What to fix before Q4, and what to leave alone

There is no shortage of Q4 checklists. Fix your tracking, tighten your negatives, audit your feed, check your site speed, plan your budget early, get your creative ready. It is all sound advice and we would give you most of it ourselves.

What almost nobody writes is the other half of the sentence.

Because the expensive mistakes in Q4 are rarely things people failed to fix. They are things people fixed at the wrong moment, in the wrong week, on a campaign that was already working.

Why this matters

Every meaningful change you make to a campaign has a cost, and the cost is not the change itself. It is the time the platform needs to relearn.

Edit an ad while it is running and it re-enters the learning phase, which slows optimisation while the algorithm reworks out who to show it to (Mabo). Scale the budget too quickly and you trigger the same reset, driving your cost per acquisition up at exactly the moment you were trying to buy more volume (Evoke Creative).

In March that is an irritation. In the third week of November it is the quarter.

This is what makes Q4 different. Not that the stakes are higher, though they are. It is that the recovery time you normally rely on has gone. You cannot make a change on the twentieth, discover it was wrong on the twenty seventh, and fix it in December, because nearly all of December's shopping happens in the first two weeks (Mabo).

Everything is also more expensive and slower while you do it. CPMs rise across the quarter and ad approval times get longer, which is why the advice is to build and schedule campaigns at least five days before they need to be live (Evoluted).

So the real question in September is not what should I improve. It is what am I still allowed to touch.

Worth fixing now, while a mistake is still cheap

Tracking, first and without argument. This is where the largest recoverable losses sit. One agency reports incoming accounts leaking 25 to 30% of spend to default settings, automation drift and broken tracking, with a single client recovering $20,000 a month from one tag fix (Go Fish Digital). Conversions API alone can improve cost per acquisition by 13% or more (Rocketer).

The feed, if you sell products. On Shopping and Performance Max there are no keywords, so the feed is what tells Google which searches you belong in. A vague title in September is a product that never appears in November.

Anything structural. Consolidation, new campaign types, bidding changes, a fresh conversion action. All of them reset learning, so do them now while there is time to be wrong and recover. Meta's shopping campaigns want at least seven days without changes to stabilise, and you want that week spent in September rather than in the run-up to Black Friday (Evoke Creative).

Your creative bank. Test hooks and formats in September and October, then move only the proven winners into the peak, so you are not paying November prices for unproven visuals (Evoke Creative).

Budget headroom, decided in advance. For ecommerce the guidance is holding roughly 40% of the annual budget for Q4 (Rocketer). Decide the number and the scaling rule now, because a sensible pace is something like 10 to 20% every 48 hours while performance holds (Evoke Creative), and that is not a decision anyone makes well at 9pm on Black Friday.

Leave alone

Anything currently working. The hardest discipline in the account and the one that costs the most. There is always a tempting optimisation on a campaign that is performing, and in Q4 the downside of being wrong is far larger than the upside of being slightly right.

Audience testing during the peak. Black Friday is not the time for it (Evoluted). It is the single most expensive week of the year to be learning something you could have learned in September for a fraction of the price.

Structural changes after roughly mid-October. Pick a date, write it down, tell whoever else touches the account. After that you change budgets and creative, nothing else.

Your attribution settings. Change a window mid-quarter and you can no longer compare November to October. You will spend January working out whether performance moved or the measurement did.

Very short bursts. One-day offers limit reach and give the algorithm no time to optimise (Evoluted), which is why longer runs generally outperform them (Mabo).

What people overestimate

Doing more. The instinct when performance wobbles in November is to intervene, and intervention is usually what turns a soft week into a bad month. Sometimes the right action in Q4 is to look at the account, decide it is fine, and close the laptop. That is a real decision, and it is much harder than making a change.

Budget as the lever. More money into a fatigued campaign buys more repeats to the same people, and scaling too fast resets learning anyway (Evoke Creative). Q4 is usually constrained by creative and by margin, not by budget.

October and November as the months that matter. September is the month that decides Q4. It is the last point at which you can make a structural change, be wrong, and still recover. Everything after that is execution.

Discounting as the whole strategy. It matters, 80% of buyers say discount offers influence their purchase and 86% are open to switching brands (Rocketer), and in Australia nearly 44% now consider cheaper or private-label alternatives (Elephant in the Boardroom). But a discount on top of broken tracking is just a cheaper way to lose money.

The end of December. Most businesses stop thinking after Christmas. Roughly 41% of holiday shoppers are still active in the period between Christmas and New Year, and CPMs drop by about 16% in that window (Rocketer). It is the cheapest attention of the quarter and almost nobody is bidding for it.

Their own discipline. Nearly everyone agrees with a freeze in principle. Almost nobody holds it, because the fifteenth of November is exactly when the pressure to do something becomes hardest to resist.

One more thing, if you are selling in Australia

Most Q4 advice is written for the northern hemisphere, and it quietly assumes a cold, indoor, gift-led quarter. Ours is not that. We are heading into summer, holidays and heatwaves at the same time as the retail peak (Elephant in the Boardroom).

Two practical consequences. Melbourne Cup, the first Tuesday of November, is a genuine commercial moment here and appears in no international playbook. And Boxing Day is one of the biggest shopping days of the Australian year, which means the quarter does not end on the twenty fifth the way most imported calendars imply (Elephant in the Boardroom).

If your plan came from a US template, it is probably one date short at the front and a fortnight short at the back.

What to do with the next three weeks

Give yourself a short list and a hard date.

Between now and the end of September, fix the tracking, clean the feed, make any structural change you have been putting off, and put the page and the checkout through a proper test at volume. Test your creative now so the peak only ever sees proven winners. Write down the budget number and the scaling rule while nothing is on fire.

Then set your freeze date, tell everyone who touches the account, and hold it. From that day you change two things only, budget and creative. Everything else waits for January, and January is closer than it feels.

The businesses that have a good Q4 are rarely the ones that optimised hardest during it. They are the ones that arrived in October with nothing left to fix, and then had the discipline to leave a working account alone while everyone around them fiddled.

At Dadek Digital we run exactly this review with businesses in September, so the account goes into the quarter with the fixes done and a clear line about what is off limits. If you would like a straight read on what is worth changing before October and what should not be touched, we are happy to take a look.