The Amazon product launch playbook for sellers

Before day one: the checklist that decides the launch

Launches are lost in the fortnight before they start, usually because something on this list was left until "after we go live". Work through it in order.

Stock, and the arithmetic behind it. Nothing kills a launch faster than going out of stock in week three, because the sales velocity you paid for evaporates and the rank goes with it. Plan inventory against your ambition, not your forecast: if you intend to sell thirty units a day by week six, you need cover for that plus lead time, and running out is more expensive than overstocking by a wide margin.

Brand Registry, finished. Without it you cannot run Sponsored Brands, cannot use A+ content, cannot see Brand Analytics, and have no protection against hijackers. All four matter during a launch more than at any other time. If it is not done, delay the launch rather than launch without it.

The listing, written for the 2026 fields. This is the step most published playbooks have not updated. Amazon announced that "Starting July 27, 2026, titles in all categories except for media will need to be 75 characters or less including spaces", with a new 125-character Item Highlights field that began displaying beneath the title on 10 August. Amazon states both are search inputs and neither is prioritised.

So a new listing has 75 characters for brand plus the one or two terms that say what the product fundamentally is, and 125 for the attributes a buyer filters on. If you write the title the old way it will be rejected or rewritten for you — Amazon gave brand owners 14 days to review AI-generated recommendations and updated non-compliant titles gradually afterwards.

Photography, treated as the main budget line. The main image decides click-through rate on a results page, which decides what every advertising dollar buys. Seven images minimum: hero on white, in-use, scale reference, feature callouts, what is in the box, a comparison against the obvious alternative, and one addressing the most common objection in your category’s reviews.

A keyword set, tiered. Fifteen to twenty terms per ASIN, sorted by relevance before volume, with the placement decided in advance. Our keyword research guide covers the method; the important part for a launch is that the list is short enough to act on.

Price, set against the page rather than the spreadsheet. Look at what the page-one results charge and what review counts they carry. A new listing at the top of the price band with no reviews is asking shoppers to take a risk they have no reason to take.

The honeymoon period, and what it really is

New listings frequently see an early surge of impressions that is hard to explain from sales alone. Sellers call it the honeymoon period. Amazon has never confirmed it exists, publishes no duration, and anyone quoting you an exact window is repeating folklore.

What is observable, and what actually matters, is simpler: a new listing has no conversion history, so Amazon has to test it to learn anything. That testing produces impressions. What you do with them decides whether the listing keeps them.

Which reframes the whole question. The point of the first few weeks is not to exploit a bonus before it expires. It is that you are being assessed on a small sample, and a listing that converts poorly during that sample takes longer to recover than one that never had the impressions in the first place.

Three consequences worth planning around:

• Do not launch before the listing is finished. Traffic against a half-built page spends your one cheap sample proving you do not convert.
• Do not launch into an out-of-stock risk. The assessment does not pause while you restock.
• Do not launch on a Friday. Something will need fixing in the first 48 hours, and Seller Support is slower at the weekend.

Treat the early window as an audition rather than a gift and you will make better decisions than any theory about its length would produce.

The corollary is worth stating plainly, because it is the most expensive mistake in this article: a soft launch is usually a wasted launch. Going live quietly to "see how it does" before the photography is finished, before the keyword set is placed, or before stock has landed in full, spends the one period when Amazon is actively looking at you on a version of the listing you would not choose to be judged on. If you are not ready to support it with budget and attention, leave it unpublished.

Days 1–30: buy data, expect to lose money

The first month has one job: find out which keywords convert for this product. Sales are the mechanism, not the target.

Campaign structure, day one. Three campaigns and no more:

• Auto, modest budget, running permanently. This is a research instrument that happens to make sales. It will find phrasing you did not think of.
• Exact match, your fifteen to twenty researched terms. One ad group, generous bids. You are buying placement to generate conversion history on terms you intend to own.
• Product targeting against three to five competitor ASINs. Specifically the ones a shopper would buy instead of you. Expensive, and it puts you in front of exactly the right person.

Resist adding broad match, Sponsored Brands, Sponsored Display or anything else in month one. You cannot read the results of five variables changed at once, and the simplicity is the point.

Bid higher than feels comfortable. A launch bid is not a profit calculation, it is a purchase of placement. Our own working range at this stage is 50–80% ACoS, which is deliberately unprofitable, and the discipline is that it has an end date rather than that it is low.

Harvest weekly, without exception. Every week: pull the search term report, move converting terms into their own exact-match ad groups, add non-converting spenders as negatives. Four cycles of this during month one is most of the value the month produces.

Reviews, inside the rules. Enrol in Vine if you are eligible — it is Amazon’s own programme and the only sanctioned way to get early reviews. Turn on Request a Review automation. Put a well-designed product insert in the box that asks for honest feedback without conditioning anything on a rating.

What not to do, at any price: buying reviews, incentivising positive reviews, review gating, or giveaway schemes designed to manufacture rank. Amazon suspends the seller for these, not the supplier, and in the US the Federal Trade Commission’s rule on consumer reviews makes buying and selling them a civil-penalty matter. Our review work is built entirely inside the permitted mechanisms for this reason.

Days 31–90: concentrate, then consolidate

The second and third months are where a launch either compounds or stalls, and the difference is almost always whether anyone acted on month one’s data.

Days 31–60: concentrate

You now have four weeks of real conversion data. Use it to make the budget narrower rather than larger.

Rank your terms by conversion rate, not by volume. The top five to eight get most of the budget. Terms that took clicks and never converted get paused or negated, however good they looked in research.

Promote the winners into the listing. Terms that converted consistently for four weeks have earned a place in the title, Item Highlights or bullets. This is the moment advertising data becomes SEO work, and it is the step most launches skip entirely.

Add Sponsored Brands if you have two or more products worth showing together, and defend your own brand term, which will be cheap and convert well.

Start reading organic rank on your main terms weekly. This is the number that tells you whether the money did anything durable.

Days 61–90: consolidate

Reduce bids on terms where organic rank has arrived — one at a time, watching position for a fortnight after each. Reduce rather than pause: total sales velocity on a term includes the paid sales, so switching off abruptly can cost the organic position you just bought.

Watch TACoS rather than ACoS. A falling total advertising cost of sale at steady spend means organic revenue is growing underneath the advertising, which is the entire point of the ninety days. There is more on the distinction in our piece on ACoS and TACoS benchmarks.

Decide what this product is. By day 90 you should know whether it is a winner worth further investment, a steady contributor to run at target ACoS, or a mistake to clear and stop funding. Making that call on schedule is worth more than any tactic in this article.

External traffic, and whether it is worth it

Sending traffic to Amazon from somewhere else — an email list, paid social, a creator, your own site — comes up in every launch conversation, and the honest answer is that it is useful in a narrower set of cases than it is recommended in.

Use Amazon Attribution or do not bother. It is free, it gives every external source a tagged link, and without it you are sending traffic you cannot measure and will never know whether it worked. An untagged Instagram campaign that coincides with a launch is indistinguishable from the launch.

The quality problem is real. Amazon shoppers arrive ready to buy. Someone clicking from a social feed is browsing, and browsing traffic converts worse. Because conversion rate on a term feeds ranking, a burst of poorly-converting external traffic can be actively unhelpful during exactly the weeks you are being assessed.

Where it genuinely helps. An existing email list of people who already want the product, because they convert like Amazon traffic rather than like social traffic. A creator whose audience matches the product closely. And a category so thin that Amazon’s own search volume cannot produce the velocity you need, where you have no alternative.

Where it usually does not. Cold paid social to a brand nobody knows, on a product that competes on price, during a launch that also needs its advertising budget. That combination spends money in two places to achieve what one would have done better.

Send them to the right place

If you do run external traffic, send it to a search results page for your brand plus product rather than straight to the detail page. Amazon reads a search-then-purchase sequence as demand for those terms, which is worth more than the same sale arriving cold. And keep the landing consistent with the ad — external traffic that bounces immediately is worse than no external traffic at all.

The four numbers to watch daily

Launch dashboards fail by showing forty metrics nobody reads. Four numbers, checked each morning for the first month, catch almost everything that goes wrong.

The pairs matter more than the individual figures. Units flat and ACoS rising means you are paying more for the same sales, which is usually a conversion problem rather than a bidding one. Units rising and organic rank static means you are renting the position and will keep renting it. Conversion falling with everything else steady usually means a competitor changed price or image, or your reviews took a hit.

Why launches stall after a strong start

Four causes, in order of how often we see them.

Stock ran out. The most common and the least recoverable. Velocity stops, rank decays, and restarting costs more than the original launch.

The budget was cut too early. Someone saw a 70% ACoS in week two, panicked, and halved the bids. The launch then fails to reach the velocity it needed, which is read as the launch failing.

Nobody promoted the winning terms. Month one produced the data and nothing was done with it, so the listing never earned the organic position the advertising was buying.

Reviews did not arrive. Conversion plateaus around the point where shoppers start comparing review counts, and a listing with four reviews against competitors with four hundred converts worse every week as the traffic gets less forgiving.

What a launch costs

No honest answer is a single number, because it scales with your category’s cost per click and your price point. But the shape is consistent, and thinking in shape rather than total is more useful.

Work backwards from units. Decide how many units a day you need to reach the organic position you want, then work out how many clicks that takes at your conversion rate, then what those clicks cost at your category’s going rate. That is your launch budget, and it will be a larger number than you expected in a competitive category and a smaller one than you feared in a thin one.

Budget the loss, not the spend. If break-even ACoS is 40% and you plan to run at 70% for thirty days, you are choosing to lose roughly 30% of advertised revenue. Write that figure down before you start. It converts an anxious daily decision into a plan you already agreed to.

Do not forget the non-advertising line. Photography, copy, A+ modules and Vine units are real launch costs, and they are the ones that make every advertising dollar go further. A launch that spends everything on clicks and nothing on the page is the most expensive way to run one.

Keep a reserve for month two. Launches that spend their whole budget in the first thirty days cannot act on what they learned, which wastes the thirty days. A reasonable split is half the advertising budget in month one, a third in month two, and the remainder held for whichever terms turn out to be worth defending.

Account for the fee floor. Amazon raised FBA fees by an average of $0.08 per unit from 15 January 2026, and referral fees range from 5% to 45% by category with a $0.30 minimum. On a thin-margin product that is the difference between a launch that reaches break-even in month four and one that never does, so calculate with your own category rate rather than a rule of thumb.

Launching in more than one marketplace

Do not run a second marketplace at the same time as the first. Each one has its own search behaviour, competitive set, fee structure and, increasingly, its own listing requirements. Land one, learn it, then take the proven keyword set and re-research it locally rather than translating it — US and UK English diverge in ways that cost real money.

If you would like a second read on a launch before you commit the budget, our launch process is published in full, and a free written audit comes back within 48 hours with the listing and keyword gaps named before you spend anything on clicks.

Questions people also ask