide goes deeper on fixing the page itself.

Free toolListing & Title CheckerCheck the advertised listing's title, bullets and content for gaps before you blame the campaigns for a low conversion rate.Open tool →

Free toolReorder Point CalculatorMake sure your best-advertised products will not run out of stock mid-campaign, which wastes the rank and data your ad spend built.Open tool →

11. Sponsored Brands and Sponsored Display, briefly

Most of the audit applies to Sponsored Brands and Sponsored Display too, with a few differences that change how you read the numbers. See Sponsored Products vs Brands vs Display for the full comparison.

• Attribution is longer and wider. For sellers, Sponsored Brands credits sales of any product from your brand within 14 days of a click. Sponsored Display also uses 14 days, and vCPM (cost per thousand viewable impressions) campaigns can count views as well as clicks; Amazon updated view attribution for vCPM display campaigns from 1 January 2026. Never compare their ACoS directly with Sponsored Products.
• Judge them on the whole brand. Look at TACoS and total brand sales over the same period, plus new-to-brand metrics, which show how many orders came from shoppers who had not bought your brand recently.
• Search terms still matter. Sponsored Brands keyword campaigns have their own search term report. Apply the same click threshold and negative checks.
• Check landing pages and creative. A store page full of out-of-stock products wastes clicks no matter how good the bids are.

12. Launch vs mature: different products, different scorecards

The same numbers mean different things depending on how old the product is. An audit that applies one target to everything will either strangle launches or let mature products waste money. Our product launch strategy guide covers the launch side in detail.

Free toolFBA Profit CalculatorConfirm the real profit per unit, with current FBA and referral fees, before setting a launch loss limit or a mature target ACoS.Open tool →

13. How often to audit: a weekly, monthly and quarterly rhythm

An audit works best as a routine, because waste comes back if nobody watches it. Split the work by how quickly each problem can cost money.

Also run an unscheduled check after anything that changes the economics: a fee update, a price change, a new shipment at a different cost, a big deal, or the start of the holiday period.

14. Common audit mistakes

Our list of general Amazon PPC mistakes covers campaign management. These are the ones specific to auditing.

1. Starting with keywords instead of the numbers. Without break-even and target ACoS, nothing has a standard to be judged against.
2. Auditing the account average. Averages hide the products losing money.
3. Cutting everything above target. Some high-ACoS terms are key ranking searches. Lower the bid rather than delete them, and check TACoS first.
4. Raising budgets before removing waste. You fund the same bleeding search terms at a larger scale.
5. Blaming PPC for a product problem. Check conversion rate, Buy Box, stock and price first.
6. Changing everything at once. You cannot tell what worked. Make the biggest change first and give it one to two weeks.
7. Not saving the data. Without saved reports, next quarter's audit has nothing to compare against.

Prioritize by dollars, not by count

Finish the audit by putting a rough monthly dollar figure next to each finding, then work from the top. Using the examples in this guide for one product:

Note that the word-level savings may overlap with the wasted-spend total, so do not add every line together. The point is order: do the biggest items first, then the rest.

Quick-reference formulas

Closing

A good Amazon PPC audit follows the money in order: set your break-even and target ACoS, check whether advertising is growing the business through TACoS and organic sales, then clean search terms, fix bids from revenue per click, set placements from their own conversion rates, tidy the structure and feed budget to what works. Before cutting anything, make sure the product itself can still convert. Do a light version every week, a full one every month and a structural one every quarter, and keep a log so each audit builds on the last.

Interactive

Score your account

Tick every statement that is true for your account today. Your score updates as you go and is saved in this browser only.

Audit score0/100

Grade—

Checked0/27

ResetCopy fix list

Frequently asked questions

Amazon Ads features and limits change often. Checked in October 2026. Sotrics is not affiliated with Amazon.