Black Friday Google Ads: set the campaign now, fix the site before the sale
The November budget is approved, the campaign is half built, and the page the clicks will land on is the one that converted 3% in March. What to set now, what to leave alone in sale week, and where the remaining weeks pay back most.
- Late changes cost more than they fix. New campaigns, restructures and big target changes force Smart Bidding to relearn, and in sale week the relearning runs at the year’s highest CPCs.
- Leave seasonality adjustments off for Black Friday. In Optmyzr’s 2024 cohort, accounts that used them paid 34% CPC inflation against 17% and lost 15.7% ROAS while the others gained 5.7%.
- Sale week raises the price of the leak, not its size. At a 3% conversion rate, 970 of every 1,000 paid visitors still leave; each one simply cost 17% more to bring, so a point of conversion rate is worth more in November than in any other month.
- The one site change that needs no campaign change is a page that keeps the ad’s promise per visitor. Measure it on a 50/50 split of the same week’s traffic with bots removed, and count each purchase once.
Every Black Friday the same account does two things at once: it pays the highest cost per click of the year, and it sends that click to a page built for a March visitor. Make the campaign changes now, leave Smart Bidding alone once the sale starts, and spend the weeks in between on the page the clicks land on. The first two are cheap and well documented. The third is where the budget is decided, because the sale-week visitor sees exactly what every other visitor sees unless the page adapts to the ad that brought them.
This is written for an online store planning its November Google Ads budget. The conversion event is a purchase, measured as revenue per paid visitor rather than conversion rate alone. The seasonal framing stops here: peak-week paid traffic behaves the same way at a January clearance or a product launch, so the body is meant to be useful in March as well.
What the sale week does to a Google Ads campaign
Three things happen to a campaign in sale week, and two are outside your control. Demand rises and the cost per click rises with it. Smart Bidding sees the conversion rate climb and raises bids on its own: in Optmyzr’s study of 5,886 accounts published in October 2026, accounts that made no manual adjustment still saw CPCs climb 17% over their pre-sale baseline in 2024, while their ROAS improved 5.7%. The third is the one you control: whether the system is learning or executing when the expensive clicks arrive.
The same study is blunt about the learning point: “New campaigns, restructures, and big target changes all force the bidding system to relearn.” A learning period in a quiet week costs you a quiet week. One that starts the Tuesday before Black Friday runs through the most expensive clicks of the year, and the account pays full price for every bid the model gets wrong while it adjusts.
Optmyzr’s three-year study (November 2025) shows the same direction every year from 2022 to 2024, with the conversion rate lift Smart Bidding found on its own shrinking from 17.5% to 7.5% as more advertisers crowded into the same auctions. That shrinking lift is the point of this article: the auction is getting more efficient at pricing the sale, so the margin you can win by outbidding competitors is thinning, and the margin you can win on the page after the click is untouched.
One more number sets the calendar. In Optmyzr’s comparison of 5,134 e-commerce accounts (December 2025), Black Friday 2025 produced 34% more conversion value than Cyber Monday 2025 ($144.8 million against $107.9 million), with higher ROAS on the Friday. For an online store, Friday is the event and Monday is the tail. Anything that has to be ready “for the weekend” has to be ready by the previous Monday.
Set it in October, leave it alone in sale week
Anything that forces a relearn gets done now, while a bad week is cheap. Anything that only moves money within a structure the system already understands can wait until the sale. One category of change should not happen in sale week at all. The table pulls the Optmyzr checklist (October 2026) and Google’s own guidance into those three columns.
| Campaign change | Do it now | Leave alone in sale week | Never in sale week |
|---|---|---|---|
| New campaigns, restructures, new ad groups | Launch now, let them settle | Yes: each forces a relearn | Launching a new campaign on the Wednesday |
| Bid strategy and targets (tCPA, tROAS) | Set the sale target now | Yes: target changes restart learning | Switching strategy because Thursday looked slow |
| Seasonality adjustments | Remove any scheduled ones | Off. Smart Bidding already manages this event | Adding one mid-sale to “catch up” on volume |
| Conversion tracking | One primary purchase conversion; GA4 for comparison only | Yes: a second primary counts each sale twice | Adding or editing conversion actions |
| Budgets | Raise caps where campaigns capped out profitably | Moves within existing campaigns are fine | Funding a campaign that lost money in October |
| Product structure | Split by discount, margin or stock, each with its own target | Shift spend between groups as stock runs out | Rebuilding the feed or the product groups |
| Alerts and rules | CPC +30% on the prior seven days; spend 20% ahead of pace | A person approves what the alert suggests | Letting a rule change bids without sign-off |
Two rows deserve a sentence each. On seasonality adjustments, Google’s own help page says they are “ideal for short events of 1 to 7 days” and should not be used for events that “Smart Bidding already manages”. They are for the flash sale Google could not have predicted, not the event it has watched for a decade. On budgets, the useful signal is impression share lost to budget on campaigns whose October results you would accept at sale-week prices. Everywhere else, extra money buys the same leak at a higher price.
- Launch new campaigns and restructures
- Set the sale targets and bid strategy
- Remove scheduled seasonality adjustments
- One primary purchase conversion
- Split products by margin or stock
- Landing page per ad promise
- Proof, offer and next step per visitor
- Control group live on the same traffic
- Raise caps where campaigns cap out profitably
- Shift spend as stock runs out
- Alerts: CPC +30%, spend 20% ahead of pace
- Compare the two halves of the same week
- Count each purchase once
- Keep what beat the control
Common mistake: adding a seasonality adjustment on the Wednesday because the forecast says volume will double. The forecast is right, and Smart Bidding already knows it. The adjustment makes the model bid for a lift it treats as exact; in Optmyzr’s example a predicted +50% against a real +40% overbids every click by 7.1%. If leadership wants volume over efficiency, lower the ROAS target in October and say what it costs.
Where the money goes: the page, not the bid
Pathmonk’s reference arithmetic shows it. You pay for 1,000 visits at $3.40 a click, which is $3,400. At the 3% conversion rate the industry treats as normal, 30 people buy, 970 leave, and $3,298 of the $3,400 was spent on the visitors who left. Sale week raises the price of every visitor who leaves; it does nothing to the number who leave. Add the 17% CPC inflation from the Optmyzr study and the same 1,000 clicks cost $3,980, of which about $3,861 goes on people who saw the page and left. WordStream’s 2026 benchmarks (13,474 US search campaigns, April 2025 to March 2026) put the median CPC at $4.14 for shopping, collectibles and gifts and $4.44 for apparel, fashion and jewelry, so most stores start above $3.40.
Now run the sale week at two conversion rates instead of two CPCs, because that is the lever you still have time to pull. The inputs are hypothetical; the structure is not. Revenue per paid visitor is conversion rate multiplied by average order value, and it decides whether a sale-week click pays for itself.
| 1,000 clicks at $3.98, order value $120 | 3.0% converts | 4.0% converts |
|---|---|---|
| Spend | $3,980 | $3,980 |
| Purchases | 30 | 40 |
| Cost per purchase | $132.67 | $99.50 |
| Revenue | $3,600 | $4,800 |
| Revenue per paid visitor | $3.60 | $4.80 |
| ROAS | 0.90 | 1.21 |
Compare what each lever costs. Buying 1,000 more clicks for 30 more purchases costs $3,980 at sale-week prices and is capped by the auction. Moving the page from 3% to 4% produces 10 more purchases per 1,000 clicks at no extra media cost, on every click for the rest of the season. That is why conversion rate sets CPA and ROAS, and why whether ad spend is fuelling growth or burning cash is usually settled on the site, not in the account. Ruler Analytics’ 2026 benchmark (May 2026, more than five million conversions) puts retail and e-commerce at 2.4%, so a store at 3% is already above average; the point it needs is not a benchmark gap but a page that keeps the promise of the ad.
Making the page match the ad for each visitor
In sale week the ad account does something the page cannot: it shows a different promise to every segment. The search ad says 30% off the winter range; the Shopping ad shows one product and one price; the remarketing ad recalls an abandoned cart. Then all three clicks land on the same page, with the same hero, proof and first button, and the page has to hope the visitor finds the promise they came for before they leave. At 3%, 970 of them do not.
The fix is not a landing page per ad; at sale-week volumes there are too many ads and the pages would be stale by Thursday. The fix is a page that opens on the promise the visit was made with, which is what Pathmonk does on the site while the campaign stays untouched in the ad account. One snippet goes in, with no developer and no redesign, and from then on the page is personalised in real time for every visitor by source: a visitor from a Google Ads campaign sees the headline, proof and next step that continue that ad, while a visitor from an AI answer or organic search sees the version that continues theirs. Elements change while the visitor browses, not after a test, and the engine decides per visitor and learns from every purchase instead of running a segment tree somebody maintains. Half the visitors see it and half never do, so the sale-week result is read against a control group on the same week’s traffic, not against last year. Personalisation only changes what the page shows, the way Pathmonk’s CRO and personalisation and its landing pages are built to work.
Two online stores on the Pathmonk homepage show the shape of the result. Joto Imaging Supplies, a B2B printing supplies store, recorded +73% sales in 2 weeks, through Black Friday. Porter Art Guild, an online art store with flat results from its Meta ads, recorded +24% conversions with Pathmonk on half its traffic; Pathmonk’s PPC case study lists the same store at +58% conversions from paid, on the same budget, over months 1 to 3. The two Porter figures cover different windows and are not to be added. What all of them share is the same visitors, the same ad spend, and a page that stopped showing everyone the same thing.
Search adWinter range: 30% off this week, free returns
Measuring the sale without fooling yourself
Three errors make most Black Friday reports unreadable, and all three are avoidable in October. The first is double counting: an account with two primary purchase conversions reports every sale twice, inflates the conversion rate Smart Bidding learns from, and overbids on it. Optmyzr’s checklist is explicit: one primary purchase source, GA4 for comparison only. The second is comparing against last year, when the auction, the product mix and the page were all different. The third is counting visits that were never people, which in a week of inflated CPCs costs twice: once for the click and once for the wrong conversion rate it feeds back to the bidding.
Sale week raises the price of every visitor who leaves; it does nothing to the number who leave.
The clean design runs anyway when the page adapts per visitor: half the sale-week visitors see the personalised page and half see the unchanged one, bot traffic is detected and excluded before either half is counted, and the two halves are compared on revenue per paid visitor over the same days. Pathmonk’s guarantee is written on that design: 20% more conversions against your own control group, measured as a live A/B test from month two, or your money back. For a sale week it means “did the site change work” has an answer by the Monday, from your own traffic.
Read the result as revenue per paid visitor on each half, not as conversion rate: a sale that lifts conversion rate by moving everyone to the cheapest discounted item can lower revenue per visitor, and only the control half shows which happened. Write the pass criterion before the Friday, and do not touch the bids while you wait for it.
When this advice does not apply
Below roughly 50,000 monthly visits, a 50/50 split of four days of paid traffic does not produce enough purchases to separate the halves from noise. Run the campaign checklist, land every ad on your single best page, and read the result over the whole season.
If the sale-week margin after discount and CPC inflation is below zero at your current conversion rate, no page change rescues the arithmetic in time. If the 4% column still shows a loss with your own order value, the decision is which products to advertise at all, the product-split row in the table, not a conversion problem.
Lead generation. In Optmyzr’s 2025 comparison, conversion value was flat between the Friday and the Monday across 14,463 lead-gen accounts. A B2B store with a sales-assisted purchase should treat the sale as a retail event for its self-serve range and leave the rest of the account on its normal calendar.
Key takeaways
- New campaigns, restructures and big target changes force Smart Bidding to relearn. Do them now, while a bad week is cheap, and lock them before the sale.
- Leave seasonality adjustments off for Black Friday. Smart Bidding already manages the event; in 2024 the accounts that added one paid 34% CPC inflation against 17% and lost ROAS while the others gained it.
- Black Friday produced 34% more conversion value than Cyber Monday for e-commerce in 2025. Friday is the event; be ready the Monday before.
- Sale week raises the cost of every visitor who leaves and does nothing to the number who leave. At 3%, 970 of 1,000 paid visitors still walk, each 17% more expensive.
- One point of conversion rate on the same clicks is worth more than any bid change you can make in November. Show the calculation with your own order value.
- The page should open on the promise of the ad that brought each visitor, per visitor, with the campaign untouched in the ad account.
- Count each purchase once, exclude bots before counting, and measure the sale as a 50/50 split of the same week’s traffic on revenue per paid visitor.
- Below the traffic floor or below zero margin, the weekend cannot prove anything; run the checklist and read the whole season.
FAQs on Black Friday Google Ads
Should I use a seasonality adjustment for Black Friday?
No, not for Black Friday or Cyber Monday. Google’s help page says adjustments are for short events Smart Bidding does not already anticipate, and Optmyzr’s study of 5,886 accounts found that in 2024 the accounts that added one saw CPCs rise 34% against 17% for those that did not, with ROAS down 15.7% against up 5.7%. Save adjustments for a flash sale or an email-only promotion Google could not have predicted.
When should I make my Black Friday campaign changes?
At least four weeks before the sale, and earlier if you can. New campaigns, restructures and target changes all start a learning period, and a learning period that overlaps sale week runs at the year’s highest CPCs. Set the targets you want for the sale now and leave them alone once it starts.
What should I change in Google Ads during the sale itself?
Budgets on campaigns that cap out while still profitable, spend shifts between product groups as stock runs out, and whatever your alerts flag for a person to approve. Nothing that changes the bid strategy, the targets or the structure.
Is Black Friday or Cyber Monday bigger for an online store?
Black Friday. Across 5,134 e-commerce accounts in Optmyzr’s December 2025 comparison, Black Friday 2025 produced 34% more conversion value than Cyber Monday with a higher ROAS. For lead generation the two days were flat.
Why does conversion rate matter more in sale week than usual?
Because every visitor who leaves cost more to bring. At a 17% CPC rise, 1,000 clicks that cost $3,400 in October cost $3,980 in sale week, and at 3% the same 970 visitors leave. A point of conversion rate on those clicks is worth more than it is in any other month, and it is the one lever that does not restart Smart Bidding’s learning.
Do I need a separate landing page for every Black Friday ad?
No. At sale-week volumes there are too many ads and the pages would be out of date within days. The alternative is a page that adapts per visitor to the source and the promise of the ad that brought them, with the campaign left untouched in the ad account.
How do I know whether a site change worked in sale week?
Split the sale-week traffic 50/50, exclude bot traffic before counting, count each purchase once, and compare the two halves on revenue per paid visitor over the same days. A comparison against last year or against the week before cannot separate the page change from the auction, the discount and the product mix.
- Optmyzr, “We analyzed 5,000+ Google Ads accounts: seasonality adjustments backfire on Black Friday”, 15 October 2026 (martech.org; also published on searchengineland.com)
- Optmyzr, three-year BFCM seasonality adjustments study, 19 November 2025 (optmyzr.com)
- Optmyzr, Black Friday vs Cyber Monday Google Ads 2025, 16 December 2025 (optmyzr.com)
- Google Ads Help, “About seasonality adjustments” (support.google.com)
- WordStream, Google Ads benchmarks 2026, updated 16 September 2026 (wordstream.com)
- Ruler Analytics, conversion rate by industry 2026, 26 May 2026 (ruleranalytics.com)
- Pathmonk homepage, October 2026 (reference arithmetic, Joto Imaging Supplies, Porter Art Guild, guarantee) (pathmonk.com)
- Pathmonk, PPC results case study (Porter Art Guild +58%) (pathmonk.com)