Black Friday Google Ads 2026: 8 Checks Before You Scale

Alina Popa in a MeasureCamp London hoodie working on a laptop, framed by the words Black Friday

In Romania, Black Friday doesn’t wait for the last Friday of November. eMAG, the country’s largest online retailer, runs its Black Friday on November 6, 2026, three weeks before the US date, and most Romanian stores plan their own sales around it. So by the time the international Black Friday guides come out, a lot of us are already in the middle of it, coffee in one hand and an impression share report in the other.

We’ve run Black Friday campaigns for eCommerce clients for years, and the results have been good. What hasn’t stayed the same is Google Ads. Every year there’s something new to understand before the sale: this year it’s seasonal budget adjustments, Promotion mode in Search and Performance Max, Smart Bidding Exploration, and new ad channels competing for the same budget. A plan that worked two years ago needs at least a second look.

These are the checks I go through before I let an account scale for Black Friday. Most of them have very little to do with bidding, which is a bit disappointing if you came here for a secret bid hack. There isn’t one. Sorry.

If you only want the list, you can get the printable checklist at the end. The article is the reasoning behind it.

1. Rebuild last year’s sale from the data

Before touching anything, I rebuild last year’s Black Friday from the account data, day by day and campaign by campaign. I’m looking for a few specific things:

  • How much budget the campaigns created just for Black Friday took, compared to the value they brought. The campaigns with history usually do the heavy lifting. New “Black Friday” campaigns tend to take a big share of the budget and bring back much less, because they start the busiest week of the year with no data.
  • Which days ran out of budget. Look at impression share lost to budget, day by day. If the best day of the sale was capped, that’s the first thing to fix.
  • Whether the highest spend went to the best days. It’s surprisingly common for the last day of the sale to spend the most, at the lowest ROAS, because budgets were raised in a hurry after a strong Friday. Sunday is the guest who arrives last and still eats most of the cake.
  • The pre-sale dip. Sales usually drop in the days before Black Friday, because people wait for the discount. You don’t want to read that dip as a performance problem, or spend extra into it.
  • Brand search. If it was capped by budget, you lost the cheapest, highest-intent clicks in the account.

None of this is exotic. But these are the account’s own numbers, and they make every discussion with the client (and with myself) a lot shorter.

2. Plan the total budget across channels, not just Google Ads

Black Friday budgets rarely live in one platform. Most of the eCommerce accounts we work with run Google Ads, Meta and TikTok at the same time, and this year ChatGPT ads have joined the list. Each platform gets its own budget and reports its own ROAS, and if you plan them separately, two things go wrong.

The first is that everyone claims the same orders. Meta, TikTok and Google Ads each attribute a share of the sales to themselves. Add those shares up and you often get more than 100% of the store’s revenue. It’s a group project where every member did 70% of the work. So I plan the total first, against one shared number: total store revenue divided by total ad spend across all channels (MER, or blended ROAS). That target comes from the margin after the discount (see section 4), and each channel’s budget has to fit inside it.

The second is that the channels depend on each other. Meta and TikTok create demand, and a lot of that demand ends up as a branded search on Google. A newsletter or an SMS campaign does the same. If brand search is capped by budget at the moment the SMS goes out, you pay for the attention on one platform and lose the sale on another.

One budget, five channels: how I read each one during Black Friday

ChannelIts job during the saleHow I read its numbers
Google Search (brand)Catches the demand every other channel createsBudget should never be its limit. Cheap clicks, high intent, though some of those orders would come anyway
Shopping / Performance MaxCatches product searches, including competitors’ customersSame-day ROAS is incomplete. Watch orders and impression share lost to budget
MetaCreates and re-activates demand, before and during the saleGenerous attribution. Useful for decisions inside Meta, not for the total
TikTokReach and attention, mostly creative-drivenJudged on blended numbers, not on platform ROAS alone
ChatGPT adsNew channel, little history in most accountsA small, fixed test budget. Never money you can’t afford to lose
The shared number is store revenue ÷ total ad spend. It won’t tell you which channel caused which sale, but no platform can claim it twice. Each platform’s own report is for optimising inside that platform.

In practice, this means agreeing on three things with the client before the sale: the total budget for all channels, the blended target that budget has to hit, and the timing of every email, SMS and social push, so the Google Ads budget is there when the searches come.

3. Check whether you can trust the numbers you’ll be reacting to

Most Black Friday guides skip this part. It’s where I spend the most time. During the sale you make budget decisions every few hours, so you need to know which numbers you can react to and which ones are still filling in.

What I check:

  • Which conversion actions each campaign actually bids on. For an online store, that’s normally the purchase, with the right value. Check the account’s primary goals, and then the goal settings of each campaign, because a campaign can use its own. If a newsletter signup or an add-to-cart sneaks in, Smart Bidding will happily optimise for it during your biggest weekend. I’ve seen this more often than I’d like (more on that in Your Tracking Isn’t Failing. It’s Misleading You.).
  • The attribution lag. With a long click window, the last days are always under-reported, and same-day ROAS in Google Ads is incomplete by design. During the sale, my real-time signal is the orders in the store, not the ROAS column. GA4 realtime is good for a quick sanity check (is the purchase tag still firing?), but with consent mode in the mix it doesn’t see every order, so it doesn’t decide budgets.
  • How much each platform claims, compared with the store’s revenue (section 2).
  • The purchase tag, tested end to end before the sale, including an order with a discount code. Black Friday is when checkout changes, new discount rules and new apps get added in a hurry, and when tracking quietly breaks.

If you only have time for one check, make it this one. Scaling on numbers you haven’t validated is a fast way to spend more and learn nothing.

4. Calculate break-even ROAS after the discount

Your normal break-even ROAS doesn’t apply during Black Friday. The discount comes straight out of the margin. Your accountant knows this. Smart Bidding doesn’t.

If your gross margin is m and the discount is D, the margin during the sale becomes:

margin_after_discount = (m − D) / (1 − D)
break_even_ROAS = 1 / margin_after_discount

For a few common combinations:

Break-even ROAS after the discount

Normal margin−10%−15%−20%−25%
30%4.55.78.015.0
20%9.017.0lossloss
10–15%18+losslossloss
Same discount, very different consequences. A ROAS of 8 is fine for one product line and a loss for another.

The table is optimistic. The margin doesn’t include shipping, payment fees, returns or cash on delivery (still very common in Romania), so the real break-even is higher.

It also means some products shouldn’t get Black Friday budget at all. If a 20% discount loses money on every sale of a low-margin product line, a good-looking ROAS won’t fix it. Whether to leave those products out of the sale or keep them at full price is a business decision for the client, not something to solve with bidding settings.

5. Freeze the structure early, and clean up what could wake up

About two weeks before the sale, I freeze the account structure. After that date we don’t add campaigns, change targets or touch listing groups. Budgets, assets and promotions are the only things that still move.

The most common test of that rule is a new campaign idea a week before the sale, usually Demand Gen “to warm up the audience”. I say no to that, for four reasons:

  1. Demand Gen needs a couple of weeks and a decent volume of conversions to get out of the learning phase. It reaches the peak days still guessing.
  2. The warm-up lands in the pre-sale dip, when people are waiting for the discount anyway.
  3. You can’t measure it. With a long click window, you won’t know anything reliable until well into December.
  4. In most accounts, prospecting is already Meta’s and TikTok’s job.

The idea isn’t bad, just late, like most ideas that start with “what if we quickly launch…”. Launched in mid-November, the same campaign has three or four weeks to learn before the Christmas gift-shopping peak.

The freeze is also a good moment for cleanup. Accounts collect old campaigns, and old campaigns sometimes come back to life:

  • Last year’s Black Friday campaigns that are still enabled and only held back by an end date. One edit to that date and they run with last year’s offer.
  • Seasonal campaigns from earlier in the year, enabled, with a budget and zero spend. Valentine’s Day is a classic. Nothing says Black Friday like a heart-shaped headline.
  • Campaigns below break-even even without a discount. Before pausing a Shopping campaign, I check whether its products would disappear from Shopping. Product-level data from shopping_performance_view shows which products are also serving in Performance Max, and usually most of them are.

The feed gets a freeze date too

In Shopping and Performance Max, the feed is most of the ad, and a lot of Black Friday problems start there. Before the freeze, I check:

  • sale_price with sale_price_effective_date, so the discounted price appears when the sale starts and disappears when it ends. The price on the product page has to match the feed the whole time, or products get disapproved for price mismatch exactly when you need them.
  • How often stock syncs. If the feed updates once a night, a product that sold out at 10:00 keeps collecting clicks until tomorrow.
  • The product IDs. A new ID is a new product for Google, with no history, the week before the sale. Feed plugins sometimes change them after an update, so compare a few with last month’s.
  • Variants grouped with item_group_id, each with its own price, availability and image.
  • Diagnostics in Merchant Center, a week before and again the day before. Disapprovals take time to fix and more time to get reviewed again.

6. Find out what’s limiting the campaigns before you touch bidding

The classic Black Friday advice is to lower your ROAS target so Smart Bidding can enter more auctions. Andrew Lolk’s Black Friday playbook at Savvy Revenue argues this well, for accounts running a full Black Week or Black Month in late November.

Before I do that, I check what’s actually holding the campaigns back. It can be the budget, the target, the feed, the assets or simply the demand, and often more than one at a time, so I look at several signals together: the “Limited by budget” status, the impression share lost to budget, the hour when the daily budget runs out, the ROAS against the target. When the budget is clearly the binding limit, a lower target mostly makes the same clicks more expensive. And changing the target right before the peak gives Smart Bidding one more thing to recalibrate, with no time to evaluate the change. In that case the target stays, and the weekend runs on budgets (next section) and on seasonality adjustments.

Google recommends seasonality adjustments for short events of 1–7 days where you expect a major change in conversion rate. This is how I set them:

How I set conversion-rate seasonality adjustments

  1. Measure last year’s lift. Conversion rate on each day of the sale, divided by the conversion rate in the 2–4 weeks before it. Only the conversion rate: the adjustment doesn’t cover a bigger or smaller basket.
  2. Set this year’s values below that. Too high and Smart Bidding overpays for clicks. Too low and it just reacts a bit later.
  3. Split by day only when the days really differ. If last year’s Friday jumped far more than the Sunday, they get separate adjustments. If not, one for the whole sale is simpler.
  4. End each one at 00:00 the next day, not 23:00, or the last hour of every day runs without it.
  5. Only the campaigns bidding on purchases. Not the ones optimising for something else.
The scope is set per campaign. Check it in the UI after saving, it’s easy to miss one.

What I don’t use them for

A common piece of Black Friday advice is to add a seasonality adjustment when “Google isn’t spending enough of your budget”. I disagree. The adjustment tells Smart Bidding the conversion rate will be higher. If that isn’t true, you pay more per click for the same traffic and the ROAS drops. Google’s own guidance is to use them only when you expect major changes to conversion rates. If the campaigns aren’t spending, look at the budget, the target and the max CPC limits first.

I’d also skip “update your ROAS targets the night before”. Change the target and add a seasonality adjustment on the same night, and on Friday afternoon you have no idea which of the two you’re looking at.

Then there’s Promotion mode, a beta in Search and Performance Max. For a period of 3 to 14 days, it relaxes the ROAS target (Google calls it ROAS tolerance) and can add extra daily budget, then everything goes back on its own. On days when the discount has already pushed the break-even ROAS up, relaxing the target goes in the wrong direction. I’d test it on a smaller promotion first. Four days of peak traffic is a bad moment to find out how a beta feature behaves.

7. Plan budgets by day, front-load the best day, and keep a reserve

Once the total is agreed (section 2), the Google Ads share gets split by day, and the split comes from last year’s numbers (section 1). The best day of last year’s sale gets the biggest budget, before the sale starts, not as a reaction on the day.

Example: a four-day sale, Thursday to Sunday

  • Thursday: 22% early access or soft start
  • Friday: 38% the biggest day gets the biggest budget
  • Saturday: 24%
  • Sunday: 16% no top-ups from the reserve
  • Reserve: 15–20% on top only used by rules written before the sale
Illustrative split. Yours should come from last year’s data. Bars are scaled to Friday.

The reserve only moves under rules written down in advance. The ones I use most:

  • If brand search loses more than 10% impression share to budget, its budget goes up.
  • If Performance Max has spent most of its daily budget by mid-afternoon and the store’s orders look good, it gets a top-up, on the first days of the sale only.
  • The last day gets nothing from the reserve. If its spend passes the best day’s, all increases stop.

Automating the night shift

Budgets have to change every night of the sale, and nobody on our team wants to be in the Google Ads UI at 23:30. We like our clients. We also like sleeping.

Google Ads now has seasonal budget adjustments, next to the conversion-rate ones. They’re worth a look, but they only work for Search and Shopping campaigns (not Performance Max), last between 3 and 14 days, are set per campaign, and add a single amount to the average daily budget for the whole period. For a long sale, that’s fine. For a short one, you can’t give one day more than another.

When the split by day matters, we use a small Google Ads script instead. It holds an “evening → tomorrow’s budget” schedule per campaign and runs daily at 11 PM:

// Illustrative values
var SCHEDULE = {
  '2026-11-04': { 'Search_Brand': 100, 'PMax_Main': 200 },
  '2026-11-05': { 'Search_Brand': 150, 'PMax_Main': 300 }
  // ... one entry per evening, the last one resets to normal budgets
};

The code is the easy part. Two details matter more:

  • Scheduled scripts don’t start at an exact minute, and sometimes they start late. If a run slips past midnight, the script should still apply the previous evening’s budgets instead of skipping a day.
  • It needs a test date. In Preview you set it to one of the evenings in the schedule and read the log (old → new for each campaign) without changing anything in the account. Then you clear it, save, and schedule it.

Black Friday ads work the same way. We create them days in advance and leave them paused, so they go through ad review before the sale and not during it. One automated rule enables them when the sale starts, another pauses them when it ends.

The boring checks

None of these will make a nice slide, but any of them can quietly ruin the weekend:

  • The account spending limit and the payment method, on every platform. If an account has a budget or credit limit, add up the expected spend for November, sale included. It takes 30 seconds to check.
  • Promotion assets in Google Ads, scheduled to the exact sale dates, so the discount doesn’t show up before it exists or after it ends. The same goes for Merchant Center promotions, if your country has them. Romania isn’t on the list yet, so here the discount lives in the feed and in the ads.
  • A change log. Every change made for the sale goes into it, with the date. After the sale, it’s what you use to undo everything and to explain what happened.
  • One channel with the client during the sale, plus a few answers before it starts: the discount per product line, the timing of every email and SMS, and stock for the best-selling products.

8. Decide in advance how you’ll behave during the sale, and after

Most bad Black Friday decisions get made at 15:00 on Friday, while staring at an incomplete ROAS column. That’s why the rules are written before the sale:

  • Check-ins happen at fixed times: a few times a day on the main days, twice a day on the weekend. Not every 20 minutes.
  • The store’s orders come first, impression share lost to budget second. GA4 realtime only confirms the tracking still works. Same-day Google Ads ROAS doesn’t get a vote.
  • When a product sells out, the feed has to say out_of_stock quickly, so you don’t pay for clicks on something nobody can buy. That’s usually enough, and the product can stay in the feed. If the feed updates too slowly for a fast-selling product, exclude it from the listing group until the stock is back.
  • Every automation gets a human check each morning: did the budget script run, do the budgets match the plan, did the Black Friday ads go live.
  • After the sale, budgets go back to normal the night it ends, the sale ads and assets stop, and nothing gets evaluated for at least 14 days. With a long attribution window, a verdict on Monday morning is mostly noise.

On negative seasonality adjustments after the sale, Google’s documentation says “no negative adjustment is needed when the promotion is over”, while some practitioners use them anyway to stop Smart Bidding from overspending in the days after. For a short sale, I’d rather watch CPCs for a few days and react if I need to. For a full Black Week or Black Month, the argument for a planned cooldown gets a lot stronger.

Where the work actually goes

Reading this list back, I notice how little of it is about bidding. Most of the time goes into the data, the margins, the budget across channels and the rules for the weekend. The bidding settings come last, and once everything else is in place, they’re the easiest decisions to make.

Get the checklist

I turned these notes into a two-page checklist you can print, tick off and stick next to your monitor before the next sale. Leave your email and the download link shows up right away. No waiting, no “check your inbox in 24 hours”.

If you tick the box, you’ll also get my occasional emails with practical notes on Google Ads, tracking and measurement. Occasional really means occasional: I’m usually busy checking someone’s budgets. You can unsubscribe anytime, and the details are in the Privacy Policy.

Stay curious.

Found this useful? Share it.

Similar Posts