Mid-Year Account Hygiene: Naming, Exclusions and Clean Audiences

Open your ad account right now and sort campaigns by name.

If the first thing you notice is "Copy of Copy of New Sales Campaign 4" sitting above "TEST DO NOT USE," you do not have a performance problem yet. You have a legibility problem, which is how performance problems start.

The pattern shows up at every spend level: an account you cannot read is an account you cannot manage. You end up making decisions from memory, and memory is a bad analyst.

Why does account structure matter before Q4?

Because the expensive quarter is a terrible time to find out your account is confusing.

When budgets rise, small structural mistakes get multiplied. A leaky exclusion costs a little in July and a lot in November. A mislabeled campaign is a footnote in summer and a wrong decision in the middle of a promotion.

There is a second reason. Meta keeps moving toward automation. In February, it said it was testing a streamlined Advantage+ setup, renamed shopping campaigns as Advantage+ sales campaigns, and was extending Advantage+ to leads campaigns.² I covered the details in Advantage+ Sales Campaigns. The more the platform decides, the more your job shifts to the things only you can control: clear structure, clean inputs and honest naming.

What should a naming convention look like?

Boring, consistent and readable out of context. Imagine someone reading a CSV export at midnight with no access to your head.

Build one template and never break it

Pick the fields you will always need and put them in the same order. A template I like for campaigns:

OBJECTIVE_OFFER_AUDIENCETYPE_GEO_STARTMONTH

So a campaign reads SALES_Workshop_Prospecting_US_2025-08. At the ad set level, add what you are testing: Broad, Interest A or Lookalike 1%. At the ad level, name the angle and format: Story_BeforeAfter_Video.

Include dates and the variable under test

A launch date in the name tells you how stale something is. The variable under test tells you what the campaign was supposed to teach you. Together they make reports usable months later, when you have forgotten why you ran any of it.

What should you exclude, and why?

Everything that stops you from paying twice to reach someone who is already yours.

Customers and recent leads out of prospecting

This is the other half of reactivating past buyers. If your prospecting campaigns can still reach people who bought last month, you are spending acquisition budget on retention messaging. Exclude your customer list and your recent leads from cold audiences. The cleanest setups treat prospecting as strictly "people who do not know us yet."

Retargeting pools out of each other

Layered retargeting audiences often overlap heavily. Someone who watched your video, visited your site and opened a form can sit in three audiences at once. Decide which stage wins and exclude the earlier stages from the later ones, so each person sees one clear message for where they are.

Buyers out of buying messages

Remove recent purchasers from "buy now" ads. It sounds obvious, and it is missed constantly, because purchase lists go stale and nobody schedules their refresh.

How do you clean up audiences?

Audiences pile up quietly. Every launch adds a few, and nobody deletes the old ones.

Archive what you no longer use

List every saved and custom audience. Anything not used in the last 90 days that you cannot name a reason to keep can be archived. A shorter list is easier to audit, and it lowers the odds of someone attaching an out-of-date audience to a new campaign.

Rename what you keep

Apply the same convention to audiences: CUSTOMERS_EmailList_Upload2025-06 tells you what it is and when it was last refreshed. Audience 7 tells you nothing.

Refresh the lists that feed exclusions

Customer and lead lists decay. If your exclusion list was last uploaded in January, you are excluding January's customers and re-targeting everyone since. Put a recurring reminder on the calendar. Monthly is a sensible default.

Is your account too fragmented?

Possibly, and it matters more than most people think.

Meta's help center describes a "learning limited" status for ad sets that probably will not collect about 50 optimization events within seven days of their last significant edit, and lists consolidating ad sets, adjusting budget and broadening audiences among the ways to address it.¹ The logic (also behind Scaling Meta Budgets Without Resetting Learning) is simple: ten ad sets each struggling to reach five conversions teach the system less than one ad set that reaches fifty.

Signs of fragmentation

• Several ad sets with nearly identical audiences.
• Budgets spread so thin that none of them collects meaningful results.
• A different campaign for every small creative idea.

If you see these, merge by purpose. One ad set per genuinely different audience and offer, not one per hunch.

Jon Loomer makes a specific version of this case for Advantage+ audience in a video he posted on February 26, 2024. That setting starts from your conversion history and then goes broad, so extra ad sets add overlap, not control:

His point is about Advantage+ audience, so check how your ad sets are set up before you merge anything. The principle holds either way: one ad set per genuinely different job.

Plays here, from YouTube

What does a one-afternoon cleanup look like?

You can do most of this in a few hours. The order matters.

When should the cleanup happen?

Now, while spend is calm. Budget, audience and creative edits can each restart learning, and you want any wobble to happen when the stakes are low.

Then lock the structure before Q4 (when Black Friday budget pacing gets tested) and make only the changes you can explain. I do this kind of cleanup as part of the system I build for clients. A clean account is not exciting. It is the reason you can trust the numbers when things get exciting.

Keep building.Ekai