This chapter covers the mechanics a launch's advertising is actually built from: how Amazon's per-click auction lets a listing with zero sales buy page-one placement, how to derive bids and daily budgets from price, conversion rate and target ACoS instead of accepting Amazon's suggestions, which of the three bidding strategies and which placement multipliers to use when, and how to carve an account into campaigns — by product, match type, keyword volume, and shopper intent. It then works through the two targeting families in detail: automatic campaigns as a discovery engine that demands constant negative-keyword pruning, and product/ASIN targeting for competitor sniping and brand defense, ending with how proven search terms get harvested into dedicated ranking and scaling campaigns. The material is overwhelmingly Sponsored Products; Sponsored Brands and Sponsored Display appear here mostly as budget lines and brand-defense surfaces rather than as campaign types with their own mechanics.
Amazon PPC is a real-time keyword auction. Advertisers bid on keywords, and the highest bidder wins the top ad placement for that search — regardless of the seller's organic sales history or Best Seller status (Amazon PPC Auction Model). That single property is why advertising is the first lever available at launch: a brand-new listing has no sales velocity to earn organic placement, but it can win a first-page ad placement on day one purely by bidding competitively.
Two details of the mechanism matter operationally. First, you pay per click, not per impression, so a high bid ceiling does not by itself produce a high charge — it only decides whether you're eligible to be shown. The framing to carry through the rest of the chapter is that a bid "doesn't equal your cost per click, it just tells Amazon whether or not you're in the game" (PPC Starting Bid Formula (Target ACoS × Price × Conversion Rate)); actual CPC is settled at auction and typically lands below the bid. Second, Amazon paces delivery across the day against the daily budget, which means a campaign can win auctions and accumulate impressions without exhausting its budget immediately — and, less admirably, means sellers can inflate exposure by setting daily budgets well above their intended real spend.
The urgency comes from the PPC Honeymoon Period: a short window right after a new product goes in stock during which Amazon's algorithm gives the listing a temporary organic-rank boost independent of its (nonexistent) sales history. Since over 70% of Amazon sales happen on page one and a zero-sales listing is otherwise buried well past it, that window is worth catching in full. The practical consequence is a scheduling rule rather than an advertising rule: photos, copy and A+ content all need to be finished and approved before launch (see Listing Content & Conversion Design), because every day spent polishing the listing after it goes live burns honeymoon days that don't come back, and PPC should start the instant inventory is live.
What the spend is buying, at launch, is not profit. Amazon evaluates a search term by comparing a listing's click-through and conversion rate on that term relative to the other products competing for it, and PPC sales are how a new listing generates that evidence — which is the PPC-to-Organic Rank Mechanism. Early PPC losses are therefore better read as the price of rank than as campaign failure. A second, compounding payoff is the PPC Dual-Placement Visibility Effect: with campaigns structured properly, the same ASIN can occupy both a sponsored slot and an organic slot on one results page, so the shopper sees it twice while scanning rather than once.
Two expectation-setting cautions belong here before any money moves. Reading ACoS in the first few days of a campaign is meaningless — 50–100% ACoS readings are common and expected on samples too small to act on, and negating terms on that basis cuts off keywords that would have normalized. And the obvious, dead-on keyword is not the easy win it looks like: the Exact-Match Top-of-Search Trap is assuming that because a query names your product, it will convert at top-of-search. A listing with no reviews or reputation often can't convert efficiently even on a perfect name-to-query match, because ranking and shopper trust depend on more than word overlap. The fix isn't avoiding the keyword; it's not over-committing top-of-search spend to it until reviews and rank exist to back it up (see Reviews & Account Health).
Match types control how closely a shopper's query must resemble your keyword before the ad can trigger (Keyword Match Types (Exact / Broad / Phrase)). Exact demands a close match to the keyword and is the tool for precise control over an already-validated list; broad only requires that the keyword's terms appear in some form, and is for exposure and discovery; phrase sits between them, matching queries that contain the phrase in the same word order. These apply only to keyword targeting in Sponsored Products and Sponsored Brands — Sponsored Display has no keyword targeting at all, only categories, products, audiences and remarketing, so match type is never a Sponsored Display decision.
The names have drifted. Since roughly 2024–2025 Amazon has given itself leeway to substitute words even on exact-match targets: an exact match on "wooden phone holder" can show for "wooden phone stand." Read exact as directionally tight — "almost-exact, with occasional Amazon-driven substitution" — rather than a literal guarantee, which is exactly why loosely-matched campaigns still need negative-keyword hygiene. Broad has drifted the other way, toward autotargeting: Amazon may substitute related terms with no shared words at all, so an "iPhone stand" seed can pick up "phone holder" searches. Broad is best understood as still seeded by your root terms but with far more platform latitude than exact or phrase.
The undocumented fourth type is the broad match modifier (Broad Match Modifier (Undocumented Fourth Match Type)): select broad match in Seller Central and prefix required words with a plus sign, e.g. +phone holder. Any + word is forced to appear somewhere in the customer's query, in any order and not necessarily adjacent to the other required words, while unmarked words keep broad's full substitution latitude. Its real value is forcing inclusion of an audience or avatar term Amazon has no way to infer as relevant — +mom phone holder keeps the ad on that segment while still letting Amazon swap "holder" for "stand." It's described as becoming the most important match type fairly quickly as Amazon leans further into AI-driven placement, because it's the one lever that keeps broad's flexibility while guaranteeing a chosen term isn't dropped.
That points at the chapter's most contrarian principle: the AI Ad Platform "Buffer" Reward Principle — Amazon's auction increasingly rewards advertisers who give its matching room to expand rather than constraining it tightly, the same claim extended to Facebook, YouTube and Google. The concrete data point offered is one account's pancake batter dispenser: a 56% ACoS on the term "pancake batter dispenser" under exact match versus 20% ACoS on the identical term surfaced through broad match. Framed for Amazon specifically as "Amazon's algorithm wants leeway in 2025 and beyond — it wants the flexibility to show your ad when and where it wants."
In practice the sources split on phrase match. One launch strategy skips it entirely, treating phrase queries as a subset of what broad already catches, and builds the Six-Campaign PPC Launch Structure on exact and broad only. The broad match modifier is argued to supersede most of phrase's use case, since it can force one required word while phrase can only preserve word order. Against that, a PPC agency managing a portfolio of brands doing over $6M/month reports using all four match types rather than dropping phrase — so they're not fully interchangeable in practice. The defensible use cases per type are narrower than the debate suggests: exact for keywords already proven profitable (usually surfaced in an autotargeting search term report) or isolated for tight bid control; phrase for surfacing longer-tail variants around a known root, e.g. "wooden phone holder for kids" from "wooden phone holder"; broad for controlled discovery, always reviewed against what Amazon actually matched. Where the keyword lists themselves come from is the previous chapter's job — see SEO & Keyword Strategy: Winning A9, Cosmo & Rufus.
The chapter is unambiguous that Amazon's suggested bid is the wrong number to place: "I can tell you with certainty the answer is no do not use Amazon suggested bid" — it optimizes for filling the auction, not for hitting your margin target. The replacement is arithmetic (PPC Starting Bid Formula (Target ACoS × Price × Conversion Rate)):
Starting Bid = Target ACoS × Product Price × Conversion Rate
The conversion-rate input is pulled, not guessed: Seller Central → Business Reports → ASINs tab → the Unit Session Percentage column. The same relationship is sometimes written through revenue-per-click (Revenue-Per-Click (RPC) Bid Formula): RPC = price × conversion rate, so Bid = RPC × target ACoS. It's the identical formula computed from a pre-derived RPC rather than from price and CVR directly. Whichever form you use, the output is a starting bid, not an expected CPC.
Budget has its own formula rather than a number you can afford. For a product with a sales goal, Daily Budget = Target Units Sold Per Day × Product Price × Target ACoS (Daily Budget Formula (Target Units × Price × Target ACoS)). For a product with sales history, an alternative runs off TACoS instead: monthly ad budget = target TACoS × target total revenue, divided by days in the month. The ceiling on that — the maximum daily spend before ads erase all profit — swaps in break-even TACoS (margin after COGS, shipping, returns, FBA fee and seller fee) times trailing-30-day average daily revenue. Spending right up to that break-even ceiling, at zero ad profit, is presented as a legitimate aggressive play for sellers who can absorb it: trading margin for rank and market share, not a mistake. How to compute break-even TACoS and judge whether that trade paid off is the subject of Profitability, LTV & Customer Analytics.
A new launch has no revenue to run those formulas on, so it gets its own (New-Launch Ad Budget Formula (CPC ÷ CVR × Target Sales)):
Daily budget = (CPC ÷ Conversion Rate) × Target Daily Sales
Every input is estimated. CPC comes from the ad console's historical CPC view if the account has any relevant history, or from averaging Amazon's suggested bids across several primary keywords — the one place suggested bid is treated as usable, explicitly as a budgeting stand-in rather than an endorsement of bidding it. Conversion rate comes from a prior launch, a fallback range of 5–15%, or the category-wide rate across all competing brands from Brand Analytics → Search Analytics → Search Query Performance. Target daily sales should be set close to the average daily sales of page-one competitors (found with a tool like Helium 10) rather than picked arbitrarily, since ranking means roughly matching what's already selling on page one.
One diagnostic keeps budget and bid from being confused with each other: if the daily budget is exhausted too early or barely touched, adjust the budget; but if the underlying problem is low impressions, adjust the bid. A bigger budget does nothing for a campaign that isn't winning enough auctions to spend it.
Finally, none of this becomes campaigns until it reconciles. The Campaign Planner (Pre-Launch Budget Worksheet) is a forcing function: before creating anything in Seller Central, chart every planned campaign — one row each, each with its own budget — in a spreadsheet, on paper, or in planning software, and confirm the individual budgets sum to the top-down total. The planner's rows span SP Auto, SP Exact/Broad keyword, SP Exact product targeting, SB Video, SB Custom Image, SD Product Targeting and SD Retargeting, weighted toward single-keyword ranking campaigns for an aggressive launch or toward product-targeting and video for a leaner one. The rule attached is blunt: never take Amazon's suggested daily budget, never pick a number based on what's affordable, and never set a placeholder "to adjust later." Worth noting that those Sponsored Brands and Sponsored Display rows are about as far as this chapter's material goes on those two ad types — it names them as budget lines and, later, as brand-defense surfaces, but carries no real SB or SD campaign mechanics.
Amazon exposes three bidding strategies, and the chapter treats the choice as a question of how much discretion to hand the algorithm rather than a question of bid size.
Dynamic bids — down only is the steady-state default (Dynamic Bids Down Only (Default Steady-State Bidding Strategy)). Amazon may lower your bid when a click looks unlikely to convert or when competition for a placement drops, but never raise it above your set amount. Chris Rawlings frames it as the safe default: "whenever you're in doubt, just go with down only." It suits manually-targeted phrase, broad and exact campaigns specifically because bids there are already set per keyword — there's nothing for Amazon to raise, only underperforming auctions to pull back from. In the $100K-in-90-days launch, down-only was applied uniformly across every campaign type — category targeting, semantic phrase families, brand defense and auto — with placement modifiers deliberately withheld until data accumulated.
Dynamic bids — up and down lets Amazon raise the entered bid by up to 100% or lower it (Dynamic Bids Up and Down (Amazon PPC Bidding Strategy)). Its proper use is fixing an impression problem, not a bid problem: a brand-new keyword, a new product target, or a launch that isn't getting shown. Amazon reads the setting as license to prioritize the ad — "giving Amazon more leeway so that it can use its algorithm to optimize everything about your ads." It also fits auto campaigns, where one bid has to cover many search terms of very different competitiveness. It is explicitly a bootstrapping mode: once the campaign reveals its real CPC and CTR, switch back to down-only. Because the ceiling is +100%, Rawlings offers a translation between the two — a $6 down-only bid is roughly equivalent to a $3 up-and-down bid — useful when moving a computed starting bid between strategies.
Fixed bids never move in either direction, which makes placement outcomes predictable (Fixed Bids (Amazon PPC Bidding Strategy)). Their use case is forcing a placement, most often in a single-keyword exact-match campaign built purely to push organic rank, and they're judged accordingly: by rank movement tracked in Helium 10 or DataDive, not by ACoS — "it's not about the return you get on the ad. It's about the ranking improvement that it causes." Note an unresolved wrinkle: the launch-phase guidance says fixed bids are "avoided entirely," while the ranking-campaign guidance prescribes them. Across ranking-focused campaigns generally, the down-only-versus-fixed choice is best read as a dial on aggressiveness — down-only conservative, fixed maximally aggressive.
The launch window gets its own settings (Aggressive Launch Bidding Strategy (Over-Bidding + Dynamic Up/Down)): bid roughly $1 above Amazon's suggested range, and use odd-penny amounts — $1.13 rather than $1.15, $0.53 rather than $0.55 — specifically to beat competitors who default to round five-cent increments, winning marginal tie-breaks for a fraction more spend. Paired with up-and-down bidding, this trades efficiency for maximum competitiveness during the honeymoon window, and is expected to be walked back once optimization begins.
Sitting on top of all three strategies are placement bid adjustments (Placement Bid Adjustments (Top of Search / Rest of Search / Product Pages)) — separate multipliers for Top of Search, Rest of Search, Product Pages, and a fourth for Business Buyer placements. They're a different kind of lever: they fine-tune spend by placement rather than choosing how Amazon may move the bid. Top of Search is the expensive one but carries the highest CTR and the most ranking influence — "the alpha dogs." One documented case measured roughly 9% CTR at top of search versus ~1.3% rest of search and ~0.2% on product pages, justifying a 128% top-of-search adjustment; when the goal is rank rather than efficiency the percentage can go as high as 900%. Guidance on timing conflicts across sources: one line says raise Top of Search first by roughly 20–30%, or test in 10% increments; another says start all three at 0% and leave them for at least two weeks, since which placement wins varies unpredictably by product and early adjustment is guessing. Both agree the Placements report — not intuition — should drive the number, and that placement performance varies per keyword, so a term converting at top of search while another in the same campaign converts on product pages should be bid separately rather than under one blanket premium.
When a placement underperforms, don't slash its adjustment by 80–90%. Reduce the campaign's default bid by 10–20% and reallocate that freed percentage onto the better placements. The same arithmetic underlies the Global Bid Placement Adjustment Hack, which exists because of a hard UI limit — "you can only increase your bids by a percentage that you set here. You can't decrease it." To net-decrease a weak placement: raise the strong placement's multiplier (say Top of Search +10%), then lower every individual keyword's base bid by that same percentage. The raised multiplier cancels the cut for Top of Search, while every other placement nets lower. Apply it as one global change before touching individual keyword bids.
The base structuring rule is One-Product-Per-Campaign Structuring Principle: advertise exactly one product per campaign, never grouping multiple products or variations together. Mixing them blurs attribution — clicks, spend and conversions can't be traced back to a single ASIN — which also breaks every formula in the previous section, since each one assumes a single product's price, conversion rate and target units. Inside the campaign, ad groups are largely dead weight (Ad Groups as a Split-Testing Tool Only (Otherwise Redundant)): the campaign level and its name are what matter for organization and reporting. The one legitimate use for multiple ad groups is split-testing near-identical variants of the same product — different images or listing versions — against identical keyword targets, isolating the variable while targeting stays constant. Otherwise, one ad group per campaign.
That many small campaigns need a filing system. The PPC Campaign & Portfolio Naming Convention gives a concrete template: optional company tag + a two-word product nickname + ASIN + a targeting-type tag (SPA for Sponsored Products Automatic, SPM for manual) + the match type or specific keyword/family. Once you're managing dozens or hundreds of campaigns, this is what lets you read a campaign's identity and performance straight from a bulk view without opening it. Pair it with portfolios organized by product: every campaign for one product, regardless of targeting or match type, goes in that product's portfolio, so spend and sales roll up cleanly per product.
Chris Rawlings adds a second organizing layer — a campaign setup cheat sheet with four goal categories, each carrying a prescribed starting bid, match type, targeting, bidding strategy and naming entry (Campaign Setup Cheat Sheet: Four Goal Categories (Performance / Brand Defense / Research / Ranking)): Performance (convert profitably around a shared shopper-intent theme), Brand Defense (protect your own listings), Research (surface new keyword and product-targeting opportunities, not to convert efficiently), and Ranking (move organic rank, typically on fixed bids). The category determines the bidding strategy: Ranking pairs with fixed bids, a new Performance campaign or launch with up-and-down, a matured Performance campaign with down-only.
Beyond one-product-per-campaign, the chapter names three further axes to split on.
A related sort runs on shopper type rather than intent: Branded vs. Generic Keyword Segregation. Queries containing your own brand name turn up unbidden inside generic broad, phrase and auto campaigns' search term reports. Those shoppers are brand-aware and further down the funnel, not the open-discovery shopper the generic campaign was built for, so they get broken out into a dedicated branded campaign. Similarly, terms that signal storefront or browse intent — shopping a brand generally rather than one unit — belong in a Sponsored Brands/Store campaign, not Sponsored Products.
How dense all this should be is genuinely contested. The six-campaign structure plus a four-way auto split lands around nine campaigns per product. Against that stands Chad Scaling (Simple, Aggressive Campaign Scaling Style) — Rawlings' term for running a small number of straightforward, high-conviction campaign types (chiefly single-keyword exact-match and product targeting) bid aggressively, rather than an elaborate segmented structure. That style took one supplement product past $1M in revenue in 12 months using ten total campaigns.
Autotargeting hands Amazon full control of keyword, category, audience and product targeting. The chapter's position is that auto is widely misunderstood and often abandoned in favor of exact match, but that its value was never its own ACoS (Auto PPC as Launch & Keyword-Discovery Tool (Not a Profit Center)): it's a keyword-discovery mechanism and a launch rank driver. Because Amazon itself picks the targets, auto surfaces terms your research missed — in one case, auto data revealed a seller was mistargeting search terms entirely, and rewriting the title and building exact-match campaigns around what auto was actually converting on made the listing "take off." Auto should run alongside exact-match campaigns, not instead of them; a product only shows once in results, so overlap isn't waste.
Amazon's automatic targeting sorts matches into four groups — Close Match, Loose Match, Substitutes and Complements (Automatic Campaign Four-Way Split (Close Match, Loose Match, Substitutes, Complements)). Close and loose are search-term-based; substitutes places your ad on detail pages of the same kind of product (other water bottles for a water bottle), complements on adjacent products (a cheeseboard ad on a knife-set listing). At campaign creation Amazon offers two configuration modes: set default bid, which gives Amazon full latitude across all four and leaves negative targeting as your only lever, or set bids by targeting group, which exposes a separate bid per group. The chapter consistently recommends the latter.
Whether to then split the four into separate campaigns is where its sources disagree openly. The dominant recommendation is isolation — one group per campaign or per ad group, never mixed — because a high-intent close-match click and a low-intent complements click otherwise average into the same numbers. Practical sequencing: launch with close match only, disable the rest, re-enable each as its own campaign when there's budget and attention to judge it. Close match and substitutes are reported to outperform loose match and complements early. Against that, one documented case deliberately combined all four inside a single campaign using per-group bids (Combined Auto-Targeting Campaign (Set Bids by Targeting Group)) and outperformed the split — with loose match as the surprise top performer and substitutes turned off entirely. The rationale offered is the leeway principle again: combining, while still tracking per-group bids, lets Amazon route traffic more effectively than a hard split. The honest reading is that match-type winners can't be predicted per product without testing.
Bid discovery inside auto has its own tactic. The Auto Campaign Four-Bid Ladder Strategy launches four otherwise-identical copies of the same auto campaign at different bids — starting at Amazon's minimum suggested bid and stepping up 10 cents each, e.g. 30¢/40¢/50¢/60¢, up to just below the maximum suggested bid — then compares ACoS once data accumulates and scales the winner. The claimed mechanism is that different bids place the campaign into different competitive inventory, changing who you compete against rather than just how much you spend, so "you never know what bid is going to produce the best results." The reported result: a brand with a 35% ACoS target got 18% ACoS from the 40¢ rung, dramatically beating 30¢, 50¢ and 60¢. This was discovered internally by Sophie Society teammates Chris Pollock and Patrick, not documented by Amazon, and is explicitly caveated as inconsistent — "when it does work, it can help you find a real gemstone," but it needs testing per listing. A simpler tell sits alongside it: if an auto campaign isn't spending its full daily budget, the bid is too low. One UK-denominated starting point puts auto bids at 50p–£1, raised in 10–20p increments until spend catches up, on a daily budget of at least £10 so the campaign generates enough search-term data to judge.
How much of the account auto should be is contested. A live audit prescribes roughly 10% of ad budget for auto, used purely for keyword harvesting, with the rest going to manual exact/phrase campaigns built from vetted keywords plus product targeting (Auto Campaign as ~10%-of-Budget Keyword-Harvesting Cap). The diagnosis behind it: a five-year-old candy brand plateaued at ~$5K/month partly because spend sat in broad auto and its own branded keyword — auto surfacing head terms dominated by category giants, and branded clicks that weren't incremental. Broad match shares the same discovery role and the same warning: neither is meant to be the long-term profit driver, and both get expensive left un-optimized. Pulling the other way, the leeway argument holds that auto — historically "a money pit" — can now be a top performer, and one lean launch used actively-pruned auto as one of four campaign types on the way to $100K in 90 days at 14% average ACoS. A separate, purely economic reason to keep auto running: because most sellers concentrate their competitive bidding in exact and phrase campaigns, the same search term can sometimes be won more cheaply through auto (Auto Campaign Cost Arbitrage vs. Exact/Phrase Match) — fewer competitors in that channel means lower effective CPC for the same placement, which makes auto worth testing even for terms you've already validated.
What makes auto workable in either camp is pruning. A negative keyword tells Amazon not to show the ad for a term, excluding traffic that won't convert without lowering bids on the keywords you want (Amazon Negative Keywords). They go in two phases: pre-launch common-sense negatives for obviously mismatched but similarly-worded searches, then ongoing negatives mined from search term reports as non-converting terms surface. Two controls matter — negative exact blocks one literal query, negative phrase blocks anything containing a word, so phrase-negating "metal" and "chrome" clears an entire class of traffic off a wooden phone holder listing. Negative hygiene scales with looseness: exact campaigns need the least, phrase more, broad more still, auto the most, since auto has no keyword-level guardrails at all. Timing matters as much as the list — wait for the first search term report roughly 24 hours after launch, add clearly irrelevant terms, but don't negate a term for high ACoS on a handful of clicks. Note also a distinction worth keeping straight: deactivating a keyword already inside a campaign is the tool for an underperforming term you deliberately targeted; a negative list is for terms you never targeted that broad or auto matching is pulling in. Finally, sellers run two different negation philosophies: negate only what's genuinely irrelevant and let campaigns overlap, or negate every term targeted elsewhere so each campaign is fully segregated — the second trades flexibility for clean per-keyword attribution and bid control. "Don't overthink this. You can add these as time goes on when you look at the data."
Manual targeting splits into keyword targeting and product targeting — the latter also called PAT (Product Attribute Targeting) — where you bid on products, categories or product attributes rather than search terms (Manual Product Targeting (Category vs. Individual ASIN)). Category targeting casts a wide net over many listings and is reserved for high-volume impression chasing; individual ASIN targeting hits specific competitor or complementary products, added by search, pasted list or file upload, and is precise enough to be the default for most campaigns. Both support exact and expanded match — expanded uses your chosen ASIN as a seed and extends to products Amazon judges similar — and both support negative targeting. An account running almost entirely on auto and broad match is treated as a signal of under-invested structure regardless of spend level.
One thing the targeting choice does not do is restrict placement. Amazon shows both keyword- and product-targeted campaigns in search results and on product detail pages: "whether you do keyword targeting or product targeting Amazon can show you both in Search and on product listings and it will" (PPC Dual-Placement Visibility Effect). The targeting type is a signal of intent to the algorithm, not a placement filter — the presenter is candid that he doesn't fully understand the mechanism, treating it as observed platform behavior. One stated tactic for biasing a product-targeting campaign toward detail pages is to raise the search-placement percentage (e.g. +100%) while simultaneously lowering the base bid, on the logic that the percentage applies on top of the base; the same source recommends letting the campaign run and checking the console for where placements actually land rather than tuning this upfront.
Category targeting gets sharper through Refine (Category Targeting Refine Filters (Price, Rating, Brand, Shipping)), which narrows a selected category to a more winnable subset on up to four filters: brand, price range, review star rating, and shipping. Set a price minimum above your own listing's price so deal-seeking shoppers on pricier competitors see your cheaper ad; cap the target star rating below yours (under 4★ when you're 4.5★) so your better-reviewed offer surfaces against weaker ones. The brand filter targets a whole competitor catalog and is usually used alone, picked from brands you already convert well against per the search term report. The shipping filter over-restricts the pool in practice and is normally left alone. Critically, refine on one axis at a time — stacking price and rating shrinks the eligible pool too aggressively and reportedly hurts performance. If a refined campaign starves for impressions, loosen a filter rather than abandoning the tactic. Category targeting isn't a fallback: in the $200+ Home & Kitchen bedroom launch, category campaigns drove the bulk of a 90-day $100K run — "bed frames" at 12% ACoS on $21,700 in sales, "beds" at 18% ACoS on $6,100 — outperforming a keyword-only approach in that window. The setup was Sponsored Products → manual → product targeting → Categories tab → select category → your own calculated bid, never Amazon's suggestion → dynamic bids down only → no placement modifiers yet → daily budget.
Individual-ASIN campaigns come in four sub-strategies (Product Targeting Sub-Types (Refined, High-Volume ASIN, Missing-Feature ASIN, Self-Targeting)): refined (competitors that are lower-rated and/or priced higher than you, so your comparative position is favorable), high-volume ASIN (the biggest sellers in the niche regardless of rating or price, to intercept their demand), missing-feature ASIN (competitors lacking a feature you have, so differentiation does the selling), and self-targeting. The refined variant has a name of its own — the Switchers Tactic (Product Targeting by Competitor Star-Rating Gap) — built on the logic that a shopper already comparing options is more persuadable to defect from a visibly worse-rated listing than from a well-reviewed one. In the Essential Candy audit, that meant a new PAT campaign against roughly 100 relevant competitor ASINs, selected on worse star ratings or higher prices, layered with category-level attribute targeting on price, Prime eligibility and rating to find shoppers actively comparing. For any newly-added ASIN target, start the bid low and raise it incrementally until impressions appear rather than deriving it from the target-ACoS formula — a fresh target has no click history to feed the formula — and run that discovery in Sponsored Products only, since Display and Video auctions behave differently enough to make the signal unreliable.
The defensive mirror image is Defense / "Shielding" Campaigns (Product Targeting): advertise your own product on your other own ASINs' pages, denying competitors the slot. The apparent cost — clicks that might have converted organically — is reframed as insurance, since losing the sale to a competitor is effectively infinite ACoS on that unit. Rawlings frames it as a response to an adversarial default state ("people are doing this to you"), and is dismissive of the cannibalization objection: "Some people don't use these campaigns cuz they're afraid of cannibalization or whatever. That's totally ridiculous." These are worth running at zero sales — "We would actually pay to have these ads live, even if they produced zero sales at all, solely for the purpose of pushing out the space for competitors." Cover the surface broadly: keyword, product and category targeting, across Sponsored Products, Sponsored Brands and Sponsored Display, and across the entire catalog rather than the hero SKU alone, so no listing is an uncontested placement. For an established offline brand moving onto Amazon, this pairs with branded Sponsored Products campaigns bidding on the company's own name, captured cheaply because there's little competing bid pressure on an owned term. The concrete setup: Sponsored Products → manual → product targeting → individual products → your own other ASINs (all of them, or the top ~12 for a large catalog) → exact match only, not expanded → dynamic bids down only, no placement modifiers → daily budget.
A quirk sits inside the same mechanics. The Self-Targeting Instant Retargeting Hack (Product Targeting) enters a listing's own ASIN as its own exact product target. The ad doesn't display on that listing page — instead Amazon shows it to that same visitor afterward, in search or on other listings, functioning like built-in retargeting with no DSP setup. Rawlings presents this as exploiting "a glitch" and cites real examples at 5%, 8% and 12% ACoS despite modest volume, attributing the efficiency to the delayed mechanism. Switch the same self-target from exact to expanded and it becomes a discovery tool instead: Amazon surfaces a wide set of long-tail competitor product pages it considers similar — placements manual keyword or product research wouldn't find — with sales spread thinly across many low-click targets.
Finally, product targeting has its own exclusion lever (Negative Product/Brand Targeting): exclude specific competitor ASINs or entire brands from triggering the campaign. Add one when data shows a competitor is a stronger or cheaper offer that pulls clicks and never converts, so you stop paying to sit next to an objectively better deal. Source the candidates from the search term report and from existing auto, broad and expanded campaigns — whatever shows spend without conversions.
Everything discovery-oriented — auto, broad, expanded product targeting — exists to produce a list of proven terms, and the structure only pays off if those terms get moved. Search Term Harvesting into Exact Match is that move: pull winning search terms out of broad and auto into their own exact-match campaigns, which concentrates budget on proven converters and unlocks per-keyword placement control. The threshold is an economic judgment, not a fixed rule — three to five sales is the default for most sellers, but a brand whose lifetime value supports a high acquisition cost can justify harvesting off a single sale (see Profitability, LTV & Customer Analytics for computing that). Treat the number as a starting point and adjust based on whether harvested terms stay profitable.
Harvesting also catches terms you'd never have predicted, because A9 indexes and serves ads on observed customer search behavior rather than the listing's literal attributes. A listing can convert on "CBD cream" while being legally barred from using the phrase in its own copy — "That doesn't stop customers from searching it." That mismatch is precisely why winners have to be found in the search term report rather than reasoned out from the product spec.
The harvested terms land in one of two structures. The 'Good Keywords' Scaling Campaign is a dedicated campaign holding proven, profitable keywords pulled out of the original launch campaigns — named [product] good keywords, matching each keyword's original match type, and funded at $50–$100+ daily so de-risked spend isn't capped by a launch-sized budget. It's the "harvest the winners" counterpart to negative-keywording the losers out. The other is the Single Keyword Exact Match Campaign: one keyword, alone, so its bid and placement adjusters can be tuned without dilution. Under the cheat sheet's Ranking category it pairs with fixed bids to force the placement; in practice the build is a manual keyword ad group with only that keyword as exact match, a set bid, and spend pushed via an aggressive top-of-search placement adjustment — +128% in the documented case. Managing one day to day is described as roughly 90% about tuning that placement adjustment; if top-of-search ACoS becomes uncompetitive, shift the adjustment to whichever placement is currently performing rather than abandoning the campaign.
The metric inversion is the point. These campaigns are expected to run at a high ACoS — 65% in one example — and are judged by rank movement tracked in Helium 10 or DataDive, and by whether TACoS trends down over time, not by their own ad efficiency. In one $1M-in-12-months supplement launch, five of the top eight campaigns by sales were single-keyword exact-match ranking campaigns, reflecting that for that product organic rank was everything. The caveat is stated in the material itself: that emphasis won't hold for every launch, so the number of ranking campaigns should track how much a given product's growth actually depends on organic rank. When choosing which keyword to fund, one heuristic favors long-tail phrases of five or more words in competitive niches — lower absolute volume, but a materially higher probability of reaching top rank quickly, since fewer competitors bid those exact long phrases.
A broad-match sibling exists: the Multi-Keyword Ranking Campaign (Broad Match) loads the same core "money" keywords used in the exact-match ranking campaigns but targets them as broad, giving Amazon room to surface related high performers, with phrase negatives blocking the irrelevant queries that discretion invites. In one ten-campaign launch structure it posted the best ACoS of the entire account — and the exact-match versions of the same keywords still received the heavier ranking budget, because efficiency and organic-ranking effect are different jobs and the broad campaign doesn't do the second one.
All of this together is what Chad Scaling (Simple, Aggressive Campaign Scaling Style) describes: a small number of high-conviction campaign types — chiefly single-keyword exact-match and product targeting — bid aggressively, selected and tuned from data rather than from a fixed template. Ten campaigns, $1M in twelve months. The tuning itself, though — reading search term and placement reports, diagnosing what a live account is actually doing, day-parting, DSP and AMC — is the next chapter's material: PPC Optimization, Analytics & Advanced Targeting.