There is a thing that happens every year around the second week of September.

An owner looks at the calendar, realizes Q4 is three weeks out, and decides the move is to add budget. More spend, more leads, more revenue. Simple math.

Then October hits. The leads cost forty percent more than they did in August. The ad account eats the extra money in about nine days. And by Halloween that owner is sitting there wondering why the biggest quarter of the year feels like running uphill in sand.

Here is what actually happened. They poured more fuel into a leaky tank.

Q4 does not make your advertising worse. It makes your advertising honest. Every inefficiency you have been carrying since spring gets repriced at holiday rates, and the waste you could afford to ignore in July starts costing you real money in November.

So this week, before the auction turns ugly, you are going to do the opposite of what everybody else does. You are going to subtract.

A few numbers so you know I am not being dramatic.

Average Meta CPMs in 2026 are running around thirteen and a half dollars, roughly twenty percent above where they sat last year. That is the baseline before the holidays even start. Then Q4 stacks on top of it. Depending on your category, October through December CPMs typically run thirty to eighty percent above the rest of the year, and the week around Black Friday can double or triple.

Google search tells the same story. Cost per click climbs in peak season because every retailer with a budget shows up to the same auction you have been quietly winning all year.

None of this means you should sit out Q4. Plenty of service businesses do their best work in Q4 precisely because their competitors get nervous and pull back.

It means you should stop paying rising prices for things that were never working.

Because here is the part that gets people. Waste is not a fixed cost. If thirty percent of your spend is going to campaigns that have never produced a paying customer, and the price of impressions goes up fifty percent, your waste goes up fifty percent too. You do not just pay more for the good stuff. You pay more for the garbage.

Ninety minutes fixes most of it. Seven passes. Go.

Open your ads manager. First thing, set the date range to the last ninety days. Not seven. Not fourteen.

The seven day view is where good businesses go to die. It is noisy, it is emotional, and it makes you kill things that were fine and protect things that were not. Ninety days gives you enough volume to see the truth.

Now build a column set that tells you something. Amount spent. Results. Cost per result. And then, and this is the part almost nobody does, sit down with your CRM open next to it and mark which campaigns actually produced closed revenue.

Leads are not results. Booked calls are not results. Money is the result.

You are going to find campaigns with beautiful cost per lead numbers that have never produced a single customer. You are also going to find one that looks expensive on the surface and quietly pays for the whole operation. That is normal. That is why we do this.

If you cannot connect ad spend to closed revenue at all, stop reading and go fix that first. Everything else in this article is guessing until you can answer the question of which campaign paid for itself.

Filter for ad sets with spend greater than three times your target cost per acquisition and zero conversions in the last thirty days.

Turn them off. All of them. Right now.

I know. You have a story about that one. It was working in April. The creative just needs a refresh. The audience is really good, it just needs more time.

No. If something has burned three times what a customer is worth to you and produced nothing in a month, it is not a slow starter. It is a hole. And in about two weeks the hole is going to get fifty percent more expensive.

The average account I open has somewhere between fifteen and thirty percent of monthly spend sitting in this bucket. That is not a rounding error. On five thousand a month, that is nine hundred dollars leaving the building every month with nothing to show for it.

This one is the quiet killer, and it got worse in 2026.

Look at your ad set structure. How many ad sets are targeting broad or near broad audiences in the same geography with overlapping creative?

If the answer is more than two or three, you are running an auction against yourself. Meta is not sorting that out for you. You are competing with your own dollars, driving up your own costs, and splitting your conversion data across so many ad sets that none of them ever exit the learning phase.

The fix is consolidation, and it feels wrong the first time you do it. Collapse eight thin ad sets into two or three funded ones. Fewer containers, more budget in each, more conversion signal per container.

The rule I use is simple. Every ad set should be able to generate at least twenty five to fifty conversions a week at your current budget. If it cannot, it does not deserve to exist as its own ad set. Merge it or kill it.

This single move has taken twenty percent off cost per acquisition in accounts where I changed nothing else. No new creative. No new offer. Just less self inflicted competition.

Add frequency as a column. Break it down by ad, thirty day window.

On a cold prospecting audience, frequency above two and a half in thirty days means the same people are seeing the same ad over and over and have already decided. Performance is going to decay whether you notice it or not. In Q4 that decay accelerates, because everybody else is also shouting at your audience.

While you are in there, turn on the ad relevance diagnostics columns. Quality ranking, engagement rate ranking, conversion rate ranking. If an ad is below average on two of the three, the platform is charging you a tax for running a mediocre ad. You do not fix that with better targeting. You fix that with a better ad.

Which brings up the thing you should be doing this week instead of worrying about the auction. The single biggest lever on Q4 cost is creative volume and creative variety. The algorithm rewards diversity now. It wants different angles, different formats, different hooks, and it will find cheaper inventory for the ones that earn attention.

So block a half day before October first and produce five new concepts. Not five variations of the same ad with a different background color. Five actually different angles. A customer story. A price and process explainer. A common objection handled head on. A before and after. A straight to camera take where you say the thing your competitors will not say.

Batch it once. Feed it all quarter.

Two breakdowns, five minutes each, and they usually pay for the whole exercise.

First, break down delivery by placement. You are looking for spend that landed somewhere you did not intend. Audience Network and the odd corner placements have a habit of quietly absorbing budget while producing clicks that never turn into anything. If a placement has meaningful spend and no conversions across ninety days, exclude it.

Second, break down by region. This one is embarrassing when you find it, and you usually find it. Service businesses routinely pay to advertise thirty miles outside the radius they will actually drive. I have seen a home services company in one metro spending eleven percent of budget in a state they do not operate in, for four months, because a radius setting got fat during a campaign duplication.

Tighten the radius to where you actually make money, not where you would theoretically say yes.

If you run Google, this is the highest dollar per minute work in the whole audit.

Pull the search terms report for the last ninety days. Sort by cost, descending. Filter to conversions equal to zero.

Read the list. Actually read it. You will find people searching for jobs at companies like yours, people searching for free versions of what you sell, people searching for your competitor by name, and people searching for something adjacent that has nothing to do with your business.

Every one of those gets added as a negative keyword. Then build that into a shared negative list so it applies across campaigns and you never pay for it again.

Broad match is the usual culprit here. It has gotten smarter, and it has also gotten more expensive to leave unsupervised. If you have broad match running without a serious negative list underneath it, you are funding a research project.

Half the reason this audit takes ninety minutes instead of twenty is that nobody can tell what anything is.

Campaign names like "New Campaign Copy 3" and "Test 2 FINAL" are how accounts become archaeology. Set a convention and enforce it. Mine looks like this at the campaign level: priority number, brand or client code, objective, offer. Ad sets carry the launch date, the audience, and the copy variant. Ads carry the copy variant and creative variant.

The specific format matters less than the fact that you have one. What you want is the ability to open the account cold, sort by name, and immediately know what is running, when it launched, and which variable it is testing. That turns a quarterly archaeology dig into a Monday morning glance.

You are going to free up somewhere between fifteen and thirty percent of your monthly ad spend doing this. Do not let it evaporate into the general fund.

Put it in three places, in this order.

First, the winner. Whatever campaign your ninety day revenue view identified as the one that actually pays, feed it. Increase budget in twenty percent steps every three or four days so you do not blow up the learning.

Second, retargeting and your own list. This is the part that gets forgotten in Q4 planning and it is the cheapest inventory you own. People who already visited your site, already opened your emails, already talked to you and went quiet. Those impressions cost a fraction of cold prospecting and they convert several times better. Advertisers with strong first party data consistently pay meaningfully less than advertisers relying entirely on the platform to find people.

Third, the plumbing. If your leads are sitting in a form inbox for six hours before anyone touches them, no amount of auction optimization saves you. Wire the form to your CRM so a lead triggers a text inside five minutes. I use Make.com to move data between systems that were never designed to talk, and Go High Level to hold the pipeline and fire the follow up. If your sales calls are where deals go to get forgotten, put Fathom on them so you stop relying on memory.

A five minute response time on a lead you already paid for is worth more than any bid strategy you will ever choose.

Most owners walk into Q4 with the same account they had in March, a bigger budget, and a lot of hope.

The ones who do well walk in with a smaller, tighter account, three to five fresh creative concepts, a clean negative list, a five minute lead response, and every dollar pointed at something that has already proven it works.

Ninety minutes. Between now and October first. That is the entire ask.

The auction is going to get expensive no matter what you do. The only variable you control is whether it gets expensive on the stuff that makes you money or the stuff that never did.

Go open the account.

P.S. If you would rather not do this alone, reply with the word SPRINT and I will take a look at the account with you and tell you where the money is going.

Talk Soon,
Dan

Dan Kaufman
Founder, Dead Simple Growth and Pinnacle Masters