By Ted Gibson — RippleEffect
Amazon PPC strategy is four jobs in a fixed order: find the search terms that convert, move them somewhere you control the bid, stop paying for the ones that never convert, and decide how much of the budget defends demand you already have. That list does not change with the size of the account. What changes is which of the four is your constraint, and how much structure you can carry before you lose sight of the account. This is written for a brand owner doing £20,000 a month or more on Amazon UK, or $50,000 a month or more in the US, with a catalogue rather than one hero product. Almost every published strategy guide, Amazon's own included, is written for the account below that, and none of them says so. One boundary before the rest: this is about what the account is made of and what runs inside it. What to do when the numbers are wrong is a different question, what counts as a good number is another, and how much should sit on your own brand name is a third. Structure is the one that gets skipped, because it is the least satisfying thing to change and the most expensive to undo.
Amazon publishes a strategy, and it is built for a smaller account than yours
Amazon's guide to targeting with Sponsored Products names six campaign strategies, and as a starting taxonomy it beats most of what ranks above it: expand, promote, protect, conquest, upsell and cross-sell. It gives two structural rules worth keeping, too. Avoid mixing targeting strategies inside one campaign or ad group, because if you do you cannot tell which tactic produced the number you are looking at. And if you have just a few ASINs, run one ASIN per campaign with a budget you are willing to spend on that ASIN. Then read what the best-practices guide recommends you open with: a daily budget of $10 or equivalent. A brand turning over £50,000 a month and spending 12% of it on advertising is running roughly £200 a day, and the same arithmetic in dollars gives $200 a day on $50,000. Amazon's suggested opening position is about an hour of that. The guidance is not wrong, it is written for a different business, and the phrase "if you have just a few ASINs" is the only signal it gives you.
The allocation tables disagree, and the disagreement is about the denominator
Two agencies publish a split across ad types and land on identical percentages, which is where it gets interesting. Canopy Management gives a baseline of 60 to 70% Sponsored Products, 15 to 25% Sponsored Brands, 10 to 20% Sponsored Display and zero to 10% for DSP and Sponsored TV. GigaBrands gives 60 to 70%, 15 to 25% and 10 to 20% across the same three formats. The numbers match. What does not match is who they are for. Canopy's baseline is for brands spending $20,000 to $100,000 a month on ads. GigaBrands' framework is for brands doing $50,000 to $500,000 a month in revenue, with a rule of 10 to 15% of revenue going to advertising on mature products.
Work it through and the two pages hand the same brand opposite instructions. A brand at $50,000 a month in revenue, on GigaBrands' own 10 to 15% rule, spends $5,000 to $7,500 on advertising. That is well under Canopy's $20,000 floor, and Canopy has a separate instruction for brands below it: simplify, 80% into Sponsored Products, 15% into Sponsored Brands on your top five to ten keywords, 5% into Sponsored Display, and do not fragment a small budget across too many campaign types. So on Sponsored Display one page says 5% and the other says 10 to 20%. On £7,500 of monthly spend that is £375 against £750 to £1,500, and on $7,500 it is the same gap in dollars. Neither publisher is careless. One bands by ad spend and the other by revenue, the two are roughly ten times apart, and the reader is the person who has to notice. Before you use anyone's allocation table, including ours if we ever publish one, check which number it is a percentage of. Canopy at least says inside the post that its figures reflect patterns in its own client base and should not be read as official statements from Amazon. GigaBrands opens by asserting that the average Amazon seller wastes 35 to 40% of their budget, and cites nothing for it.
The only published number that measures structure rather than tactics
Every guide on this subject asserts that structure matters and not one of them measures it, with a single exception, and it belongs to Amazon. On its product strategy guide, Amazon states that ad groups containing similar products saw 6% more units sold on average when advertising with Sponsored Products, compared to those with less similar products. The footnote reads Amazon internal data, worldwide, 1 January to 31 December 2023, with the sales ad-attributed, so it is an interested party's own unaudited figure from a calendar year that is now some way behind us. It is still the only published measurement of campaign structure in the category, and nobody ranking for this query cites it. The mechanism sits in the sentence above the number and is the more useful half: similarly priced items let spend split more evenly inside an ad group, and similar products share the same keywords. That makes grouping a spend-distribution decision rather than a tidiness one. It keeps one product in an ad group from quietly taking the budget of the four beside it.
What you stop doing
This is the half nobody writes, because adding is easier to describe than removing, and three habits stop scaling. The first is a campaign for every SKU and match type. GigaBrands says it typically runs 8 to 12 campaigns per SKU across match types and ad types, which on a 30-SKU catalogue is 240 to 360 campaigns, each carrying its own daily budget. A daily budget is not a cap you can safely ignore either: Amazon's own worked example is that a $100 daily budget may take up to $3,000 of clicks in a calendar month, so money moves between days without you. Hundreds of small budgets means hundreds of places for spend to sit idle or run ahead of you, and nobody reviews 300 campaigns a week with any real attention. Consolidate to the level at which you actually make decisions, which for most catalogues is the product family rather than the ASIN.
The second is inheriting settings you never chose. Amazon lets you add placement bid adjustments of up to 900% across top of search, rest of search and product pages. That is a different control from the dynamic bidding strategies we have written about elsewhere, and it is typically set once during a launch and then carried into every campaign copied from that one. With a handful of campaigns you would spot it. With two hundred you will not, and a ceiling that high means a decision nobody remembers making can multiply what you pay for your most expensive placement. Audit placement modifiers before you touch a single bid. The third is treating every new ad format as a decision that needs making this quarter. Canopy puts DSP and Sponsored TV at zero to 10% and says to add them only after the efficient spend in the core formats has been captured, which matches what we see: DSP earns its place after the sponsored account is in order, not while it is being fixed.
On order, Amazon's own instruction is the plainest sentence published anywhere on this: start with an automatic campaign, and let it run for about two weeks before creating a manual one. That loop does not change with size, it just runs across more products at once. Discovery, then harvest what converted into campaigns where you set the bid, then negatives, then structure, then formats. Sponsored Brands and Sponsored Display come after that sequence rather than alongside it. The three-year story behind our headline case study ran in that order, with programmatic arriving late, which is why we run the five disciplines as one system rather than as a standalone advertising job. There is also a version of this where none of it applies to you, and ruling that out is cheaper than a rebuild. Open last month's search term report and look at where the money actually went. If most of it landed on terms you deliberately chose, and you can state your campaign count from memory, structure is not your constraint, and rebuilding buys you nothing at the price of a fortnight of learning data. The problem is elsewhere, most often the listing or the price. If you cannot answer either question, that is your answer. A free account audit puts the real shape of the account in front of you in priority order, and what the work costs afterwards is published.