Peak Amazon Ads: Budget, Bids and Pacing

A bid change on Amazon needs about two weeks of data before anyone can read it. Black Friday through Cyber Monday does not last two weeks. Practitioners who tune Sponsored Products bids for a living hold a change for about two weeks before judging it, so the last bid you can honestly evaluate has to be live weeks before the event. The account you take into peak is, in almost every material respect, the account you have already built.

So the useful question seven weeks out is not what to change during the event. It is what number governs what you are already spending, and it is not a benchmark ACOS target. It is your maximum allowable ACOS, computed from your own costs: break-even ACOS minus the margin you intend to keep. What is left in your hands once the event opens is budget headroom, placement settings already in place, and creative that has already cleared review.

Which advertising deadlines fall in the next 30 days?

The dates that matter most here are not dates at all. They are standing Amazon lead times on creative approval, and they bite whether or not Amazon publishes a seasonal cut-off. The register below is checked to 3 October 2026; where Amazon has not announced a 2026 date, last year's shape is shown with the page to watch.

Prime Big Deal Days runs 6 to 7 October, the day after this is published, which means its creative window has already closed and the only useful work left on it is budget headroom and placement modifiers. Two other rows constrain the plan rather than the account. There is no point pacing budget into an item that will be out of stock, so the inbound cutoffs cap what is worth advertising at all; week 1 and week 8 of this series work that through. And the 20 October deal deadline (the Seller Central date; vendors confirm theirs with the vendor manager) decides which ASINs carry a badge, so the promotional calendar has to be fixed before the media plan can be finished; week 3 covered deal fees and discount depth.

What happens to your costs during peak?

Clicks get more expensive, and the same ACOS target at a higher price per click buys fewer of them. Industry benchmarks put the peak CPC lift during Q4 and Prime events at roughly 20% to 30% above baseline, but those are aggregated vendor figures with no category or marketplace attached, so pull your own account's last two Novembers rather than adopting the range. For direction of travel only, one benchmark compiler puts the 2026 average cost per click across categories at about $1.13, with category averages running from roughly $0.40 in books to $1.42 in health and household; the spread is what matters there, not the average.

Which is why the budget question and the bid question are one question. Hold your ACOS target flat while cost per click rises and you have quietly cut impression share. Hold impression share flat and you have quietly raised your ACOS.

What number should govern the Amazon ads budget for Black Friday?

Your maximum allowable ACOS, and it is arithmetic off your own profit and loss rather than anything Amazon publishes. Break-even ACOS is sale price minus all non-advertising costs, divided by sale price, times 100. Maximum allowable ACOS is that figure minus the margin buffer you decide to protect. Break-even is the ceiling; the maximum allowable number is the target you manage to.

Two things move the cost side before Black Friday arrives. Amazon's peak fulfilment surcharge runs 15 October to 14 January, so every unit sold through FBA in the window carries more cost and break-even falls with it; a 1P vendor carries it only on any 3P side. Discount depth does the same to the price side, which is why the deal calendar locks first. For reference, one published vendor guide puts target ACOS at roughly 30% to 50% at launch and 20% to 30% in growth, and somewhere between 10% and 30% once a product is established, with total advertising cost of sales under 10% for established products. Those come from a vendor rather than Amazon, and a band like that is a sanity check, never a substitute for your own unit economics.

One input most mid-market teams never price in changes the shape of the question. On the largest accounts, part of the advertising and promotional budget is a negotiated line rather than a media decision. When I recruited and launched the Furniture category's largest new retail vendor at Amazon, a major furniture brand, the package that got it live was built around promotional funding and vendor-funded Amazon headcount. The items were new, so there was no cost to cut; the advice to the brand was to give Amazon the sharpest pricing it could stand and hold back enough budget for advertising and promotional funding to grow the business. Some of the profitability problems on a new item also solve themselves with velocity: more units in Amazon's fulfilment centres means stock closer to customers and lower shipping cost per unit, and enough velocity can make an ASIN eligible for the Direct Import programme. The person across that table is the vendor manager, who owns the profit and loss that funding lands in and will press a new vendor for its sharpest cost. Concede too much there and the margin to fund advertising and promotions is gone, and Amazon is a pay-to-play environment, so that becomes a drag on growth. Amazon will occasionally help fund a promotion itself, when it has strong margin of its own on the item or would lose that margin anyway to its own markdown algorithms, but that is rare. So before asking how much you should spend, ask whether the cost you agreed leaves you the margin to fund it. Eligibility and funding routes differ between Vendor Central and Seller Central, so the answer is not the same on both sides.

On pacing, the accounts I have run followed one rule: advertising is always on, and the fortnight before a tentpole is a gradual ramp, not a cliff. Cutting spend before the event and pouring it in on the day hands the algorithm a campaign it has not learned yet, on the two days it most needs to have learned it. Ramp in, so the learning is done before the traffic arrives.

When does the tuning window actually close?

For most structural changes it has already closed, or it closes in the next fortnight. The read windows on Amazon's ad products are longer than the events they would be tuning for, and the practitioner methods built on them are longer still.

• A bid change needs about two weeks before it can be judged. That is the practitioner standard rather than an Amazon rule, and Black Friday through Cyber Monday does not run that long even in the long-window shape Amazon used last year.
• The adjustment method is incremental. A common practitioner method scales the current bid by target ACOS over actual ACOS and moves in small steps; one published guide cuts 15% to 20% a pass and raises 10% to 20%. Two passes with an honest read between them is a month.
• Switching bidding strategy needs history. Practitioners keep a campaign on the down-only strategy until it has a sustained record at target, then move it to dynamic up and down. If you do not have that record now, you will not have it by November.
• Sponsored Display takes longer still. View-based remarketing stabilises in 14 to 21 days, and audience or prospecting campaigns need 30 to 45 days before the numbers mean anything. Those come from vendor analysis rather than Amazon, but the consequence holds either way: an audience launched inside the peak window cannot be evaluated inside it.

Finish structural work in the next two weeks, then treat the event itself as execution.

What can you still change during the event?

Three things, none of which needs a two-week read: budget headroom, placement modifiers, and switching off what is visibly failing.

Budget headroom is worth preparing now. The advertising console marks a campaign Out of budget the day it happens, and Amazon's schedule-based budget rules can raise a campaign's budget for a date range or, in the US, for set hours of the day, so the first protection is a setting, made before the event. What the console does not give you is one hour-by-hour feed across every campaign. Amazon Marketing Stream pushes hourly campaign metrics, budget usage among them, through the Ads API into your own AWS account, covering Sponsored Products, Sponsored Brands, Sponsored Display and DSP. It needs a real Ads API integration with no interface-only version, and AWS infrastructure costs apply, so getting it live for peak is an engineering ticket this month.

Hourly data has changed a decision for me, and the clearest case was furniture. A high-ticket item carries a long consideration time, and the purchases clustered after working hours, presumably once the buyer had talked it over at home. The budget, paced evenly, was often gone by then. The fix was dayparting: hold enough of the daily budget back so the ads were still live in the evening, in front of the shopper at the moment they were ready to buy. Without the hourly view, that pattern reads as a campaign that simply ran out of money.

Placement modifiers are the in-event lever, not the base bid. Sponsored Products raises bids by a percentage, up to 900%, separately for top of search, rest of search and product pages. Vendor analyses report that top of search converts and earns clicks at a materially higher rate than the others and costs more accordingly, which is why the modifier is where peak money goes; Amazon itself publishes no placement-level conversion figure and no recommended modifier. The trap is stacking: practitioners describe modifiers as applying to the base bid first, with dynamic bidding compounding on top of the already adjusted bid, and that is how a peak click ends up costing far more than intended. It is a practitioner account rather than an Amazon statement, so check it against your own bid reports before setting an aggressive combination.

Know which bidding strategy each campaign is on before the traffic arrives. Sponsored Products offers dynamic up and down, where Amazon can raise the bid by up to 100% when a click looks likely to convert and lower it when it does not; dynamic down only, which lowers but never raises; and fixed, which leaves your bid alone. Amazon's dynamic bidding guide puts that increase at up to 100% for all placements, not top of search alone, so a campaign on up and down can double its bid on a product page as readily as at the top of the results. Vendor analysis describes Sponsored Display bid adjustments by placement (detail page, home page and off-Amazon inventory) and reports the detail page as generally the strongest performer. No Amazon page documenting that control was found, so confirm it in your own console before planning around it.

Is the creative going to clear in time?

Only if it is submitted with a week of slack and built to the right spec set. Amazon has not published a holiday creative submission deadline for 2026, so the operating rule is Amazon's standing guidance: submit Sponsored Brands and Sponsored Display ads at least one week before launch, with room inside that week for a rejection and a resubmission. Watch the Amazon Ads holidays page in case a seasonal cut-off appears.

Sponsored Brands video is where rushed peak briefs go wrong. Amazon's spec page calls for 6 to 45 seconds, with 20 seconds or less highly recommended; 16:9, square pixels, progressive; 1280x720, 1920x1080 or 3840x2160; MP4 or MOV up to 500MB; H.264 or H.265 at a minimum of 1 Mbps, with 4 Mbps or better recommended. It autoplays muted with a mute toggle at lower right, so the opening frames have to work silently. Amazon revises these pages without notice, so re-check the Sponsored Brands video specs before commissioning anything. Custom media and online video ads are a separate spec set, up to three minutes with 15 seconds or less recommended, also autoplaying muted. A three-minute asset is invalid for Sponsored Brands video, and confusing the two is an expensive, avoidable mistake in a compressed schedule.

Three constraints sit around the asset. On Seller Central, Sponsored Brands and Sponsored Display are open only to sellers enrolled in Brand Registry; vendors are eligible for both without it, and Sponsored Products needs it on neither side. Brand Store moderation takes up to 72 hours on the longer of Amazon's two published figures, and a Sponsored Brands ad built to land on that Store has nowhere to send traffic until it clears. On the detail page behind it, basic A+ content allows five static modules and Premium A+ supports up to seven per ASIN; for sellers, A+ is free with a Professional selling plan and Brand Registry, and A+ text is not indexed for search. Week 6 takes detail pages properly. On cadence, vendor analysis of Sponsored Brands creative puts typical refresh at every 60 to 90 days with click-through decay reported to begin around day 45. Treat that as a prompt to check your own click-through trend rather than a rule, and note that it describes decay, not approval time.

What has to be decided this week?

Fix the promotional calendar before 20 October so the media plan has something solid under it, then compute maximum allowable ACOS per SKU with the 15 October fulfilment surcharge already in the cost base wherever you sell through FBA. Get creative into moderation with a full week of slack, and put any Amazon Marketing Stream integration you want for peak into the engineering queue now. Make your structural bid and campaign changes in the next fortnight or accept that you will not be able to read them.

The eCommerce strategy work we do with Amazon brands starts at that ACOS calculation, because a brand advertising to a benchmark rather than to its own break-even is either buying unprofitable volume or leaving margin unspent, and neither is visible from the dashboard. What you decide this week is what runs in November. What lands after the volume is Amazon deductions, which is week 11 and a different problem.

If you want the break-even and maximum allowable ACOS worked per SKU before the structural window closes, a 45-minute working session on your campaign list and unit costs is the fastest way to get there. Get in touch to book one.