The Final Three Weeks: an Amazon Black Friday Strategy Run on Margin, Not ROAS

Three weeks before the event, an Amazon Black Friday strategy is no longer about building demand. Deals are locked, creatives are in place and most of the inventory has shipped. What remains is arbitration: which products deserve budget given their stock, which spend creates new customers, and how much margin you are willing to trade. Brands that steer this period on ROAS usually end Cyber Week with strong numbers and thin profit.

  • At this stage, the levers left are budget allocation, bids, stock coverage and listings, not new campaigns.
  • Stock decides which products should be advertised during Cyber Week before bids do.
  • ROAS rises mechanically during the event, so TACoS and new-to-brand are better guides.
  • Branded and non-branded budgets need separate rules before the auction heats up.
  • The decisions that matter most are taken in the week after Cyber Monday.

Three weeks out, an Amazon Black Friday strategy is about arbitration, not new campaigns

By early November, the structural work is done or it is too late. Event deals and coupons are usually set by now, and the inventory that will be available on November 27 is largely known. Launching a new campaign architecture now adds risk without time to learn.

What can still move, and should, comes down to four levers:

  • Budget allocation between hero products, secondary products and new launches.
  • Bids by time of day and placement, since Cyber Week traffic is concentrated on a few days and hours.
  • Stock coverage, product by product, against the expected sales of the event.
  • Listings: titles, images and deal badges that convert the extra traffic.

Amazon opened Full-Funnel Campaigns to all US advertisers at unBoxed in September 2026. Our position is pragmatic: they are worth testing on secondary products, where automation saves time. On hero products, keep manual control during the event, because that is where branded traffic and margin are decided. This builds on the demand work described in Black Friday doesn’t start on Black Friday.

Stock decides your Cyber Week before your bids do

Advertising a product that will be out of stock before Cyber Monday is the most expensive mistake of the season. The ads keep driving traffic until the last unit sells, then the product loses its sales momentum at the worst moment, and part of the demand goes to competitors.

The check is simple: divide available units by the daily sales you expect during the event to get days of cover, then compare with the length of the period you need to cover.

Illustrative example: a product has 1,200 units available. You expect 150 sales per day during Cyber Week, which gives 8 days of cover. The period from November 21 to December 1 lasts 11 days. Without action, the product runs out of stock around three days before the end of the event.

Days of cover vs event length Advertising decision
Cover longer than the event Push: this product can absorb extra traffic
Cover close to the event length Hold bids, protect the top position on your own brand terms only
Cover shorter than the event Reduce bids to stretch stock to Cyber Monday, and move budget to a product with cover

Advertising, stock, price and content work as one system. Bids that ignore stock optimize in a vacuum.

Judge Black Friday ads on TACoS and new-to-brand, not on ROAS

ROAS rises during Cyber Week for reasons that have little to do with your campaigns: conversion rates jump, branded searches multiply and retargeting audiences are full of shoppers who were going to buy anyway. A high ROAS in this period mostly measures demand you already had.

Two metrics give a truer picture:

  • TACoS (total advertising cost of sale): ad spend divided by total sales, ads and organic. It shows whether advertising grows the business or only takes credit for it.
  • New-to-brand orders: the share of ad-attributed orders from customers who had not bought your brand on Amazon in the past year.

Illustrative example: a branded campaign shows a ROAS of 8, but only 10% of its orders are new-to-brand. A non-branded campaign shows a ROAS of 3, with 60% new-to-brand orders. On 1,000 orders each, the first brings 100 new customers and the second brings 600. If your products are bought again, the second campaign is the one that builds next year’s sales. We detail this logic in our article on new-to-brand on Amazon.

The practical rule: set a TACoS ceiling for the event, and accept a higher one on products where the value of a new customer justifies it.

Split branded and non-branded budgets before the auction heats up

During Cyber Week, competitors target your brand terms on Amazon too, and your own Sponsored Products and Sponsored Brands ads on branded queries can end up taking sales your organic listings would have made. Without separate rules, branded spend absorbs the budget because it shows the best ROAS.

  • Branded terms: protect the top position when competitors appear, without overpaying when you already dominate page 1.
  • Non-branded and category terms: this is where new customers come from, and where the budget should go once branded traffic is secured.
  • Competitor terms: reserve them for products with strong stock, a competitive price and a visible deal.

At Primelis, our Primelis Market platform recalculates bids every three hours based on time of day and placement, and allocates budget between branded and non-branded terms while controlling cannibalization between paid and organic. Our Amazon team then decides where the event budget goes, product by product.

An Amazon Black Friday strategy ends after Cyber Monday, not on it

Amazon data arrives with a delay of two to three days, so the real reading of Cyber Week happens in the first week of December. That week matters as much as the event itself, for three reasons:

  • Reading the results: compare TACoS and new-to-brand by product with the weeks before the event, rather than reading ROAS on the event alone. Amazon’s Long-Term Sales metric can help, keeping in mind that it is partly a forecast.
  • Retaining new customers: the customers acquired during Black Friday only become valuable at their second purchase. Plan targeted follow-up before the holiday period closes.
  • Preparing the next period: the weeks after Christmas often bring lower competition with buyers still active, so the products and terms that performed during Cyber Week are the ones to keep running.

Lock your Amazon Black Friday strategy this week

With three weeks to go, an Amazon Black Friday strategy comes down to clear arbitration: advertise what you can deliver, judge spend on total sales and new customers rather than ROAS, separate branded and non-branded budgets, and plan the reading of results for the week after Cyber Monday.

These decisions take one working session per category, provided the stock, margin and advertising data sit in the same place. Over 15 years and 300+ global brands, we have seen that the Cyber Weeks that end with profit are the ones where each product had a clear role before the event started.

If you want a second opinion before November 27, book a 30-minute review of your Amazon account. We look at your own data and show you which products to push, which to protect, and which to hold back.

Amazon Black Friday strategy: frequently asked questions

  • When should you start preparing for Black Friday on Amazon?

    Demand building should start in October, with consideration campaigns and deal submissions. The last three weeks are for arbitration: budget allocation, bids by time and placement, stock coverage and listing checks. Starting new campaign structures in November adds risk without leaving enough time for the algorithms to learn.

  • Should you increase Amazon ad budgets for Black Friday?

    Increase them selectively. Products with enough stock, a competitive deal and a good share of new-to-brand orders can absorb more budget. Products at risk of running out, or campaigns that mainly capture branded demand, rarely justify more spend. Set a TACoS ceiling for the event to keep the increase under control.

  • What is a good metric for Black Friday on Amazon?

    TACoS and new-to-brand orders are more reliable than ROAS during Cyber Week. ROAS rises mechanically with conversion rates and branded searches. TACoS shows whether advertising grows total sales, and new-to-brand shows whether the event recruits customers who can buy again during the following months.

  • What happens if a product runs out of stock during Cyber Week?

    The product stops selling at the peak of demand, part of that demand goes to competitors, and the ads that drove traffic until the last unit lose their value. Check days of cover before the event and reduce bids on products whose stock will not last until Cyber Monday.

  • What should you do on Amazon after Cyber Monday?

    Wait two to three days for complete data, then compare TACoS and new-to-brand by product with the pre-event period. Plan follow-up for customers acquired during the event, since their value appears at the second purchase, and keep the best performing products and terms active for the post-holiday weeks.

Read our insights