Amazon PPC Guide 2025

Every Amazon PPC guide worth reading starts with the same admission: the mechanics are simple, and almost everyone still gets them wrong. You bid on keywords, shoppers click, you pay. The failure happens in the layer above the mechanics — auction dynamics that punish irrelevant ads, account structures that hide winners inside averages, match types deployed without a job description, and budgets that never shift from launch mode to profit mode. This guide covers the full system: how the auction actually prices your clicks, what each campaign type is for, how to structure an account so the data stays readable, when each match type earns its place, negative keyword strategy, the launch-versus-profitability transition, and the bidding basics that hold it together. It’s the foundation for everything else in our Amazon PPC management section.

How the Amazon PPC Auction Actually Works

Amazon runs a second-price auction, modified by relevance. When a shopper searches, every ad eligible for that term enters an instant auction, and two factors decide who appears and what they pay: the bid, and the ad’s expected performance.

You pay just above the next bidder, not your bid. If you bid $2.00 and the advertiser below you bid $1.40, you pay roughly $1.41. This matters strategically: your bid is a ceiling, not a price. Bidding what a click is genuinely worth to you doesn’t mean you’ll pay it — it means you’ll win the auctions where the click is available for less than its worth.

Relevance is a multiplier on your money. Amazon’s ranking isn’t pure bid order. Expected click-through rate and conversion rate weight the auction, because Amazon monetizes the whole search page and a high-bid ad nobody clicks earns Amazon nothing. Practically: a listing with strong CTR and conversion history can win placements over higher bidders, and pushing a weak listing with aggressive bids gets progressively more expensive as the algorithm learns shoppers don’t want it. This is why traffic can’t fix a conversion problem — if clicks are landing and orders aren’t, work through the PPC not converting diagnostic before touching a single bid.

Placement multiplies everything. Top-of-search converts two to three times better than rest-of-search in most categories, and Amazon lets you bid separately for it via placement adjustments (0% to 900% on top of your base bid). Two products with identical keywords and bids can have completely different economics depending on where their impressions land. Always read performance by placement before judging a keyword.

Campaign Types Compared

Amazon offers three self-service ad types plus DSP. Each has a distinct job, and most structural problems start with one type being asked to do another’s work.

Sponsored Products is the workhorse — highest conversion rates, direct organic rank impact, and the format every account should master first. Sponsored Brands and Sponsored Display add the brand layer once Brand Registry is in place, with Sponsored Brands Video frequently posting the best cost-per-conversion in mature accounts. Amazon DSP is programmatic scale for brands already profitable on search, not a starting point.

The rest of this guide focuses on Sponsored Products, because its structure and optimization patterns are the template the other formats borrow.

Account Structure: Single-Keyword vs Themed Campaigns

Account structure is a data-readability decision. Every structure is a trade-off between control and manageability, and the right answer depends on catalog size and spend.

Single-keyword campaigns (SKAGs, or their Amazon cousin, single-keyword exact campaigns) give each major keyword its own campaign: its own budget, its own placement multipliers, its own daypart schedule. Maximum control. The cost is sprawl — a 40-product catalog with 20 keywords each is 800 campaigns, and nobody maintains 800 campaigns well. Reserve this treatment for the keywords that deserve it: terms driving meaningful daily spend (a common threshold is $10+/day) or terms where you’re deliberately fighting for top-of-search on a rank target.

Themed campaigns group 5 to 15 tightly related keywords per campaign or ad group — “garlic press” with “garlic crusher” and “garlic mincer,” not with “kitchen gadget gift.” Bids stay per-keyword, but budget, placement modifiers, and dayparting are shared. This is the right default for most of the account. The discipline that makes it work is tight theming: if two keywords in a campaign deserve different placement strategies, they belong in different campaigns.

The hybrid most well-run accounts converge on:

• One auto campaign per product (or per variation family), split by targeting group if spend justifies it — research layer.
• Themed broad/phrase campaigns — validation layer.
• Themed exact campaigns for proven terms — scaling layer.
• Single-keyword exact campaigns for the handful of hero terms — precision layer.
• Product targeting campaigns, separated into competitor conquest and self-defense.

Two structural rules regardless of approach. First, never mix products with different margins or price points in one campaign; budget flows to whatever spends fastest, not whatever earns most. Second, name campaigns so a stranger can read the account: something like SP | Exact | GarlicPress | Hero beats Campaign - 04/11/2025 #3 every week of the engagement.

Match Types and When Each Earns Its Place

Four targeting modes, four jobs. The failure pattern is using one for everything.

Auto targeting is Amazon’s algorithm choosing your placements across four groups: close match, loose match, substitutes, and complements. Its job is research — it surfaces search terms you’d never have guessed and it reaches product-page placements keyword targeting can’t. Run it at modest bids (often 20% to 30% below your manual bids so it doesn’t cannibalize), split the four targeting groups where spend allows, and harvest it weekly. An auto campaign is never “done”; it’s a permanent scouting operation.

Broad match casts wide within keyword targeting — your term plus synonyms, related searches, and reordered variants. Its job is validated discovery: cheaper traffic than exact, more directed than auto. Broad is where keyword research hypotheses go to be tested with real money. Expect mediocre blended ACoS and judge it on what it discovers, not what it earns.

Phrase match requires your term in sequence, with words allowed before or after. It’s the middle child, and its best use is targeted: when broad is matching too loosely (your “steel water bottle” broad matching “water filter”) but you’re not ready to lock into exact, phrase narrows the funnel without closing it.

Exact match serves on your precise term and close variants only. This is where proven winners live — precise bids, aggressive placement multipliers, protected budgets. Exact should carry the majority of your keyword-targeted spend in a mature account, because it’s the only match type where you know exactly what you’re buying.

The system connecting them: auto and broad discover, phrase narrows, exact scales. A search term’s lifecycle runs discovery → 2-3 conversions → graduation to exact → negation at the source. Skip the graduation step and winners stay trapped at discovery bids; skip the negation step and your campaigns bid against each other.

Negative Keyword Strategy

Negatives are where PPC money is saved, and they’re chronically neglected because adding them produces no visible activity — no new impressions, no new sales, just spend that quietly stops leaking. In accounts without recent negative management, it’s routine to find 20% to 35% of spend on search terms that have never converted.

Negative exact blocks one specific term. Use it for individual proven losers: 15 to 20+ clicks, zero orders, no plausible relevance excuse. (Scale that click threshold to your price point — a $200 product with a 2% conversion rate needs more clicks to judge than a $15 impulse buy.)

Negative phrase blocks any search containing the phrase. Use it for wrong themes: you sell stainless bottles, negative-phrase “plastic”; you sell adult apparel, negative-phrase “kids.” One negative phrase can pre-empt hundreds of junk terms — which is also its danger. Before adding, search your term reports for the phrase and confirm nothing converting contains it.

The weekly routine, which takes 30 to 60 minutes per product line:

1. Pull the search term report, sort by spend, zero orders first.
2. Negative-exact the individual losers past your click threshold.
3. Scan for repeating irrelevant themes; negative-phrase them after checking for collateral damage.
4. Negate every newly harvested term at its discovery source, so graduated keywords don’t get double-bid.

One caution: don’t negate slow burners. A term with 8 clicks and no orders on a high-priced product isn’t a loser yet — it’s undecided. Premature negation is rarer than negative neglect, but it does throw away real winners.

Launch Phase vs Profitability Phase

The single most expensive confusion in Amazon PPC is running launch tactics with profitability expectations, or vice versa. These are different operating modes with different budgets, bids, and success metrics, and you should always know which one each product is in.

Launch phase (roughly the first 60 to 90 days): the goal is sales velocity on target keywords, because velocity drives organic rank, and organic rank is the asset you’re actually buying. That means bidding at or above break-even ACoS on your 5 to 15 target terms, prioritizing top-of-search placements, and tolerating ACoS numbers that would be alarming in steady state — 50% to 80%+ is common and, within a plan, correct. Judge launch PPC on rank movement and review accumulation, not on ACoS. Pair it with Vine enrollment and a conversion-ready listing, because launch traffic against a weak page is the most expensive mistake in the playbook. Our new product launch case study shows this arc: page one in 60 days, with ad efficiency arriving after rank, not before.

Profitability phase: once organic rank stabilizes on your core terms, the objective flips. Bids come down toward break-even on terms where you hold organic position (you’ll capture much of that demand organically anyway), spend concentrates into exact match winners, discovery budgets shrink to maintenance levels, and TACoS becomes the steering metric. The distinction between ad efficiency and business efficiency is the subject of our ACoS vs TACoS breakdown, and it’s the lens that makes phase transitions legible.

The transition isn’t a date on the calendar — it’s triggered by data: organic rank holding in the top 5 to 10 on your hero terms for 2+ weeks, review count competitive with page-one peers, and conversion rate at or above category norm. Products can also move backward: a stockout that craters your rank puts you back in launch mode whether you like it or not.

Bid Optimization Basics

Bidding has one governing number: your break-even ACoS, which is simply your pre-ad profit margin. Sell at $25 with $7.50 of margin after COGS and fees, and 30% is the ACoS where a keyword neither makes nor loses money. Every bid decision is that number plus a strategic adjustment for phase and rank goals.

The starter framework:

• Set initial bids from value, not suggestions. Target CPC ≈ price × conversion rate × target ACoS. A $25 product converting at 10% with a 30% target supports about $0.75 per click. Amazon’s suggested bids are an auction median, not advice about your margins.
• Wait for significance. 10 to 15 clicks minimum before cutting a bid; more before scaling one up. Reacting to 4 clicks is astrology.
• Move in steps. 10% to 15% adjustments weekly. Large swings reset performance patterns and make cause-and-effect unreadable.
• Check placement before bids. A keyword that’s unprofitable in aggregate is often profitable at top-of-search and terrible in rest-of-search. That’s a placement multiplier fix — raise the top-of-search adjustment, lower the base bid — not a keyword problem.
• Respect Amazon’s bidding strategies. “Dynamic bids — down only” is the sane default; it lets Amazon cut your bid in low-conversion auctions. “Up and down” hands Amazon permission to double your top-of-search bid and should be reserved for campaigns whose economics you’ve already proven.

Past the basics, the leverage is in cadence and automation rules — when tools like Scale Insights or Pacvue genuinely help, how to bid against dayparts, and how rank targets change the math. The full treatment is in our bid optimization guide, with the time-of-day layer covered under PPC dayparting.

Common Structural Mistakes

After enough audits, the same failures repeat. Check your account against this list before spending another month’s budget:

1. Everything in auto campaigns, forever. Auto is research. If auto carries most of your spend six months in, you’re paying discovery prices for traffic you should own at exact match precision.
2. One campaign, forty products. Budget flows to whichever ASIN spends fastest. Your best-margin product starves while a low-margin fast-spender eats the budget.
3. Harvesting without negating. Terms graduate to exact but keep serving in broad and auto. The campaigns bid against each other and your reports become unreadable.
4. No negative keyword routine at all. The most common finding in every audit, and the fastest fix — usually worth 15% to 30% of spend in the first month.
5. Judging every campaign on the same ACoS target. Discovery, defense, and capture campaigns have different jobs. Averaging them into one blended number hides both the winners and the leaks.
6. Budgets that cap out by noon. A profitable campaign hitting its daily budget at 11 AM is a standing order to leave money on the table. Fund winners first.
7. Set-and-forget placement modifiers. Placement performance drifts with competition. Multipliers set in January are wrong by June.
8. Traffic thrown at a listing that doesn’t convert. If your conversion rate is well below category norms, PPC amplifies the loss. Fix the page first — that’s a listing optimization problem wearing a PPC costume.

Putting This Amazon PPC Guide Into Practice

Everything in this guide is executable by one disciplined person with a spreadsheet and a weekly calendar block: run the harvest-negate-adjust loop, keep launch and profitability modes distinct, and let exact match accumulate the winners. The rest of the Amazon PPC management section goes deeper on each layer as your account grows into it. And if your spend has scaled past the point where the weekly loop actually gets done — or your ACoS has sat above break-even for two months without an explanation — a structured audit is the fastest way to find out where the money is going. That’s exactly what our Amazon PPC advertising service starts with: 60 to 90 days of search term data, torn down into wasted spend, structural gaps, and a rebuild plan with specific numbers attached.