When to Start Q4 Holiday Facebook Ads (2026)

Test cheap, scale expensive: the one rule of Q4 timing

By the time the Black Friday emails go out, the outcome is mostly locked. Q4 is not won in the last week. It is won in October, by the brands that already know which ad works before the auction turns against them. The reason is price. Meta ad rates climb through the fourth quarter and spike hard around the Thanksgiving-to-Cyber-Monday window, so every dollar you spend during the peak buys less reach than the same dollar spent a month earlier. Gupta Media, which tracks Meta CPMs across tens of billions of impressions, measured a Cyber Monday 2024 CPM of $17.70, the single most expensive day of the year and 138% above Meta's 2024 annualized average of $7.43. Black Friday itself ran $16.85, more than double the baseline.

Put those two facts together and the strategy writes itself. Testing is expensive by nature: you spend budget on ads that will not work in order to find the one that will. Do that testing during the peak and you are paying the year's highest prices to learn things you could have learned in October at less than half the CPM. So the rule for the whole quarter is short: test while reach is cheap, scale while it is expensive. Everything below is the calendar that makes that possible.

The other half of the picture is demand, and it is real. Adobe Analytics put US online spending for the 2024 holiday season at $241.4 billion, up 8.7% year over year, with Cyber Monday alone at a record $13.3 billion. That is why the expensive auction is still worth entering: the buyers are genuinely there. Your job is not to avoid the peak. It is to arrive at the peak with a proven ad and a warm audience, so you can spend into that demand with confidence instead of gambling on an untested hook.

The Q4 countdown, week by week

Here is the whole quarter as a timeline, anchored to Black Friday because that is the fixed point everything else orbits. The dates in brackets assume a late-November Black Friday, so slide them to match the calendar in any given year. The point is the sequence and the lead times, which do not change.

Read the table top to bottom and one thing stands out: the expensive weeks carry the least work. That is deliberate. All the heavy lifting (choosing the offer, producing creative, wiring tracking, finding a winner) happens before the auction gets pricey. By the peak, you are executing a plan, not building one. The brands that scramble during Black Friday week are almost always the ones that started in November instead of October.

Clear the learning phase before the auction heats up

The learning phase is the single hardest reason to start early, so it is worth being precise about the mechanics. When you publish or significantly edit an ad set, Meta's delivery system explores who responds best. The ad set exits the learning phase once it gathers about 50 optimization events (a purchase, a lead, or whatever you optimize for) within roughly 7 days, per Meta's Business Help Center. Below that threshold, results are too noisy for the system to deliver stably, and your cost per result swings around unpredictably.

That 50-event figure is what turns the calendar into arithmetic. Work a real example. Suppose a purchase costs you $12 and you optimize for Purchase. To gather 50 events in a week you need about seven a day, which is roughly $85 a day per ad set just to reach the threshold. Run three test ad sets in parallel to compare concepts and you are spending about $255 a day, or a little over $1,750 across the week it takes to exit learning. Then you want a second week of stable data before you trust the read, because a winner that looks great for three days can regress. That is two to three weeks per test round before you can name a confident winner.

Now stack a losing round on top, because your first batch of creative will not all work. That is why six to eight weeks out is the floor, not a comfortable buffer. Start in early October and you have room for a first round that flops, a second that finds the angle, and a stable read on the winner, all finished before CPMs climb. Start in mid-November and you are trying to exit the learning phase in the exact window when clicks are most expensive and every edit costs you dearly. If your cost per result is higher than $12 or your budget is thinner, you gather events more slowly and need to start earlier still, not later.

To turn all this into your own start date, count backwards from Black Friday in plain blocks. Reserve one week to exit the learning phase, a second week for a stable read on the winner, and a third block of one to two weeks for a losing first round, because the first batch of creative rarely lands. Add a few days of creative approval buffer in front of that. Sum it up and you land four to six weeks of pure testing runway before the sale, which is why launching in October (not November) is the honest answer for most accounts. If you optimize for a rare, expensive event or run a small budget, add a week; if you have a big budget and a cheap event that gathers 50 conversions in a couple of days, you can shave one off. The floor moves with your economics, but it never moves later than early November.

One more timing trap sits inside the learning phase: pausing an ad set for more than 7 days resets it. That matters in Q4 because the instinct to pause a "too expensive" ad on Black Friday can quietly throw away the stable delivery you spent October earning. If an ad is profitable, high CPM and all, leave it running through the peak.

Q4 CPM inflation: why the calendar taxes late starters

CPM is the price of reach, quoted per thousand impressions, and in Q4 it is set far more by the calendar than by your account. Gupta Media's tracker makes the shape of the year plain. The blended Meta CPM sat around $8.19 across full-year 2025 and $7.43 across 2024, but the annual average hides the peak. In 2024 the quietest reach of the year was in January near $6.05, and the most expensive was Cyber Monday at $17.70, which is 138% above that year's average. The week containing Black Friday, ISO week 48, averaged $13.42, making it the priciest week of the year. Reach roughly doubles in price during the peak, and it does so for everyone at once.

That is the crucial point about Q4 CPM inflation: it is not a signal that your creative fatigued or your account slipped. It is thousands of advertisers bidding for the same feed on the same days, and no amount of bidding harder on your side changes the auction dynamic. You cannot out-bid the season. You can only decide when to be in it.

Which reframes the whole timing question as a budgeting question. A dollar of testing in October buys more than twice the reach of a dollar of testing on Cyber Monday, so the cost of learning late is not a rounding error, it is a doubling. Spend the cheap weeks discovering what works and the expensive weeks pushing it, and your blended cost per result across the quarter stays sane. Do it the other way around and the peak CPM lands on top of your least efficient, still-learning ads, which is the most expensive mistake in the Q4 playbook.

There is a slower, structural version of this too. Meta reported that its average price per ad rose about 9% across full-year 2025, so the baseline itself drifts upward every year regardless of the season. That does not change the timing advice, but it does mean last year's peak CPM is a floor for planning, not a ceiling. Budget for reach getting a little more expensive each year on top of the Q4 spike.

Creative and offer lead time: what has to be done before November

The countdown only works if the creative is ready to launch when the calendar says go, and creative is where most Q4 plans quietly slip. You are not producing one hero ad. You are producing a batch, because you will test several concepts and keep only the two or three that earn their place. A realistic test is four to six distinct concepts, meaning genuinely different hooks, formats, or angles, not the same ad with a new headline. That is a production schedule, and it has to finish before the cheap-testing window opens in October, not during it.

Approval is the lead time people forget. Meta's ad review is largely automated and typically completes within about 24 hours, but the word "typically" is doing real work. Restricted categories (health and beauty claims, before-and-after imagery, alcohol, financial products, and more, per Meta's Advertising Standards) can take longer and need permissions sorted out in advance. Busy periods stretch review times. And any edit to a live ad's text, image, audience, or format sends it back through review, so a last-minute tweak can pause a running ad in the middle of the peak. Upload early, confirm every ad is approved, and treat "approved" as a milestone with days of slack in front of it, not an assumption.

The sale creative itself deserves its own pre-approval pass. Your prospecting ads and your Black Friday sale ads are usually different: the sale creative carries the actual offer, the deadline, and the urgency. Build those in early-to-mid November and push them through review then, so on the morning the sale opens you are switching on pre-approved ads, not submitting fresh ones and praying they clear before noon. Keeping research, creative production, and launch in one place, such as AdPlay.ai, compresses that lead time, but the discipline matters more than the tool: whatever your setup, the creative has to be finished and approved before the auction turns against you.

Adobe's read on 2024 reinforces why the offer timing itself is moving earlier. Its analysts noted that early discounts were strong enough that many shoppers "hit the buy button" earlier in Cyber Week, with Cyber Monday becoming a last call rather than the only call. Shoppers now expect deals to open before the traditional peak, so a brand that waits until Black Friday morning to show its offer is late to its own sale. Warm the audience with the offer in the days before the peak, then let the peak be the crescendo, not the debut.

The budget ramp: scaling into the peak without a reset

Once October's testing has handed you a profitable, stabilized ad set, the peak is about pouring budget into it, and the mistake here is doing that too fast. A large budget jump counts as a significant edit, which throws the ad set back into the learning phase precisely when you cannot afford the instability. So the ramp is a schedule, not a single move.

Work the example. Say a winning ad set is stable at $100 a day the week before Black Friday and you want it at $400 for the peak. Jump straight there and the 4x change resets learning at the worst moment. Raise it about 20% every day or two instead: $100, then $120, $144, $173, and so on. Reaching $400 is roughly eight of those steps, so a jump that size wants a 10-to-14-day runway. That is exactly why the ramp starts in early-to-mid November: begin then and the ad set is sitting at peak budget, still out of learning, by the time the sale opens.

If you need to scale faster than a 20% ladder allows, duplicate the winning ad set and run the copy alongside the original. The duplicate goes through its own short learning phase, but the proven ad set keeps delivering untouched, so you are adding capacity rather than risking the thing that already works. Either way, batch your changes and judge each move on a couple of days of data, not a couple of hours.

A note on daily-budget nerves during the peak. Meta paces a daily budget across the week and can spend somewhat over on a high-opportunity day, then balance it back, so a single spiky Black Friday day is not a reason to slam the budget down in a panic. Slamming it down is itself a significant edit. Set the budget you can afford for the week, let the system pace it, and read the week, not the hour.

Post-peak: the retargeting window most brands leave on the table

The week after Cyber Monday is one of the most efficient of the entire quarter, and plenty of brands switch everything off right when the math turns in their favour. Two things happen at once after the peak: CPMs fall back toward the annual baseline as competitors pull their budgets, and you are now sitting on the largest, warmest retargeting pool you will have all year, built from everyone who visited, added to cart, or engaged during the ramp and the peak. Cheaper reach plus warmer audiences is the best combination on the calendar.

So plan the post-peak, do not just let it happen. Keep retargeting live for one to two weeks after Cyber Monday, aimed at cart abandoners and product viewers who did not buy, with a clean reminder of the offer while it lasts. As mid-December approaches, shift the urgency to shipping deadlines, which is a genuine, honest scarcity that converts: "order by the 18th to arrive by the 24th" does real work without a discount. Then, as physical-gift shipping windows close, pivot the message to digital gift cards, which have no shipping deadline and sell right up to the holiday itself.

January is the quiet bonus round. Reach is cheap again, gift-card recipients are looking to redeem, and post-holiday clearance gives you a fresh offer for a warm list. It is the mirror image of October: low CPMs, so a good time to test the next quarter's creative on cheap traffic while you run a clean-up sale to the audience Q4 built. The brands that treat the last week of the holidays as the finish line miss the cheapest, warmest weeks bracketing the peak on both sides.

A Q4 timeline you can actually run

Strip it all back and the whole quarter comes down to putting the hard work before the expensive weeks. Six to eight weeks before Black Friday, lock the offer, finish and upload the creative batch, and verify that the Pixel and Conversions API are firing clean events, because the learning phase can only count what your tracking actually reports. Launch your prospecting campaigns in October the same way you would run any Facebook ad, and let them run untouched long enough to clear the learning phase (about 50 events in 7 days) and hand you a confident winner, allowing for one round that flops.

By early November, testing is done, and the job flips to preparation: build retargeting audiences off the traffic you have gathered, push the sale creative through review while there is slack, and start ramping the winners in roughly 20% steps so they arrive at peak budget still out of learning. During the peak itself, do as little as possible: scale the proven ads, run the pre-approved offer, and resist every urge to edit settings or pause a profitable ad over its CPM, because $16.85 reach that converts is cheaper than $8 reach that does not. Then keep going after the peak, when CPMs ease and your warm pool is at its biggest, through shipping-deadline urgency in December and gift-card redemption into January.

None of this needs a bigger budget than the brand next to you. It needs an earlier one. The single highest-leverage decision in Q4 Facebook advertising is not the offer or the creative or the audience. It is the date you start, because that date decides whether you enter the most expensive auction of the year with a proven winner or a hopeful guess.