Prime Big Deal Days 2026: The Complete Amazon Ads Playbook

Amazon confirmed the dates today. Prime Big Deal Days 2026 runs October 6 and 7. Forty eight hours, Prime members only, more than 35 categories. From the day this publishes, that is about three weeks of runway.

Most sellers will read that and put two days in a calendar. That is the expensive move. Prime Big Deal Days is not a two-day event in any sense that matters to an ad account. It is a five-week event with a two-day spike in the middle. The money is made in the three weeks before, when consideration is still cheap and the auction has not tightened, and in the two weeks after, when you hold the largest pool of warm traffic you will have until Black Friday. The two days themselves are what everyone competes for, which is exactly why they are the least efficient hours in the sequence.

The full playbook follows: what Amazon confirmed, why October outweighs July, what to do in each phase, how to build the scaffolding, a dated calendar, and the five mistakes that cost the most.

What Amazon confirmed

Amazon Prime Big Deal Days 2026 runs October 6 and 7, starting at 12:01 a.m. PT on Tuesday, October 6 and ending at 11:59 p.m. PT on Wednesday, October 7. A clean 48 hours, and the boundaries are worth being precise about, because the start time is what catches people out.

12:01 a.m. Pacific is 3:01 a.m. on the east coast. If your campaigns run on a profile-local schedule with anything paused overnight, the opening hours pass with your ads dark while a meaningful share of the country is already awake and already buying. That is the best hour-for-hour traffic of the quarter, gone by the time most sellers log in.

The event is exclusive to Prime members and spans more than 35 categories. Breadth matters more than depth: your category is almost certainly inside the sale whether or not you have a deal approved, and the auction will be crowded either way.

Three events get treated as interchangeable and should not be. July Prime Day is the volume event: the largest raw traffic spike on the Amazon calendar, a shopper buying for themselves with no deadline beyond the sale. If that is the one you are optimising for, the Prime Day PPC strategy guide is the read. October is the intent event: less raw traffic, but the shopper is buying for someone else, inside a season, against a delivery date they already feel. Black Friday and Cyber Monday is the price event: the highest competition and highest CPCs of the year. Our earlier coverage of what Big Deal Days is has the background.

Why October matters more than the July event

July Prime Day is a bigger day. October is a bigger decision. If you keep one line from this guide, keep that one.

The first reason is cost, and it is structural rather than seasonal. CPCs across most categories begin climbing in the first week of October and, in the accounts we manage, do not come back down before Black Friday. Budget pressure from every seller building toward Q4 enters the auction at roughly the same moment and then stays in it. Whatever your cost base looks like in the second week of October is approximately your cost base for the quarter. You are not buying clicks at an October price and returning to September prices in November. You are setting a floor.

The second reason is the audience. Every impression and detail-page view you buy across the ramp and the event is an asset you can retarget in November. A shopper who viewed your listing on October 6 and did not buy is not a loss. They are inventory, and the cheapest audience you will reach all quarter. A seller who sits out October does not simply skip a sale. They arrive at Black Friday paying Black Friday prices to talk to a cold audience.

The third reason is behavioural, and the one operators underrate most. Amazon's delivery system learns from what you do. The budgets and bids you set during the ramp anchor how your campaigns buy for weeks afterward. A campaign that spent early October capped and spending out by lunchtime carries that history into November. A campaign funded and competitive across the same window carries a different one. That is why one jump on October 5 does not buy what three weeks of graduated increases buys. If you are planning the whole quarter, the Q4 ads planning guide has the September-through-December sequence.

The three-phase playbook

Three phases, and they are not equal in length. Phase 1 is three weeks. Phase 2 is two days. Phase 3 is about two weeks. Most sellers run Phase 2 competently and skip the bookends entirely, which is why they extract a fraction of the value from an event available to everyone on identical terms. Run all three. The bookends are cheaper than the middle and they are where the compounding lives.

Phase 1, the ramp (September 15 to October 5)

In the three weeks before the event you are not buying conversions. You are buying consideration. That distinction changes which numbers you are permitted to look at, and it is the hardest part of this for a disciplined operator to accept.

During the ramp, the metrics that matter are impressions, reach, and new-to-brand orders. Same-day ACoS is the wrong lens, because you are paying for exposure that converts in October rather than today. Judge ramp spend on ramp-day return and you will cut it in week two, then arrive at October 6 with a cold account and an expensive auction. Expect ACoS to look worse across this window. That is the intended trade, not a failure. Write it down before you start, so the version of you at the dashboard on September 28 does not quietly undo the plan.

This is where Sponsored Brands does the heavy lifting. Headline and video placements buy recall at the top of a category before the auction gets expensive, and recall is what converts on October 6 when a shopper scans a crowded results page. Sponsored Products captures demand that already exists. Sponsored Brands creates the demand Sponsored Products will capture three weeks later.

Ramp progressively rather than in one move. Begin lifting Sponsored Brands bids and budgets around 10 to 14 days out and step them up two or three times, instead of making one large increase on October 5. Placement history accrues, and Amazon's system rewards a campaign that has been consistently competitive over one that appeared yesterday holding a large bid. The algorithm also needs conversion signal at the new spend level before the event rather than during it. A campaign that triples its budget on the morning of October 6 spends day one learning, and day one of a 48-hour event is not a good classroom.

Take Northlane Goods, a mid-size home and kitchen seller. Rather than holding flat through September and going hot on October 5, they step Sponsored Brands budgets up in three moves: a modest lift mid September, a second around the 25th, a third at the start of October, with headline and video placements running against category terms rather than only their own brand terms. By October 6 the brand has sat in front of that category for three weeks, and the event-day auction costs them less per acquired customer than it costs the seller who turned up that morning.

The ramp is also when you clean the account before you spend into it: negatives, search terms, out-of-stock ASINs. Raising budgets on a messy account just spends faster in the wrong places.

Phase 2, the event (October 6 to 7)

For 48 hours, everything runs higher than normal. Bids, daily budgets, the campaigns you would ordinarily throttle. This is the one stretch of the quarter where the correct instinct is to worry more about being absent than about being expensive.

The most common failure is budget exhaustion before noon on day one, and what makes it worse than an ordinary bad day is that it cannot be recovered. A campaign that spends out at 11 a.m. on October 6 does not get those hours back. No make-good, no catch-up, no evening where the traffic returns. You bought the first few hours of the highest-intent window of the quarter, then went dark through the afternoon while everyone else kept buying. The failure modes are asymmetric: overfunding costs some inefficient spend, underfunding costs the rest of the day. More in our note on budget-exhausted campaigns.

The second failure is subtler, and it is about dayparting. Your ordinary-week schedule is built on an ordinary week. Hours that are reliably dead on a Tuesday in September, the overnight block in particular, are not dead during a 48-hour Prime-exclusive global sale. Shoppers check deals when they wake up, on the commute, at lunch, and at midnight when the next wave drops. An aggressive overnight pause left switched on through October 6 and 7 costs you money in precisely the hours where the marginal shopper is most likely to buy without comparison shopping.

Widen or suspend the schedule for the window rather than deleting it, because deleting is how sellers end up running around the clock for the rest of October without noticing. Whatever you do, make sure it reverses itself on October 8. If your schedule was built on the reasoning in dayparting through a Prime event, the same logic carries. The schedule is not wrong. The week is.

Pace across both days instead of treating them as one long day. The first is almost always the more expensive: every seller with a budget is in the auction on October 6, deal coverage is heaviest, and much of the traffic is still browsing. Day two frequently comes in cheaper per click and converts at least as well, partly because competitors have spent out and partly because the Wednesday shopper has already done their looking. Hold something back for it.

Harbor Kitchen learned this the expensive way. In an earlier October event they set one large increase across both days, front-loaded bids on day one, and were capped across most of their catalogue by mid Wednesday morning. The following year they split the window: funded day one generously but short of exhaustion, held a reserve for day two, and ran the schedule wide across all 48 hours. Same total spend, materially more of it landing in hours where it could convert.

Phase 3, the tail (October 8 to roughly October 20)

Phase 3 is where most of the unclaimed money sits, and it goes unclaimed for a boring reason. The event is over and everyone has moved on.

Consider what you own on the morning of October 8. You have just paid for a very large volume of traffic across 48 hours, and most of it did not convert, because most traffic never does. Those shoppers viewed your detail pages, compared you against a competitor, added to cart and abandoned, or ran out of time. They are warm, recent, and addressable. Sponsored Display retargeting against those views, and against your own detail pages, is the cheapest meaningful traffic you will buy all quarter, cheapest in the ten to fourteen days right after the event.

Run views-based retargeting against the audience the event built, and product targeting across your own catalogue for the cross-sell and the second purchase. Two weeks is roughly the right horizon. After that the audience cools and the money is better spent on your Black Friday ramp. The BFCM prep guide covers that handoff.

The other half of Phase 3 is restoration, the mistake that costs the most while making the least noise. On October 8, every elevated budget and bid you set for the event is still elevated, and nothing inside Amazon turns them back down for you. Raise daily budgets across a catalogue for a 48-hour sale with no restoration plan and you will spend the rest of October buying ordinary traffic at event prices. Sellers usually discover this in November, finding two weeks of spend they cannot account for. Preventable by deciding on October 1 how the numbers come back down rather than trying to remember on October 8.

Sunhollow Supply treats this as a hard rule: nothing goes up for an event without a written restore date attached before the increase is made. Their post-event routine is two things only. Confirm every budget and bid is back at baseline by October 8, then spend the following fortnight retargeting the audience the event built.

Setting this up in Off Hours

One thing to be straight about first, because the strategy above spans all three ad types and Off Hours does not. Off Hours automates Sponsored Products campaigns. Sponsored Brands and Sponsored Display are on the roadmap and are not automated by Off Hours today. The Sponsored Brands ramp in Phase 1 and the Sponsored Display retargeting in Phase 3 are work you do yourself in the Amazon Ads console. What Off Hours carries is the Sponsored Products execution layer, plus the scheduling, budget and event scaffolding underneath the whole plan: the part that fires on time, holds the window, and puts everything back where it was. Off Hours supports Amazon Seller Central accounts in North America, meaning the US, Canada and Mexico. Vendor Central is not supported.

Event rules carry the October 6 and 7 window

The event rule is the centrepiece, because this is the shape it was built for: a one-shot tent-pole with a ramp, a peak and a tail. Define the ramp into October 5, the peak across the two event days, and the tail after. The rule captures your original values before it changes anything and restores them cleanly when the window closes.

That last part is not a convenience feature. It is the direct answer to the Phase 3 mistake. Budgets never come back down because coming back down is a manual task scheduled for a day when everyone is exhausted. An event rule makes restoration part of the same object that made the increase, so you cannot set one without the other. How event rules work walks through ramp, peak and tail, and the event rules page has the short version.

Budget rules carry the Phase 1 lifts

The progressive ramp is a series of deliberate, dated increases, which is precisely what budget rules are for. They raise and reduce daily budgets on recurring schedules, date ranges, or daily windows, and budgets restore to baseline when the window ends. Set your Sponsored Products ramp steps as dated ranges in September and the account gets the graduated lift without you making the same change by hand three times, and without any of those increases becoming permanent by accident. That accidental permanence is how a two-day event becomes a six-week spend problem. See budget rules.

Dayparting carries the 48 hours

Dayparting is a 7x24 grid, one cell per hour of the week, enabling or pausing campaigns, checked every 15 minutes. For the event, widen the grid across October 6 and 7 rather than throwing away a schedule you spent months tuning. Open the overnight hours and the early morning block that would otherwise sleep through the 12:01 a.m. PT start, then restore the ordinary schedule on October 8.

The 15-minute check is what makes those boundaries usable. An hour that opens at midnight opens within 15 minutes of midnight. On a normal Tuesday that is a nicety. On October 6 it is the difference between catching the opening wave and reading about it. Dayparting rules covers the grid in full.

Performance rules are the guardrail, with one honest caveat

Performance rules react to prior-day metrics: ACoS, spend, budget utilization, CTR, ROAS. They run once daily in the early morning rather than every 15 minutes, a deliberate design decision rather than a limitation we are working around. Amazon's reporting lags somewhere between 3 and 12 hours, so a rule evaluating ACoS every 15 minutes would react to numbers that are not finished yet, and would thrash your account doing it.

Be clear-eyed, then, about what they do inside a fast 48-hour event. They are a safety net and a next-morning correction, not a real-time throttle. A rule firing on the morning of October 7 acts on settled October 6 data, the best read you will get before you commit day two. It will not catch something going wrong at 2 p.m. on day one.

So prefer Alert actions to Pause during the event itself. The three action types, Alert, Adjust Budget and Pause, run in ascending order of how much damage they do if they fire on the wrong information. A destructive action firing on lagged data in the middle of your highest-value 48 hours is a bad trade even when the rule is right most of the time. Let it tell you, and make the call yourself. How performance rules work goes deeper on the timing.

Bid adjustments do the quiet compounding

Bid adjustments step keyword and target bids up or down on each target's own trailing ROAS, using a step size you choose, a cooldown between moves so nothing runs away, and a ceiling you set. Every change is undoable. Through the ramp and the tail this is the layer that works while you are not watching, and the ceiling is what makes it safe to leave running while CPCs rise underneath you. During the event, keep the ceiling honest. A ceiling you raise in a panic on the afternoon of October 6 is not a ceiling, and the point of setting one in September is that September you is calmer than October you.

Spend Alerts sits underneath all of it

Spend Alerts watches each account's daily spend against that account's own 14-day rolling baseline and tells you when something has moved out of pattern. On by default, no setup, free on every plan. During an event where you have deliberately made your spend abnormal, it is what notices when abnormal turns into wrong, and what catches the October 9 morning where a budget quietly failed to come back down.

The calendar

The whole sequence on dates. The right-hand column is where each job actually lives, which is the part that gets lost when a plan moves from a document into an account.

Five mistakes that cost the most

The first is budget exhaustion before noon on day one. The most common and the most expensive, because it is the only one here that cannot be undone later the same day. Work backwards from the hour your campaigns currently spend out on a strong day, and fund the event so that hour arrives in the evening rather than before lunch.

The second is leaving an ordinary-week dayparting schedule running through the event. Your schedule encodes months of learning about a normal week, which is the right thing to encode and the wrong thing to apply to a 48-hour Prime-exclusive sale. The dead hours are not dead. Widen the grid for the window, set it to restore on October 8, and do not delete it.

The third is having no restoration plan. This one is quiet, which is why it is dangerous. Nothing breaks. Your elevated event budgets simply keep running through mid October, and you find out when you reconcile the month. Put the restore inside the same rule as the increase rather than inside a note to yourself.

The fourth is ignoring the tail. You spent real money assembling an audience over 48 hours, then walked away from it on the morning of the eighth. Skipping Phase 3 does not save money. It writes off an asset you already paid for.

The fifth is changing everything at once on the morning of October 6. New bids, new budgets, new schedule, restructured campaigns, all at 8 a.m. on day one. Nothing afterward can be attributed to any single change, nothing rolls back cleanly, and if something goes wrong you have no idea which of the nine things you did caused it. Make your changes in the ramp, freeze from October 3, and let the event run against settings you trust. That is not caution. It is the only way to learn something from this event that you can use in November.

Off Hours runs the dayparting, budget, event, performance and bid rules you set on your Sponsored Products campaigns, every 15 minutes, on every connected account, with Spend Alerts watching free on every plan. Flat $149 per account. Start a free 14-day trial.