Amazon PPC optimization is not a project you finish, it is a short list of tasks run on a fixed cadence: search terms and bids every week, budgets and placements every two weeks, structure and impression share every month. This is the routine task by task, with Amazon's own thresholds for when a term has enough data to act on.

Why Amazon PPC optimization is a routine, not a project

Most accounts are optimised in bursts. Someone notices ACOS climbing, spends a Saturday cutting bids and adding negatives, and then nothing happens for six weeks. The account improves for a fortnight and drifts straight back, because the thing that caused the drift was never a one-off.

Almost every input moves on its own. Competitors raise and drop bids. Amazon keeps feeding new shopper searches into your automatic campaigns. Demand shifts with the season, your stock level changes what you can afford to sell, and a keyword that was profitable in September can be bought by somebody with deeper pockets in November. An account is not a thing you fix; it is a thing that needs reading.

So the useful unit of Amazon advertising optimization is a cadence, not an effort. A fixed list of tasks, each on a fixed interval, with one change made at a time so you can still tell what the change did. That discipline is most of what Amazon advertising management actually is, and it is reproducible on your own account without any tool you do not already have.

Every week, part one: the search term report, worked in money order

Open the Sponsored Products search term report and sort by spend, descending. You are building two piles. Terms that are producing orders at or below your break-even ACOS go into the harvest pile. Terms that are spending and producing nothing go into the negation pile. Work from the top, because the money is at the top.

Negation is where sellers go wrong in both directions. Amazon's own targeting guidance is to evaluate a search term's performance after it has received at least 20 clicks before acting on it as a negative, and to use phrase or exact match rather than broad when you add one. Cutting a term on three clicks throws away keywords that would have worked; leaving an obvious loser running for a quarter is just paying for it. How to find wasted search terms covers the sorting, and negative exact and negative phrase covers which one to use.

Harvesting is the other half: a proven search term is moved into its own exact-match campaign or ad group, where you control its bid directly instead of inheriting whatever the automatic campaign decided. What search term harvesting is walks through the move. On a brand-new automatic campaign, Amazon recommends letting it run about two weeks before you build a manual campaign from what it found, so new accounts should expect the harvest pile to be empty for the first fortnight.

Every week, part two: one bid move, toward break-even

Bids are set against your break-even ACOS and nothing else. Break-even ACOS is your margin after every Amazon fee and the cost of goods; above it you are buying rank or volume on purpose, below it you are making money. An industry-average target is somebody else's cost structure. How to calculate break-even ACOS is the arithmetic, and it takes ten minutes once.

With the number written down, the weekly bid pass is small: move the bids that are clearly on the wrong side of it, in modest steps, and leave the rest. Amazon's targeting guidance also gives a sensible default for manual campaigns running several match types on the same keyword, which is the highest bid on exact, a lower bid on phrase and the lowest on broad, so the match type that should win the auction usually does.

Two restraints make the routine readable. First, Amazon recommends testing match types for at least one to two weeks to gather enough data before pausing any keywords, which is a fair floor for judging a bid change too. Second, do not change a bid, a match type and a placement modifier on the same campaign in the same week, because then no result can be attributed to any of them. If you are deciding between fixed bids and dynamic bidding, Amazon recommends dynamic bids up and down to maximise performance, and suggests dynamic down only when you are optimising for return on ad spend; what a bid strategy is explains the difference.

Every two weeks: budgets and placements

Amazon's own budget guidance is to check how well your current budget is serving your campaigns at least once every two weeks, and that is a good interval because nothing here changes daily. The question is narrow: which profitable campaigns are running out of budget, and which unprofitable ones are not capped?

A campaign that goes out of budget stops showing for the rest of the day, and Amazon's guide is explicit that ads cannot show again until the daily budget resets at midnight. On a campaign that is beating your target, every one of those hours is a gap you paid nothing for and earned nothing from. The Budgets page in the console puts a number on it: Amazon's recommended budget there is the estimated missed clicks multiplied by the estimated cost per click, where missed clicks are modelled against similar campaigns that stayed in budget and the CPC comes from your own history.

One thing not to panic about. Amazon averages the daily budget over a calendar month, so a single day can come in under the amount or up to 25% over it. A $10 daily budget showing a $12.50 day is the system working as documented, not an overspend to chase. While you are in there, the same fortnightly slot is the right place for placement bid adjustments across top of search, rest of search and product pages, which Amazon allows up to 900%; what a placement modifier is covers how to read the placement report before you move one.

Every month: impression share, structure and the numbers a weekly pass hides

Monthly is when you ask how much of the category you are actually getting, not just how your own spend performed. The search term impression share report for Sponsored Products, downloaded from the Report Center, gives you that directly: Amazon's example is that an impression share of 20% for a search term means your ads won 20% of all Sponsored Products impressions for that term on that date, and an impression rank of 3 means you received the third most impressions for it. What impression share is explains how to act on a low number, because the fix is sometimes budget and sometimes relevance, not always bid.

Monthly is also when structural drift shows up. The same keyword running in three campaigns at three different bids, so you are bidding against yourself. Automatic campaigns that have never been harvested, where proven terms are still spending at a bid you do not set. Profitable campaigns sitting under a budget cap that was right six months ago. None of these appear in a weekly search term pass, because none of them is a search term.

Finally, the monthly read is where ACOS stops being the only number. Total advertising cost of sale, and a split between branded and non-branded spend, tell you whether advertising is growing the business or just taking credit for sales you would have made anyway. Both move too slowly to be worth looking at weekly, and both are the first thing a serious report should show.

What the routine deliberately leaves alone

No re-cutting of negatives inside two weeks. Amazon's guidance is to let negative keywords run for two weeks or more before changing strategy again, and a negative that is reversed after four days has told you nothing except that four days is not a sample.

No weekly restructure. Rebuilding campaign structure resets the learning on everything inside it, so it belongs to a planned change with a reason, not to a routine. And no reacting to a single bad day: the whole point of a cadence is that you compare a week to a week.

The biggest thing the routine leaves alone is the listing. If your conversion rate is the bottleneck, every hour of Amazon PPC optimisation is spent making a page that does not convert slightly cheaper to send traffic to. Check unit session percentage in the Business report before you accept that the account's problem is the ads at all.

The calendar, in one place

Weekly: search term report sorted by spend; harvest the proven terms into exact; negate the spenders with no orders, once they have the clicks to justify it; one bid pass toward break-even ACOS. Fortnightly: budgets, out-of-budget profitable campaigns, placement adjustments. Monthly: impression share, structure, TACOS and the branded split, and a written note of what changed and what it did.

An hour a week covers it on a small account, and the written note at the end is what turns the hour into something you can learn from. If you already run the routine and want a second opinion on the decisions rather than someone to take the keyboard, that is what Amazon consulting is for.

If the hour is the thing you never have, this list is exactly the work our Amazon ads management services do on a schedule, and the free audit that starts it needs three exported reports and no account access.

Sources