Solved: Campaign Overlap Between Scaling and Testing Campaigns?

Hi there,

Thanks for reaching out!

Happy to give you some initial thoughts on this. Your suspicion is bang on the money, and it’s a very common trap people fall into. You've basically sent multiple salesmen to knock on the same door at the same time, and now you’re wondering why none of them are closing the deal and why it’s costing you more to have them stand there.

The short answer is yes, you almost definately have an audience overlap issue, and it's likely the direct cause of your rising CPMs and stalling conversions. Let's get this sorted. I'll walk you through why this is happening, how to build a proper account structure to prevent it, and how to think about testing and scaling in a way that actually works instead of just burning cash.

We'll need to look at the Audience Overlap issue...

First things first, let's get right to the heart of the problem. You have multiple CBO campaigns all targeting the same interests. Then you launched a new CBO campaign, which also targets those exact same interests in one ad set. This is a cardinal sin in Meta advertising.

You have to understand how the Meta auction works. It's not just about your ad against a competitor's ad. When you have multiple ad sets or campaigns targeting the same pool of people, you are also bidding against yourself. Each time a user from that overlapping audience logs into Facebook or Instagram, Meta has to decide which of your ad sets to show them. Your "testing" campaigns and your "scaling" campaign are now in a direct fight for the same eyeballs. This internal competition artificially inflates your bids, which is why you're seeing your CPMs shoot up. The algorithm is confused. It can't effectively learn or optimise because it's being pulled in multiple directions by your own account structure.

Think of it this way: Meta's goal is to find you conversions at the lowest possible cost within the constraints you give it. By creating this massive overlap, you've created a messy, chaotic environment. The algorithm can no longer reliably predict which ad set will perform best, so performance becomes erratic, and costs rise. Your old campaigns stopped converting because the new, likely higher-budget, scaling campaign started winning the internal auction for the best users within that audience, but even it can't perform optimally because it's still fighting the remnants of your old campaigns for scraps. You're paying more for worse results.

There's a common myth about "brand awareness" campaigns where people pay Meta to find the cheapest, least-likely-to-convert users. You've accidentally created a version of this inside your conversion campaigns. By forcing your ad sets to compete, you're paying a premium to confuse the algorithm and prevent it from efficiently finding the people who will actually buy from you. It's an expensive mess.

The immediate fix is simple: Pause all those old testing campaigns. Right now.

You cannot test and scale the same audience at the same time. It doesn't work. The purpose of testing is to identify winning audiences and creatives. The purpose of scaling is to exploit those winners with increased budget. They are two distinct, sequential phases, not parallel activities. Once you've identified a winning interest or creative in a test, you move it to a scaling campaign and you turn the test off. You don't let them both run.

I'd say you need a proper account structure...

Pausing the old campaigns is a short-term fix. The real, long-term solution is to build an account structure that prevents this from happening in the first place. A lot of people I see when I audit accounts are just throwing ad sets at the wall to see what sticks, with no clear system. This leads to the exact problem you're facing. What you need is a logical, funnel-based structure that segments your audiences by their intent and relationship with your brand.

I typically prioritise audiences in a very specific order, moving from coldest to warmest. This ensures you're speaking to people correctly based on where they are in their journey with you, and it allows you to use campaign exclusions to create clean, non-overlapping audiences. For an eCommerce business, it would look something like this:

META ADS FUNNEL STRUCTURE

1. Top of Funnel (ToFu) - Prospecting: This is for finding new customers who have never heard of you. Your goal here is to drive traffic and initial conversions from cold audiences.

-> Lookalike Audiences: These are your best bet for cold traffic once you have data. The algorithm finds people similar to your existing customers. The priority is:
   1. Lookalike of Highest Value Customers
   2. Lookalike of All Purchasers (180 days)
   3. Lookalike of Initiated Checkouts (180 days)
   4. Lookalike of Adds to Cart (180 days)
   5. Lookalike of All Website Visitors (180 days)

-> Detailed Targeting: This is where your interest-based targeting lives. Group related interests into themes. For example, if you sell high-end coffee gear, one ad set might target interests like 'James Hoffmann', 'Specialty Coffee Association', 'Fellow Products', while another targets competing brands like 'La Marzocco' or 'Rocket Espresso'.

-> Broad Targeting: This is what you're already testing. It can be incredibly powerful, but only once your pixel has thousands of conversion events. You give Meta no targeting other than age, gender, and location, and let the algorithm find buyers. It works, but it needs a lot of data to learn.

2. Middle of Funnel (MoFu) - Warm Retargeting: This is for people who have shown some interest but haven't made a move towards buying yet. They've visited your site or engaged with an ad.

-> All Website Visitors (30 Days)
-> Facebook & Instagram Page Engagers (90 Days)
-> 50% Video Viewers (90 Days)

3. Bottom of Funnel (BoFu) - Hot Retargeting: These people are on the verge of buying. They've added a product to their cart or started the checkout process. They need a final push.

-> Added to Cart (14 Days)
-> Initiated Checkout (14 Days)
-> Viewed Cart (14 Days)

The magic is in the exclusions. Your ToFu campaign must exclude all MoFu and BoFu audiences. Your MoFu campaign must exclude all BoFu audiences and past purchasers. Your BoFu campaign must exclude past purchasers. This creates a clean waterfall where a user only ever lives in one campaign at a time, eliminating overlap completely.

Here’s how you could set it up with seperate campaigns:

This structure gives the algorithm clarity. It knows exactly who to target in each campaign, and you can tailor your ad copy and offers accordingly (e.g., a discount code for the BoFu audience to recover abandoned carts). This is how you build a scalable advertising machine.

You probably should rethink your approach to 'testing'...

Your current method of "testing" isn't really testing; it's just creating chaos. True testing is a methodical process of isolating variables to find out what works. You shouldn't be running four CBOs to test the same interests. That tells you nothing.

Instead, testing should happen within your new prospecting campaign. You have your CBO campaign set up for ToFu. Inside it, you create multiple ad sets. One ad set is one audience test. For instance:

-> Ad Set 1: Broad
-> Ad Set 2: Lookalike 1% of Purchasers
-> Ad Set 3: Interest Stack A (e.g., Competitor Brands)
-> Ad Set 4: Interest Stack B (e.g., Related Hobbies/Publications)

You let them run against each other with the same set of creatives. The CBO will automatically shift budget to the best-performing ad set. After a few days (or once an ad set has spent 2-3x your target cost per purchase without a sale), you turn off the losers. Then you introduce a new ad set to test against the surviving champions. This is a perpetual cycle of test, learn, kill, and repeat. You do the same for creatives: put your winning audiences in one ad set and test multiple different ads against each other.

But the bigger question is what you're testing. You mentioned you're testing "interests." This leads to the most common mistake I see: marketers defining their customer by sterile demographics or broad interests. Forget that. Your Ideal Customer Profile (ICP) is not a demographic; it's a nightmare. It's a problem state.

Your ad needs to speak to that problem. Let's say you sell high-quality camera bags. Your ICP isn't "people interested in photography." That's uselessly broad. Their nightmare is being on a once-in-a-lifetime trip and missing the perfect shot because their gear is disorganised, or worse, damaged because their bag failed. They are terrified of their £5,000 lens smashing on the pavement. Your targeting and your ads must reflect this world. You should be targeting interests like specific camera brands (Sony, Canon, Fuji), professional photographer names (Peter McKinnon), gear review websites (DPReview), and software they use (Lightroom, Capture One). This intelligence is the blueprint for your targeting. Targeting "photography" is lazy and a waste of money. You must target their world, their tools, their heroes, and their fears.

Do this work first. Dig deep into who your customer really is and what keeps them up at night. Then build your interest stacks around that. That's when your testing will start to yield real, scalable results.

You'll need to focus on what actually matters for scaling...

You mentioned your goal is "scaling," but you're worried about a rising CPM. This tells me you might be focused on the wrong metric. CPM, CPC, CTR... these are vanity metrics. They are indicators, but they are not the goal. The only metric that truly matters is the relationship between how much it costs you to acquire a customer (CAC) and how much that customer is worth to you over their lifetime (LTV).

The real question isn't "How low can my CPM go?" but "How high a CPL can I afford to acquire a truly great customer?" The answer lies in its counterpart: Lifetime Value (LTV).

Let's do some quick, back-of-the-napkin maths. You need to understand this to make smart scaling decisions. For an eCommerce store, a simple LTV calculation might look like this:

1. Average Order Value (AOV): What's the average value of a single purchase? Let's say it's £60.
2. Purchase Frequency (per year): How many times does the average customer buy from you in a year? Let's say it's 2.5.
3. Gross Margin %: What's your profit margin on each sale after cost of goods? Let's say it's 70%.

Calculation for 1-Year LTV:

LTV = (AOV * Purchase Frequency * Gross Margin %)
LTV = (£60 * 2.5 * 0.70)
LTV = £150 * 0.70 = £105

In this example, each customer is worth £105 in gross margin to your business in their first year. A healthy LTV:CAC ratio is typically 3:1. This means you can afford to spend up to £35 to acquire a single new customer (£105 / 3) and still run a very healthy, profitable business.

Suddenly, a rising CPM doesn't seem so scary, does it? If your CPM goes from £5 to £10, but your Cost Per Acquisition (CPA) stays at £25, who cares? You're still well within your profitable range. This is the math that unlocks aggressive, intelligent growth. It frees you from the tyranny of cheap clicks and allows you to confidently scale your budget, even with Broad targeting, because you know your absolute upper limit for a profitable acquisition.

This is why a solid funnel structure is so important. We've used this exact approach for our clients to achieve massive returns. I remember one women's apparel client where we hit a 691% Return on Ad Spend (ROAS) precisely because we stopped worrying about front-end metrics and focused on optimising the whole funnel against their target CAC. For another eCommerce client selling cleaning products, we drove a 633% return. This is what happens when you have a clear structure and you focus on business metrics, not platform vanity metrics.

This is the main advice I have for you:

I know this is a lot to take in, so I've broken down the core strategy into a clear action plan. This is the framework for moving from a chaotic, unprofitable setup to a scalable, data-driven advertising system.

Running paid advertising effectively is more than just knowing how to click the buttons in Ads Manager. It's about having a deep understanding of auction dynamics, marketing psychology, and business finance. The framework I've laid out is the foundation, but the real work lies in the day-to-day execution: constant monitoring, analysing data, iterating on creatives, and making hundreds of small decisions that compound over time.

This is where expert help can make a huge difference. An experienced hand can implement this structure quickly and manage it efficiently, avoiding costly mistakes and accelerating your path to profitable scale. It frees you up to focus on what you do best: running your business, developing products, and talking to your customers.

If you'd like to have a more in-depth chat and have me look over your account with you, we offer a completely free, no-obligation strategy session. We can walk through your specific setup and build a tailored action plan for you. Feel free to book a time if that sounds helpful.

Hope this helps!

Regards,

Team @ Lukas Holschuh