Amazon PPC Mistakes: 10 That Waste Spend and How to Fix Them

The Amazon PPC mistakes below all waste money in ways your own reports can show: search terms nobody reviews, bids changed before sales have settled, budgets that run out on the campaigns that work, and ads on products about to run out of stock. Each comes with the symptom, how to confirm it and the fix, in the order we check them on a new account.

Most are Sponsored Products mistakes. The ones on negatives, branded terms, attribution and budgets apply to Sponsored Brands too.

1. Running broad and auto campaigns without negatives

Broad and auto campaigns find new search terms, and without negatives they keep paying for the irrelevant ones too.

Symptom: search terms with clicks and no orders, such as other product types, sizes you don't sell, or words like "free" and "second hand".

How to check: sort the search term report by spend, using dates that ended at least 7 days ago (14 for vendors) so sales have settled. Compare each zero-order term's clicks with the clicks you expect per order, which is 1 ÷ conversion rate (orders ÷ clicks). Worked example: at 10% conversion, 25 clicks without an order is two and a half orders' worth of chances. Check the term's longer history and stock first; a term that went quiet during a stock-out needs patience, not a negative.

Fix: add single irrelevant terms as negative exact and irrelevant themes as negative phrase; give relevant terms that haven't converted yet a lower bid instead. In auto campaigns, compare the four targeting groups (close match, loose match, substitutes, complements) and cut the bid on the most wasteful. More in our guide to finding and adding negative keywords.

2. Never moving converting search terms into exact match

A search term that keeps converting in an auto, broad or phrase campaign needs its own exact-match keyword; only then do you control what you pay for it.

Symptom: the same converting terms appear in discovery campaigns month after month, with no exact-match keyword for any of them.

How to check: filter the search term report for terms with orders in auto, broad and phrase campaigns, and look each one up in your exact-match keywords.

Fix: add the term as exact match in a manual campaign, with a maximum cost per click (max CPC) worked out from its own numbers:

Max CPC = average order value × conversion rate × target ACoS

Worked example: ₹999 order value × 10% conversion × 25% target ACoS = about ₹25. Then add the term as negative exact where it came from, so the two don't compete. Search terms that are ASINs, shown when your ad appeared on that product's page, belong in product targeting.

3. Mixing branded and non-branded search terms

Branded searches convert cheaply because the shopper already wants you, so mixing them with category terms hides what the category terms cost.

Symptom: a campaign's ACoS looks healthy, but a large share of its sales comes from search terms containing your brand name.

How to check: filter search terms for your brand name and its misspellings, and work out ACoS for each group. Worked example: a campaign spends ₹30,000 for ₹1,20,000 of sales, a 25% ACoS. Branded terms took ₹5,000 for ₹60,000 (8.3%); non-branded took ₹25,000 for ₹60,000 (41.7%), well above a 30% break-even ACoS (margin before ad spend ÷ price).

Fix: give branded terms their own campaigns and add your brand name as negative phrase in the non-branded campaigns. Judge branded campaigns on protecting your brand's search results, and non-branded ones on ACoS against break-even and on new-to-brand orders (from shoppers who haven't bought your brand on Amazon in 12 months).

4. Putting unrelated products in one ad group

Every product in an ad group shares its keywords and bids, so unrelated products end up paying for clicks meant for each other.

Symptom: one ad group holds products with different uses or prices; spend piles onto one or two, and search terms that suit one product show up against another.

How to check: run the advertised product report and compare spend, sales and ACoS for each product within the same ad group.

Fix: give each product theme its own ad group, with products that share keywords and have similar prices and margins. Sizes and colours of one product can share; a low-priced accessory and a premium main product can't, because max CPC rises with order value (mistake 2).

5. Using "up and down" bidding on unproven terms

With the dynamic bidding strategy "up and down", Amazon can raise your bid in real time when it predicts a sale is likely, most of all at top of search. That suits terms that already convert at a profit; on unproven terms it adds cost before you know they work.

Symptom: new keywords or discovery campaigns show an average cost per click close to or above your bid, with few orders.

How to check: note the campaign's bidding strategy and placement adjustments, which stack, and compare average cost per click with the bid on targets with few orders. Worked example: a ₹10 bid with a +50% top-of-search adjustment is already ₹15 there before "up and down" raises it further.

Fix: run new and discovery campaigns on "down only" (Amazon may lower the bid when a sale looks less likely, never raise it) or fixed bids. Get the base bid right first, since the strategy only adjusts around it, and move to "up and down" once settled ACoS is below target.

6. Ignoring placement performance

A keyword can convert very differently at top of search, in the rest of search and on product pages, and the campaign average hides it.

Symptom: a borderline campaign whose placements nobody has checked, or a large top-of-search adjustment never tested against results.

How to check: in the placement report, compare spend, sales, ACoS and conversion rate for each placement, on settled dates.

Fix: placement adjustments only raise bids, so to pay less on product pages, lower the base bid and raise the top-of-search adjustment to hold top of search where it was. Worked example: on a ₹10 bid, product pages run at 45% ACoS and top of search at 20%. A ₹7 base bid with +43% at top of search keeps top of search at about ₹10 and drops every unadjusted placement to ₹7.

7. Changing bids before attribution settles

Bids changed on the latest few days are changed on incomplete sales: Amazon credits orders back to the click date for 7 days on seller Sponsored Products and 14 days on vendor Sponsored Products, Sponsored Brands and Sponsored Display.

Symptom: bids move every few days, often up then down again, and targets that looked expensive look fine once the window passes.

How to check: note last week's ACoS the day after it ends and again once the window has passed; the gap is how far recent numbers overstate cost. The campaign's history tab shows bids going back and forth.

Fix: decide only on dates older than the window, once a target has had enough clicks to expect a couple of orders (worked example: about 20 clicks at 10% conversion), and change each bid at most once per window. More in our guide to TACoS vs ACoS.

8. Letting profitable campaigns run out of budget

A campaign that hits its daily budget stops showing ads for the rest of the day, so a profitable campaign with a small budget loses sales while weaker ones keep spending.

Symptom: campaigns with ACoS below target show as out of budget in campaign manager on many days.

How to check: list the campaigns below target ACoS on settled dates and count the days each ran out of budget. Hourly data shows when spend stops.

Fix: move budget from campaigns above break-even ACoS to capped ones below target. If total budget is fixed, lower bids on the capped campaign's weakest targets so it lasts the day, or, if sales cluster at certain hours, use dayparting to put more budget there.

9. Advertising products that are low on stock

Ads on a product about to run out bring the stock-out forward. Amazon stops them only once the product is out of stock or loses the Buy Box, and its seller help lists availability and sales history among the factors behind search placement, so a stock-out costs rank too.

Symptom: ad spend on a product holds steady or rises while its stock falls, then its ads stop delivering.

How to check: compare each advertised product's days of cover with its restock lead time, the days from ordering stock to having it available to sell.

Days of cover = units available ÷ average daily units sold

Fix: when cover falls below lead time, lower that product's bids and budgets so stock lasts until the restock lands, and move the budget to products and variations still in stock. Restore bids when the restock is received, not when it ships. This slows a stock-out; only reordering earlier prevents one.

10. Judging everything by ACoS alone

ACoS tells you what ad-attributed sales cost, not why a campaign is expensive or whether ads grew the business, and a bid cut fixes only one of its causes.

ACoS = CPC ÷ (conversion rate × average order value)

Symptom: you cut bids on every campaign above target and total sales fall, or ACoS improves while TACoS (ad spend ÷ total sales) rises.

How to check: find which part of the formula moved. A higher CPC points to the auction; a lower conversion rate to the listing (price, reviews, images, variant stock); a lower order value to cheaper variants or a price cut. Worked example: at an ₹8 CPC on a ₹999 product, a 5% conversion rate costs ₹160 per order, a 16% ACoS, while a 12% rate costs ₹67, a 6.7% ACoS, on the same bid.

Fix: set a break-even ACoS for each product from its margin (our profit calculator works it out after fees and GST), fix the part that moved, and judge the account on TACoS and total sales. More on the causes in why ACoS gets too high.

Which Amazon PPC mistakes to fix first

If you act on only two Amazon PPC tips from this list, make them weekly negatives (mistake 1) and bid changes on settled dates only (mistake 7): neither costs anything to start. Structure comes next: exact-match harvesting, separate branded campaigns and cleaner ad groups. Bidding strategy, placements and budgets come last, because judging them needs settled data from the new structure. For our clients, cutting 30–40% of wasted ad spend is a typical outcome of this work, never a guarantee.

ScaleSKUs flags several of these patterns from Amazon's official API data, including search terms to negate or harvest, profitable campaigns held back by budget and advertised products that are out of stock. Each fix comes to you as a task, and nothing changes on Amazon without your approval. See how it works.

Frequently asked questions

Why are my Amazon ads not converting?

Check three things in order. First, whether the search terms match the product, in the search term report. Second, whether the listing converts shoppers overall, using unit session percentage (units ordered ÷ sessions) in Business Reports; if it is low there too, the problem is the listing or the price, not the ads. Third, whether the product is in stock and winning the Buy Box.

How often should I optimise Amazon PPC campaigns?

Review search terms, budgets and stock weekly. Change a target's bid at most once per attribution window: 7 days for seller Sponsored Products, 14 days for Sponsored Brands, Sponsored Display and vendor Sponsored Products. Review structure and placements monthly.

Should I turn off auto campaigns?

No. Auto campaigns find search terms and product pages you might never target yourself, and they feed your exact-match campaigns. Keep them, add negatives, set bids by targeting group and harvest what converts.