Black Friday Google Ads Prep, Run as an AI Checklist

Black Friday falls on 27 November this year, and the accounts that have a good one will mostly have earned it in September. The classic prep list — budgets, bidding signals, creative, query hygiene — has not changed in years; what changed is the cost of executing it. Every item below is a prompt for an AI connected to the account — it reads your own history and runs the change on your command, in seconds — which turns "we should really start earlier" from a yearly regret into an afternoon.

The eight-week timeline, as prompts

Weeks out	Focus	The prompt that does it
8	Baseline	"Pull last November's daily spend, conversions and conversion rate by campaign; mark the surge days."
6	Budget plan	"Model this year's peak-week headroom per campaign from that history and current run rates; draft new caps."
5	Creative audit	"List every campaign whose ads mention no offer, and every asset older than six months."
4	Negative sweep	"Find queries with spend and no conversions in the last 90 days; draft negatives by level."
3	Structure freeze	"List everything still in learning; flag changes that would reset it before the peak."
1	Seasonality set-up	"Draft seasonality adjustment values from last year's peak-day conversion rates, per campaign type."
Peak	Daily check	The five-minute readout below, every morning.

The rest of this page is the reasoning behind the table, so the prompts get judged, not just pasted.

Why eight weeks, specifically

Working backwards from 27 November fixes the deadlines. Bid strategies need stable conversion data in the weeks before the event, so structural changes — new campaigns, restructures, bid strategy switches — want to be finished and settled well before November; hence the three-weeks-out freeze. Creative approvals, especially where compliance reviews are involved, reliably take longer than promised; hence staging at five weeks. And the budget conversation — the one where someone above you signs off on peak-week caps — is a September conversation because finance calendars make it one. Eight weeks is not superstition; it is the sum of the lead times, plus margin for the year something slips.

Budget headroom: model the surge from your own history

The only defensible surge forecast is your own last November. Have the agent rebuild it: daily spend and conversion rate per campaign through the peak, the multiple by which the surge days beat an ordinary day, and which campaigns carried it. This year's plan is that shape applied to this year's baseline — not a number borrowed from someone else's vertical.

Two mechanics matter when translating the model into caps. Google may spend up to twice a campaign's average daily budget on any day, balancing within the month — so surge days partly self-fund, and caps should be set with that doubling in mind rather than at it. And mid-flight budget moves are exactly the kind of reversible, spend-affecting change that belongs in an approval loop: the agent proposes the reallocation with the pacing maths attached, you click once. The month-long version of that discipline is budget pacing; peak week is simply pacing at higher stakes.

Seasonality adjustments: telling Smart Bidding what is coming

Smart Bidding learns from history, and Black Friday breaks history — the one week when conversion rates leap in hours, faster than any automated learner reacts on its own. The seasonality adjustment exists for exactly this: you declare the expected conversion-rate change in advance, bidding leans into it for the event, then reverts automatically.

The discipline is in the constraints. Google positions the tool for events of one to seven days and cautions against use beyond fourteen — it is a scalpel for the spike, not a Q4 mood. Set it days before the event, scoped to the campaigns that genuinely convert differently, sized from last year's measured jump — which the eight-weeks-out baseline pull already produced. Retail checkout rates justify aggressive values; lead-gen rarely does. And after the event, resist the symmetrical mistake: no downward adjustment is needed for the return to normal — the adjustment expires with the event and bidding resumes on its own evidence. Misuse runs one direction: an optimistic adjustment without a real conversion-rate jump behind it simply tells bidding to overpay for a week.

Creative: refresh with countdown discipline

Peak-week ads fail administratively more often than creatively — the offer went live but the ads still say last quarter's message, or the promo asset expired mid-sale. Stage the work three tiers deep, weeks early: evergreen RSAs refreshed against current queries, offer-specific variants drafted and held back, and promotion assets with start and end dates that match the calendar you actually announced. Drafts cost nothing while paused — the point is that on launch morning the only action left is "enable", which is one approval, not a writing session. For Performance Max, the same pass covers asset groups: stale imagery and no-offer copy are what the automated surfaces will otherwise amplify.

The negative sweep: clean the account before traffic multiplies

Whatever your account wastes per hundred clicks, the peak multiplies it — same leak, more water through it. So the cheapest Black Friday move is running the waste hunt before volume arrives: ninety days of search terms, every query with spend and no conversions, clustered into themes, returned as draft negatives at the right level for one approval session. Peak-specific candidates deserve their own pass: bargain-hunter queries that never convert for you, competitor-brand tourists, and — for lead-gen — the retail-intent queries that only appear in November. An account that sweeps in October polices strays in real time during peak week; an account that skipped it does archaeology in December.

The daily five-minute check during peak week

Peak week itself should be boring, because the work is done. One prompt each morning:

Since yesterday: spend and conversion delta by campaign; anything pacing to exhaust its budget before evening; any disapproved ad or asset; any search term over my cost threshold with zero conversions; anything I approved yesterday that now looks wrong. Rank by money at stake.

Five minutes, evidence-first, and the two or three interventions it surfaces — a budget nudge, a stray query negated, a disapproval appealed — each pass through the same approval click as any other week. Retail teams running deeper stacks can extend the same rhythm across feeds and the rest of the e-commerce setup.

What the daily check deliberately excludes matters too: no bid strategy changes mid-peak, no restructures, no reacting to a single strange afternoon. Auction prices spike and settle within hours during the peak; the strategies you tuned in October are reading the same signals you are, faster. The daily check exists to catch operational failures — exhausted budgets, disapprovals, broken pages — not to relitigate strategy at the most expensive possible moment.

The day after: read the event while it is fresh

The cheapest improvement to next year's Black Friday is a post-mortem prompt run in the first week of December, while context is intact:

Compare peak week to my plan: spend versus modelled caps by campaign, conversion rate lift versus the seasonality adjustments I set, queries that surged, ads and assets that carried the volume, and every intervention from the daily checks with its outcome. Write it up as next year's starting baseline.

That document is next September's eight-weeks-out prompt, pre-answered. Accounts that keep the loop running year over year stop re-deriving their peak playbook from scratch — the event becomes cumulative knowledge instead of an annual improvisation.

The September start is the whole trick, and delegating both the reading and the running is what makes September real. Start a free pilot and run the eight-week table against your own account — the baseline prompt takes less time than reading this page did.