An Amazon product launch PPC strategy is a phased plan for advertising a new product before it has any sales history. Prepare the listing and a short keyword list, go live with exact match on that shortlist plus a capped auto campaign for discovery, judge the launch on conversion rate rather than ACOS, and hand the product into normal management once its numbers clear preset benchmark gates.
In 6+ years on Amazon Ads, the launches I have seen go sideways rarely failed on bids. The listing went live before it could convert, the keyword list was the category's priciest head terms, or nobody wrote down what the launch could lose. This guide covers the four phases I run, the launch sequence, the budget math, the exit gates, and a checklist.
The four phases of an Amazon product launch
A PPC launch runs in four phases. Pre-launch readies the listing, the keyword shortlist, and the loss you will accept. Launch weeks buy relevant clicks and read conversion. Stabilization prunes and harvests once real data exists. Hand-off moves the product into routine management when its numbers clear the gates for the next benchmark phase.
The phases run on conditions, not dates. I will not promise a timeline for any of them; category competition, your offer against the listings already ranking, and review pace all move it. The logic is constant: a new ASIN earns its place search by search, through orders on the terms you target, which is why PPC and organic ranking are one conversation and conversion is the launch's scoreboard.
Before the first ad: what a launch needs in place
Three things must exist before a launch campaign spends anything: a listing that can convert, with stock to survive the push; a keyword shortlist the product can realistically win; and a written limit on what the launch may lose. Missing any one turns launch spend into noise.
A listing that can convert. Ads deliver shoppers; the detail page sells. Lock the main image, title, bullets, and price before launch, and carry enough stock that a stockout cannot halt the orders the spend exists to create. If shoppers click and leave, no campaign setting fixes it, which my guide to PPC clicks with no sales explains.
A shortlist you can win. Group the terms that describe the product into families by root word and pick the one family it can compete in today. That is rarely the biggest head term: where page one belongs to listings with years of reviews, shoppers pick the established option, and your lost orders read as weak conversion. Look for a longer phrase with real demand and softer incumbents. Hold broad category searches back too, since browsing shoppers drag conversion down. My keyword research guide builds the full list.
A loss budget in writing. Decide what the launch may spend above your eventual target, and for how long. Without that number, someone panics at week one's ACOS and cuts spend before any verdict exists, or nobody ever declares the launch over. The worked example below sizes it.
How to launch a product on Amazon with PPC, step by step
Launch day runs a deliberately small structure: exact match on the shortlist, a few capped rank-push campaigns on its best terms, one auto campaign on a discovery budget with down-only bids, and brand terms walled off. Everything else waits until the first search term data shows it has earned a place.
- Build the structure as one bulk file. Write every campaign before go-live. The free Campaign Builder turns a SKU and a shortlist into an upload-ready bulk file, negatives included, with a launch plan that splits budget by how hard you want to push.
- Estimate each bid's ceiling, then open below it. With no history, the ceiling is price times a conservative conversion rate (the 3% awareness floor works) times target ACOS; how to set Amazon PPC bids has the method.
- Go live with exact match on the shortlist. One product per ad group, placement adjustments at 0, and dynamic bids set to down only. These campaigns carry the relevance-building orders.
- Add rank-push campaigns for the best few terms. One keyword per campaign, a deliberate Top of Search premium, a hard daily budget cap, and a written end condition. Placement adjustments only raise bids, up to 900%, so step into the premium.
- Run one auto campaign on a capped budget. All four targeting groups on, each bid separately, dynamic bids down only. Its matches show whether Amazon understands the listing.
- Wall off brand terms. Negate your brand name in every non-brand campaign, so stray branded searches cannot flatter the launch's numbers.
- Read conversion weekly, one change at a time. Judge each shortlist term once its spend passes the fair-judgment threshold, and change one variable per cycle so movement has a traceable cause.
- Start the harvest loop as trials finish. Promote converting search terms to exact match, negate them where they were found, and negate proven losers. Stabilization starts here.
The Amazon launch campaign structure
A launch structure is the mature per-product stack with its middle held back. Auto and exact match go live on day one, single keyword rank-push campaigns sit on top for a few chosen terms, and broad and phrase research, product targeting, and brand defense join as the data and the branded searches arrive.
Some launch playbooks keep auto off for the first month, fearing stray early matches teach Amazon the wrong associations. I run it anyway, capped and on down-only bids, because if close match keeps landing on searches for a different product, the listing is describing itself badly, and week one is the cheapest time to learn that. Broad research and competitor ASIN targeting wait for evidence; my campaign structure guide maps the full stack they join, and the auto campaigns guide covers the 40% auto-share ceiling for this phase.
Why conversion rate, not ACOS, grades a launch
A launch's ACOS is set mostly by its conversion rate, not its bids. At the same click price, a listing converting at the awareness-phase floor of 3% spends roughly three times as much per order as one converting at 9.1%. So launch weeks are graded on conversion, and early ACOS is read as a symptom of it.
Take the sample account I use throughout this hub: $36,303 of spend, 45,672 clicks, 4,140 orders, and $111,058 in ad sales, or 9.1% conversion at a $0.79 average CPC. Hold the click price and average order steady and change only conversion:
| Conversion rate | Clicks per order | Ad spend per order (fair-judgment threshold) | Revenue per click | ACOS at $0.79 CPC | Highest CPC for a 40% ACOS |
|---|---|---|---|---|---|
| 3% (awareness floor) | 33 | $26.33 | $0.80 | about 98% | $0.32 |
| 6% (market share floor) | 17 | $13.17 | $1.61 | about 49% | $0.64 |
| 9.1% (sample account, actual) | 11 | $8.77 | $2.43 | 32.7% | $0.97 |
At the 3% floor, one order costs $26.33 in ad spend, nearly the whole average order, so ACOS lands near 98% with no bid set wrong. At 6%, each order costs half that; the fastest ACOS fix in a launch usually lives on the detail page. And the fair-judgment threshold, the price of one fair trial for a search term, triples at the floor.
That is how I size a loss budget. My harvest bar is two orders, so giving one shortlist term a real chance to clear it at 3% takes about 67 clicks, roughly $53 at $0.79, and ten terms need about $530 just to finish those trials. Your figures will differ; the method holds. Terms times trial cost at a realistic launch conversion rate, plus the rank-push caps, is the floor under the budget.
Stabilizing a launch: what changes once real data arrives
Stabilization starts when shortlist terms have finished their fair trials. The work shifts from buying evidence to acting on it: harvest terms that converted into exact match, negate them at the source, park shortlist terms that keep losing, add broad and phrase research, and start earning placement premiums with data instead of guessing them.
At launch, nearly every targeting row sits in the Low Visibility bucket of my four-bucket model: unproven. Stabilization sorts each row into one of the other three on evidence. Low ACOS terms get protected. High ACOS converters get repriced, not paused. High Spend, Non-Converting terms past their trial become negatives; my free Wasted Spend Finder lists them worst first, and the Negative Keyword Finder formats them for upload. And each term harvested into exact gets a negative exact at its source the same day, or a new account manufactures its own search term bleed.
Shortlist terms that lose get a different verdict: parked, not negated. A term you chose on purpose usually fails for a fixable reason: a price gap against the listings above it, a main image that loses the side-by-side, or a search that means something slightly different. Fix it, then bring the term back. And once the first keyword family holds, open the next rather than spreading budget across all of them at once.
When is a product launch over? The hand-off gates
A launch ends when the product's numbers clear gates, not when a calendar says so. I call it stable when conversion holds above the awareness floor and trends toward the next phase's bar, the rank-push campaigns have hit their end conditions, auto share sits under its 40% ceiling, and the harvest loop is running on its own data.
| Metric | Launch bar (Awareness) | Next bar (Market Share) | What I want to see at hand-off |
|---|---|---|---|
| Conversion rate | > 3% | > 6% | Above the floor and climbing; weighted most |
| Account ACOS | < 40% | < 30% | Falling as conversion rises |
| Top-of-Search ACOS | ≤ 60% | ≤ 50% | Rank-push premiums walked back |
| Wasted spend % | < 50–60% | < 35–45% | Shrinking as negatives land |
| Auto campaign % | ≤ 40% | ≤ 25% | Falling as harvests graduate terms |
| Branded sales % | ≤ 5% | ≤ 25% | Brand campaign live once branded searches appear |
Two more signals set the timing. TACOS runs high during the push, since nearly every sale is ad-driven, and if the launch is working it drifts down as organic orders grow; my ACOS vs TACOS guide shows how to read that trend. And rank-push premiums come off when their end conditions hit, because a leftover launch Top of Search modifier is one of the costliest leaks I find in audits.
Then the product moves to the next benchmark column and gets managed like everything else. When I run a launch for a client, it carries a $499 flat, one-time fee per new product line, meaning a new parent ASIN; a new color or size of a product I already manage does not count. Once the gates clear, it folds into the monthly fee, which is never a percentage of ad spend.
The Amazon product launch checklist
This is the checklist I work through on every launch, in order. Each row has a done-when condition, because an undefined checkbox gets ticked early. Anything unticked in the pre-launch rows is a reason to delay the ads, not to launch and patch it later.
| Phase | Checkpoint | Done when |
|---|---|---|
| Pre-launch | Listing and stock | Images, title, bullets, and price are final; inventory covers the push plus reorder time |
| Pre-launch | Shortlist | One keyword family chosen; broad category terms held back |
| Pre-launch | Loss budget | Written down: trial costs, rank-push caps, and how long they run |
| Launch weeks | Structure live | Exact shortlist, capped rank-push campaigns, capped auto, brand negatives |
| Stabilize | Harvest loop | Winners in exact and negated at source; losers negated; losing shortlist terms parked with a named cause |
| Hand-off | Gates cleared | Conversion above 3% and climbing, auto share under 40%, rank-push premiums removed, TACOS trending down |
Frequently asked questions
How much should I spend on PPC to launch a new Amazon product?
There is no honest universal figure, so size it from arithmetic. Count your shortlist terms, price each one's trial at a realistic launch conversion rate (clicks per order times CPC, doubled if you want two orders before judging), and add capped budgets for the rank-push campaigns. Decide how long you will fund it before the first ad runs.
Should I use auto campaigns when launching a new product on Amazon?
Yes, but capped. The worry that early auto matches teach Amazon the wrong keywords deserves respect, so in my launches auto runs on a small discovery budget with down-only bids while exact match on the shortlist carries the push. Its matches double as a listing check: strange matches mean the listing needs work, not the bids.
What ACOS should I expect during an Amazon product launch?
Higher than your long-run target, driven mostly by conversion rate. My awareness-phase benchmarks allow under 40% account ACOS and up to 60% at Top of Search, but a listing converting near 3% can run far above that with no bid set wrong. If conversion climbs, ACOS follows it down; if it stalls, check the listing and price first.
Is the Amazon honeymoon period real?
Sellers use the phrase for an early visibility boost Amazon supposedly gives new listings. Amazon does not document one, so I never build a launch that depends on it. I front-load the first weeks anyway, because a product with no history is judged on the little it has, and early orders on the right searches are most of that.
When should I start running PPC on a new Amazon listing?
When the listing can convert, not the minute it goes live: final images and copy, a price you intend to hold, inventory that covers the push, and campaigns already built. Ads on an unfinished listing buy clicks that do not convert, and those become the first data Amazon collects on the product. Thin reviews at launch are normal; an unfinished page is not.
Sources
- Guide to dynamic bidding with Sponsored Products, Amazon Ads
- Best practices for your Sponsored Products ads, Amazon Ads
- A guide to targeting with Sponsored Products, Amazon Ads
Planning a launch? Check the account it lands in
A new product inherits the account it launches into, so if the current campaigns leak brand traffic or keep paying for search terms that never convert, the launch leaks too. The free Account Health Snapshot grades eight metrics from a 60-day bulk file against the benchmarks for your phase, with no email, no account, and nothing stored. Then bring the product and your shortlist to the free 30-minute diagnosis call, and I will tell you what I would build first, whether or not you hire me.