The November ad budget, week by week, with a stop rule.

When do you stop raising the budget in November?

Most plans answer that question in October, with a number. The number is usually last year’s spend plus whatever growth the business wants, spread across the weeks by a feel for when people buy. Then November arrives, the account behaves differently from the spreadsheet, and the number becomes either a ceiling nobody trusts or a target somebody chases.

There is a better answer, and it is not a number. It is a rule you apply every few days while the month is happening. This piece covers only that: the budget curve and the stop rule. The BFCM 2026 playbook owns everything else about the peak.

The stop rule in one sentence

Marginal ROAS is the return on the last dollar, not the average dollar: the extra revenue you got divided by the extra spend that got it. Breakeven ROAS is the return at which a dollar of ads pays for itself and nothing more. A 40% contribution margin means breakeven is 2.5x. A 25% margin means 4.0x.

The phrase “at the prices you are actually selling at” matters more in November than in any other month. If you are running 20% off, your margin this week is not your margin in March. The offer arithmetic shows how fast breakeven climbs with the discount, and the breakeven ROAS calculator does it per product.

Why average ROAS cannot tell you when to stop

Every budget curve bends. The first dollars reach the people most likely to buy, and each extra dollar reaches someone a little less likely. The average keeps looking healthy long after the last dollar has stopped paying, because the good early dollars carry it.

That is the trap. An account at 4x average ROAS can be spending its last $2,000 a day at 2x. If breakeven is 2.5x, that last $2,000 loses money every day it runs, and no average in any dashboard will show it. MER and ROAS both answer “how efficient is the whole budget.” The stop rule needs a different question: what did the last step buy?

Build the weekly curve from your own last November

Before the stop rule can correct a plan, you need a plan. Build it from your own history, not from a generic split. How your demand bunches into the peak depends on your offer, your category, how big your email list is, and whether your customers buy for themselves or as gifts. A percentage that suits someone else’s store tells you about their store.

1. Export last year’s daily store revenue for November and the first days of December.
2. Align it by event day, not by date. Black Friday 2025 was November 28 and Black Friday 2026 is November 27, so every 2026 date lines up with the 2025 date one day later.
3. Sum it into this year’s weeks and compute each week’s share of the period.
4. Apply those shares to the total you are prepared to spend. That is the draft.
5. Adjust for what is different this year: a new offer, a bigger list, a channel you did not run, a product that sold out.

Here is the draft for an illustrative brand planning $300,000 of spend for November 2 to 30. It is arithmetic, not a client result.

Two things to notice. First, the draft follows demand, which is a reasonable starting point because spend converts best when people are ready to buy. Second, it is only a starting point. Nothing in it knows where your response curve bends. That is what the stop rule is for, and in practice it will move money both up and down from this table.

One exception to following demand: the first three weeks. If part of your plan is to warm up new prospects before the weekend, those weeks deserve prospecting spend that will look inefficient on the week and pay off on Black Friday. Decide that deliberately and write the amount down. It is a choice, not something the stop rule should talk you out of.

How to estimate marginal ROAS during the ramp

You cannot see marginal ROAS directly. You estimate it by changing spend on purpose and watching what changes. The method is simple enough to run in a spreadsheet.

1. Hold a baseline for three days and record daily spend and daily store revenue. Use the store, not the platforms, because the question is what the business gained.
2. Raise total spend by a planned step, 20% to 30%, and hold it for two or three days.
3. Adjust for demand. Compare against the same weekdays, and index to last year’s aligned days, so a rising week does not get credited to the ads.
4. Divide the change in daily revenue by the change in daily spend. That is your marginal ROAS for the step.
5. If it is above breakeven, take the next step. If it is below, step back to the last level that cleared.

Make the steps planned and few. Google Ads can spend up to twice the average daily budget on a single day while keeping the month within its limit, so a daily figure is not a hard cap on any one day. Meta documents a learning phase that ad sets can re-enter after significant edits, budget changes among them. Many small reactive edits make both systems noisier. A handful of deliberate steps gives you cleaner reads.

Here is a ramp for the same illustrative brand. Contribution margin at event pricing is 40%, so breakeven ROAS is 2.5x. Figures are demand-adjusted daily averages for each block. It is arithmetic, not a client result.

Step 3 is where the stop hits, and it is the row that an average-ROAS view gets wrong. The account is still at 4.16x. Everyone looking at the dashboard would call it a great week. But the extra $2,000 a day bought $4,800 of revenue, which at a 40% margin is $1,920 of contribution. The step lost $80 a day. Step 4 would have lost $800 a day while the average still read 3.72x.

Notice also that the rule found room the draft did not. The draft had about $6,000 a day for the second week. The ramp says the account can carry $8,000 before the next dollar falls below breakeven. A plan built only from a spreadsheet would have left that on the table.

Treat a single read close to the line with care. 2.40x against a 2.5x breakeven is inside the noise of three days of data. The honest move is to step back and re-read, not to cut hard. A read far below the line, like step 4, is a clear answer.

Peak week: the curve moves, the rule does not

From Thanksgiving to Cyber Monday, conversion rates jump and the whole response curve shifts upward. Spend levels that failed the stop rule on November 13 can clear it comfortably on November 27. That is why the draft puts well over half the budget into Thanksgiving week and Cyber Monday, and it is why you should not carry the mid-month stop level into the peak.

What changes is the clock. You do not have time for three-day steps. Pre-plan the daily levels from last year’s aligned days, read in half-day blocks against the same hours last year, and apply the same rule faster. Seasonality adjustments are the bidding tool for the peak days; the BFCM playbook covers how to set them.

Google’s Performance Planner can forecast how conversions respond to budget, and it is a reasonable way to sanity-check your draft. Treat it as a forecast. The stop rule is measured on your own store revenue, which is the thing you actually bank.

What to do when CPMs rise

I am not going to give you a figure for how much November CPMs rise. It depends on the platform, the audience, the category, and the year, and I have not seen a primary source that makes one number honest. Pull your own CPM by day from last November and this October. That is the only curve that describes your auction.

What you can reason about is the arithmetic. ROAS moves with the revenue each thousand impressions produce, divided by what those impressions cost. If CPM rises 30% while revenue per thousand impressions also rises 30%, because conversion rates rise with demand, ROAS does not move. If revenue per thousand impressions rises only 10%, ROAS falls to 1.10 divided by 1.30 of where it was, about 15% lower. A 3.0x campaign becomes about 2.54x.

So a rising CPM is not, on its own, a reason to stop. The stop rule already includes it, because marginal ROAS is measured on what you got for what you paid. If CPMs rise faster than demand, the rule will tell you to stop earlier. If demand rises faster, it will let you go further. Watch the outcome, not the input.

What “spend it all on the weekend” gets wrong

The common advice is to hold budget back through early November and pour it into Black Friday week, because that is when people buy. It is half right. The peak does deserve the most money, and the draft above gives it almost half.

What the advice gets wrong is the idea that the right amount is decided by the calendar. It is decided by the curve. A brand whose curve bends early can overspend on Black Friday just as easily as it can underspend on the Tuesday before. And a brand that starves the first three weeks arrives at the weekend with fewer warm prospects to convert, which lowers the return on every peak dollar. The calendar tells you when demand arrives. Only the stop rule tells you how much of it is worth buying.

When this advice is wrong

• →You are deliberately buying customers, not profit. Then the stop line is a new-customer cost ceiling you have justified from repeat purchase, not breakeven ROAS on the week. What a new customer really costs covers how to set it.
• →Your daily volume is too small to read. If a step moves revenue by less than normal day-to-day variation, use weekly steps and larger changes, or accept a rougher plan.
• →Tracking or the store is broken. The rule reads store revenue for a reason, but if orders stall because checkout failed, no budget decision is valid until it is fixed.
• →You are running out of stock. Spending into a product that sells out on Saturday buys nothing on Sunday. Cap spend by inventory before the rule gets a say.

What to do this week

1. Compute breakeven ROAS at your November prices, per offer, not at full price.
2. Export last year’s daily store revenue, align it by event day, and build the weekly draft.
3. Decide how much, if any, of the first three weeks is warming spend, and write that number down separately.
4. Plan three or four budget steps for the first half of November, with dates, so the ramp is a schedule rather than a series of reactions.
5. Set up a simple daily sheet: spend, store revenue, same weekday last year, and a marginal ROAS column.
6. Pull your own CPM by day from last November, so you know what your auction did last time.

The peak season budget planner models a November curve from your current spend and ROAS and marks the point where the next dollar falls below breakeven. Use it to sanity-check the draft you built from last year. It gives you the plan. The ramp gives you the truth.

If you want the ramp run on your account by people who read it every day through the peak, our ecommerce PPC team does exactly that, and the free audit is the place to start.