Performance Max in Q4: What to Change and What to Leave Alone

Performance Max is the campaign type most likely to be blamed for a disappointing Q4 and least likely to be the actual cause. Its performance is largely decided by three things you control outside the campaign: your product feed, your asset variety, and how the campaign is structured around your margin.

All three need to be right before peak volume arrives. This post covers what to fix in the next fortnight, what can safely be adjusted while peak is running, and what should not be touched at all once budgets rise.

Why does the feed matter more than the settings?

Because Performance Max is largely a Shopping campaign with extra placements, and Shopping is decided by product data. Titles, attributes, images and availability do more for delivery than any bid or budget change.

Three feed jobs are worth doing before the end of October. Titles that lead with the terms a buyer would actually type rather than with your brand name. Complete attributes, particularly size, colour, material, age group and gender, since these are what the system matches queries against. And accurate availability, because paying for placement on an out of stock product is the quietest waste in the account.

That same product data increasingly decides whether you appear in AI assisted shopping surfaces too, so the work pays in more than one place.

How should the account be structured for peak?

By margin and by role, rather than by product category.

The useful split separates products you are happy to discount from products you are not, and separates your genuine best sellers from the long tail. Grouping by margin lets you set different targets against genuinely different economics, which a single campaign covering the whole catalogue cannot do.

Keep the number of campaigns low. Splitting a catalogue across many asset groups fragments the data each one learns from, and during a six week window there is no time to recover from that.

What should be finished before the freeze?

Everything structural. New campaigns, new asset groups, conversion setup changes, value rules and feed rebuilds all belong before the site and account freeze two weeks out from Black Friday.

The reason is the learning period. A campaign with material changes needs roughly a fortnight to settle, which means a change made in mid November is still settling during the days that matter most. Across our portfolio the operating rule is no cold launches in Q4, and Performance Max is the campaign type where that rule is broken most often, usually with good intentions.

The full freeze sequence is in the peak season freeze calendar.

What can you safely change during peak?

Budgets, targets within limits, and creative assets.

Budget increases are safe when they are gradual. Our own operating rule is scaling by no more than 30% in any seven day period, because larger jumps reset the campaign's learning at the worst possible moment.

Target return or target cost per acquisition can be adjusted, but treat each adjustment as a small change rather than a correction. Moving a target sharply is functionally a relaunch.

Adding new creative assets to an existing asset group is the safest change available and usually the most useful, since fatigue is real across a six week window and fresh assets do not carry the same reset risk as structural edits.

How do you change an offer without resetting learning?

Through the feed rather than through the campaign. Promotional pricing, sale price fields and catalogue level overlays change what the ad says while leaving the campaign structure untouched.

This is the single most useful technique in a promotional quarter, because the alternative, building a new campaign for each offer, forfeits everything the account has learned at the point of maximum spend.

What should you not touch?

Conversion tracking, attribution settings, and the conversion actions the campaign optimises toward, once peak is running.

Changing any of these mid season breaks comparability with your own baseline, which means you lose the ability to tell whether performance actually moved or only the measurement did. Verify all of it in October and then leave it alone.

Audience signals are worth loosening early rather than tightening late. Narrow demographic guardrails cost reach in the weeks when reach is cheapest, and our practice is to loosen them from early October through to mid January.

How should you judge it at peak?

On new customer cost and contribution, not on the return the campaign reports.

Performance Max reports confidently and includes brand search within its attributed revenue unless you have deliberately separated it. A campaign showing a strong return while your blended numbers stay flat is usually harvesting demand other activity created.

The wider commercial frame, including the maximum acquisition cost to set before November and the portfolio cost curve peak follows, is in the Q4 Profit Playbook, free and written for founder led DTC brands on Shopify at $5M to $30M.

Webtopia runs Google and Meta together against contribution margin for founder led brands as an ecommerce paid media agency.

Frequently asked questions

Should I create new Performance Max campaigns for Black Friday?

No. Use feed level promotional pricing and catalogue overlays to change the offer instead, so the campaign keeps everything it has learned.

How much can I raise a Performance Max budget at once?

Gradually. Our operating rule is no more than 30% in any seven day period, because larger increases reset learning during the weeks that matter most.

When should Performance Max changes be finished before peak?

Structural changes should be done two weeks out from Black Friday at the latest, since a campaign needs roughly a fortnight to settle after a material edit.

Why is my Performance Max return higher than my blended numbers?

It often includes brand search unless you have separated it, so the campaign can report strong performance while harvesting demand created elsewhere.

What is the highest impact Performance Max fix before Q4?

The product feed. Titles led by buying terms, complete attributes and accurate availability do more for delivery than any bid or budget adjustment.