Amazon Product Launch: What The First 90 Days Should Actually Look Like

An Amazon product launch is won in three phases across roughly 90 days: buy data, buy rank, then buy profit. Here is what each phase costs, what to measure in it, and why a high ACOS in week two is a plan rather than a problem.

Strategy

An Amazon product launch works in three phases across roughly 90 days. Weeks one to three buy data, weeks four to eight buy rank, and weeks nine to thirteen buy profit. Each phase has a different job, a different budget and a different acceptable ACOS, and treating all ninety days as one long attempt to be profitable is the most common reason launches stall.

Written by Epsilon Ads  ·  18 August 2026  ·  Verified Amazon Ads Partner

We have launched products inside accounts spending $500 a month and accounts spending six figures, across 16 marketplaces and most categories. The launches that work are rarely the ones with the biggest budget. They are the ones where somebody decided, in advance, what each week of spending was for.

Why an Amazon product launch takes 90 days

Ninety days is not magic. It is roughly how long Amazon’s ranking system takes to form a settled opinion about a new listing on a competitive keyword. Sales velocity, conversion rate and relevance all feed that opinion, and none of them can be established in a fortnight because there simply is not enough data yet.

Long tail phrases move much faster. A specific four word search with modest volume can be yours inside two weeks. Head terms, the ones with the traffic everybody wants, take the full stretch and sometimes longer. That difference dictates the whole strategy: start where you can win quickly, use those wins to build velocity, and let the head terms come to you rather than attacking them on day one with a bid you cannot sustain.

If you want the mechanics of how organic position is actually earned, we covered it properly in Amazon SEO in 2026, including a real listing we took from position 15 to position 3. The launch plan below is the advertising half of that same machine.

What has to be right before you spend a penny

Advertising sends traffic. It does not sell anything. If the listing does not convert, ad spend simply buys the same failure more expensively, and we see this constantly in audits: a launch declared dead when the actual problem was the third image.

Before you switch anything on, check these. Your main image is clean, fills the frame and reads at thumbnail size on a phone. Your title carries the primary keyword naturally and describes the product in the first sixty characters. Your bullets answer the questions a buyer would ask before they ask them. Your price sits within the range of the products that currently rank for your keyword, or you have a visible reason why it does not. Your backend keywords are filled and not repeating the title. And, if you are brand registered, your A plus content exists.

Two of those deserve extra attention because they quietly wreck more launches than anything else. Most Amazon traffic is on a phone, and most listings are approved on a desktop, which is exactly the gap we wrote about in your listing is a mobile listing now. And A plus content is where brands most often lose a sale they already paid for, which we broke down in why your A plus content is not converting.

Phase 1, days 1 to 21: buy data

The job in phase one is information, not profit. You are finding out which searches lead to sales, what a click actually costs in your category, and whether the listing converts at all. Everything else waits.

Structure. Run one automatic campaign on a modest budget with default bids to discover search terms, and one manual exact campaign holding the ten to twenty keywords you are confident about. Some accounts add a broad campaign as a middle layer. That is fine, but three campaigns is plenty for one product and more than that splits your data into pieces too small to read.

Bids. Start near Amazon’s suggested bid rather than above it. Aggressive opening bids buy you the top of search on day one at a price you cannot hold, and that produces a spike that teaches you almost nothing.

Patience. Change nothing structural for the first fourteen days. This is genuinely difficult and it is the discipline that separates launches that work from launches that thrash. Amazon needs a couple of weeks to settle, and any conclusion you draw on day four will be drawn from ten clicks.

What to watch. Impressions tell you whether your bids are competitive at all. Click through rate tells you whether your main image and price are working. Conversion rate tells you whether the listing does its job once someone arrives. Ignore ACOS almost entirely in these three weeks. It will be ugly and that is expected.

Phase 2, days 22 to 60: buy rank

By now the search term report has opinions. You know which searches convert, roughly what a sale costs, and which keywords are worth fighting for. Phase two narrows the aim: take the money spread across everything and concentrate it on the terms that can actually carry the product.

Pull your search term report and split it. Terms with sales at acceptable cost get promoted into their own exact match keywords with deliberate bids. Terms with clicks and no sales past your break even click count get negatives. If that process is unfamiliar, the full method is in the Amazon search term report guide, and doing it properly in this window is worth more than any bid tweak.

This is also where placement matters. Top of search converts substantially better than rest of search or product pages in most categories, and phase two is when it is worth paying for that position on your core keywords, because sales there build the velocity that drives organic rank. Use placement modifiers deliberately rather than raising base bids across the board.

And guard your stock like it is the campaign itself, because it is. Running out mid launch does not pause progress, it reverses it: the listing loses velocity, organic position slips, and you restart from a worse place than you began. We see this wipe out six weeks of work more often than any other single mistake.

Notice what the chart does not show: a straight line down from day one. ACOS falls because organic visibility rises and starts carrying sales the ads used to pay for. That is the entire economic argument for tolerating an expensive first month, and it only pays if the sales velocity actually happens.

Phase 3, days 61 to 90: buy profit

The product is ranking on some terms, converting predictably, and the account has real data. Phase three turns a launch into a business.

Tighten the bids on keywords where you now hold organic position. This feels counterintuitive and it is the most valuable single move in the whole ninety days. If you rank first organically for a term, paying top dollar to sit above yourself buys sales you were about to get free. Reduce, watch total sales rather than ad sales for two weeks, and if total holds you have just found margin.

Add Sponsored Brands if you are brand registered, because by now you know which keywords deserve the extra real estate. Consider Sponsored Display for retargeting shoppers who viewed and did not buy. And if the product is doing volume, this is the point where Amazon DSP starts to make sense, though not before: DSP is for accounts that have already taken the search results, not for products still trying to.

Finally, set the steady state target. Not a number from a blog post, a number from your own profit and loss. Ten percent total advertising cost of sale is excellent for one brand and quietly disastrous for another, which we argued at length in there is no correct TACOS. Our own book runs a 17.84% lifetime ACOS across very different categories, and the best account in it sits at 8.68% TACOS, but neither of those is your target until you have checked it against your margins.

How to size the budget from your own numbers

Forget category rules of thumb. Work backwards in four steps.

First, look at the products currently ranking on page one for your main keyword and estimate what daily sales it takes to sit there. Review counts, best seller rank and the tools you already have will get you close enough. Second, take your listing’s conversion rate, or 10% as a placeholder if you have no history, and divide the daily sales figure by it to get the clicks you need per day. Third, multiply those clicks by the average cost per click Amazon suggests for your keywords. Fourth, multiply by 1.4 for the discovery spend in phase one that will not convert.

Step	Example	Running number
Daily sales to hold page one	8 units a day	8
Divide by conversion rate	10% conversion	80 clicks a day
Multiply by cost per click	$0.75 average	$60 a day
Add phase 1 discovery	Multiply by 1.4	$84 a day for the first three weeks

If that number is uncomfortable, the honest options are to choose a less contested keyword to launch on, improve conversion rate so each click goes further, or delay until the budget exists. Launching at a third of the required spend is not a cheaper version of the same plan. It is a slower plan that usually never reaches the point where organic rank takes over, which means you pay for the expensive early phase and never collect the payoff.

Five ways launches go wrong

Judging week one on ACOS. Week one ACOS is a measurement of how little Amazon knows about your listing. It is not a verdict.

Running out of stock in week five. The single most expensive mistake available. Order for the launch you are planning, not the sales you have.

Changing everything every three days. Bids, budgets, keywords, price, images, all at once, twice a week. When something moves you have no idea which change did it. One change at a time, then wait.

Attacking head terms on day one. Buying clicks on the most contested keyword in the category before the listing has converted anything is how a budget vanishes in eleven days.

Treating advertising as the whole launch. Reviews, price, images, stock and content all move the same needle. Ads only decide who sees the listing, not whether they buy.

When to stop, and how to decide in advance

Write your kill criteria before the first pound goes out, because in week six you will be too invested to judge fairly. A reasonable standard: 30 days elapsed and at least 300 clicks across your core keywords. At that point, if conversion rate is below roughly half your category norm, the problem is the offer or the listing and more advertising will not fix it. If conversion rate is healthy but ACOS is far above break even with no downward trend, your cost per click is too high for your margin, which is a pricing and positioning problem rather than a bidding one.

Stopping is not failure. Stopping at day 30 with a clear reason costs you a month of budget. Refusing to stop, and repeating the same launch for eight months because each individual week felt nearly there, costs considerably more. The sellers who build big catalogues are usually the ones who are quickest to call it on the products that were never going to work.

Frequently asked questions