Solved: Testing a New Product with CBO and Multiple Ad Sets

Hi there,

Thanks for reaching out!

I saw your query about testing a new product with your proposed ad strategie and I'm happy to give you some initial thoughts and guidance based on my experience. It's a common situation, especially when you're starting out, to want to test as much as possible to find what works quickly. Your instinct to test multiple angles is good, but the way you've structured it might not give you the clear answers you're looking for. There's a few things to unpack here, and I'll do my best to lay out a path that might be more effective for you.

Paid advertising can feel like a bit of a maze when you're new to it, but once you get the fundamentals of testing and structure right, it becomes a much more predictable tool for growth. Let's get into it.

I'd say you need to reconsider the campaign structure...

The first thing that jumps out at me is the structure you've proposed: one CBO campaign at $55/day, with 7 ad sets. CBO (Campaign Budget Optimisation) is a powerful tool, but it works by distributing your budget to the best-performing ad sets in real-time. With a budget of $55 spread across 7 ad sets, you're looking at an average of just under $8 per ad set per day. That's a very small amount of budget for the algorithm to work with.

Here's the problem with that: Meta's algorithm needs data – specifically, conversion events – to learn who your ideal customer is and which ad set is the 'winner'. To get that data, it needs to spend enough money to generate clicks and, most importantly, conversions. If an ad set only has a few dollars to spend each day, it might get a handful of clicks but it's very unlikely to get a sale or whatever your main conversion goal is. We've seen this happen with clients before; they have dozens of adsets running on a small budget and none of them ever get out of the 'learning phase' because there's simply not enough spend or data points per adset for the system to make an informed decision.

The result is that the CBO might spend the whole $55 on one ad set that got a lucky click early in the day, starving the other six of any budget at all. Or, it might spread the budget so thinly that none of the ad sets acheive anything meaningful. You'll end the day with very little conclusive data, not knowing if an angle failed because the creative was bad, the audience was wrong, or if it just never had a fair chance to perform due to budget constraints. It makes your test unreliable.

You also mentioned mixing videos and images in the same ad sets. While testing different formats is a great idea, lumping them together in the same ad set can sometimes muddy the waters. Meta's algorithm might favour one format over the other (e.g., it often pushes video more), meaning your images might not get much delivery. This again makes it hard to know if an image creative was actually bad, or if it just wasn't shown to enough people. It's better to isolate variables where you can. We'll get into a better testing structure later on.

We'll need to look at your campaign objective...

Before we even get to structure, we have to talk about the objective. You mentioned testing a new product, so I assume your ultimate goal is to get sales. This means you should be optimising your campaign for 'Purchases'. A lot of beginners make the mistake of optimising for cheaper events like 'Link Clicks' or 'Landing Page Views' because it feels like they're getting more results for their money. But these are vanity metrics. You'll get lots of low-cost clicks from people who are happy to browse but have no intention of buying. It's a false economy.

When you tell Meta to find you 'Purchasers', it uses its vast dataset to find people who have a history of buying things online and who look like others that have bought from you (once you have some data). This is far more powerful, but it's also more expensive. A 'purchase' conversion is the most valuable action a user can take, so it costs more to acquire. This is a fundemental concept that underpins everything else.

This directly ties back to the budget issue. If you're optimising for purchases, you need to set a budget that gives you a realistic chance of getting them. A common rule of thumb is to have a daily budget per ad set of at least your target Cost Per Acquisition (CPA). If you don't know your target CPA yet, you need enough budget to get at least 1-2 sales per day across the campaign to give the algorithm something to work with. With a $55/day budget, this might be a stretch, which brings us to the next point.

You probably should budget for realistic costs...

So, what should you expect to pay for a sale? This is the million-dollar question, and the answer is always "it depends". It depends on your product, your price point, your target audience, your website conversion rate, and the country you're targeting. However, based on the campaigns we've run for numerous eCommerce clients, we can establish some realistic ballpark figures.

For eCommerce stores targeting developed countries (like the US, UK, Canada, Australia), a typical conversion rate is around 2-5%. The Cost Per Click (CPC) often lands between £0.50 and £1.50. Let's do the maths on that.

As you can see, the expected cost for a single sale could be anywhere from £10 to £75 (that's roughly $12 to $95). Suddenly, that $55 daily budget looks very different. In a worst-case scenario, you might not even have enough budget to get one sale a day. How can you possibly test 7 different angles if you're not even generating one conversion? It's impossible.

I remember one campaign we worked on for a cleaning products company. We achieved a 633% return and a 190% increase in revenue by implementing a solid Bottom of Funnel retargeting strategy. We started with a realistic budget that allowed us to gather data, test properly, and scale the winning audiences. If we'd stuck with a budget that was too low, we wouldn't have achieved those results.

You'll need a more structured testing aproach...

Okay, so given the budget and the need for clear data, how should you approach this? I'd scrap the idea of testing 7 angles at once. Instead, you need a more methodical, phased approach. The goal is to isolate variables. First, you find a winning AUDIENCE. Then, you test your CREATIVES within that winning audience.

Phase 1: Audience Testing

Instead of 7 ad sets, I'd start with just 2 or 3. Put your entire $55 budget into this test. This gives each ad set a much healthier daily budget of around $18-$27. This is still on the low side, but it gives the algorithm a much better chance to find you some conversions.

What audiences should you test? Since this is a new product, you'll be starting with Top of Funnel (ToFu) audiences. You don't have past purchaser data yet. So you'll rely on 'Detailed Targeting' (interests, behaviours).

The key here is to choose interests that are as specific as possible to your ideal buyer. I see a lot of people making the mistake of choosing very broad interests. For example, if you're selling a niche kitchen gadget, targeting the interest "Food" is far too broad. You'll reach everyone from gourmet chefs to people who just like posting pictures of their dinner. You need to think deeper. What magazines do your customers read? What brands do they love? What TV shows do they watch? What specific hobbies are related to your product?

Let's say you're selling a new type of eco-friendly yoga mat. Here’s how you could structure your initial audience test:

In this phase, you should use the SAME 2-3 ads in EACH ad set. These should be what you believe are your strongest "control" creatives. This is important. By keeping the ads the same, you ensure that any difference in performance between the ad sets is due to the AUDIENCE, not the creative. You're isolating the audience variable.

Run this for a few days, until each ad set has spent at least 2-3x your target CPA (or for at least 4-5 days if you're not getting sales). By the end, you should have a clear winner – one audience that is delivering a lower CPA or higher ROAS than the others. Now you move to Phase 2.

Phase 2: Creative Testing

Now that you've found a promising audience, you can unleash your 7 creative angles. You'll create a new campaign (or duplicate the winning ad set) and this time, you'll keep the AUDIENCE the same (your winner from Phase 1) and create 7 different ADS inside that one ad set. This is where you test your videos vs images, different headlines, different calls to action, etc.

This method, often called "Angle Grinding" or systematic creative testing, is far more effective. You're now testing your creatives against a proven audience, so you can be confident that the performance differences you see are due to the ads themselves. You're isolating the creative variable.

After a few days, you'll know which 1-2 angles are your top performers. These are the creatives you'll then use to scale your campaigns.

I'd say you should build a proper funnel...

This phased testing aproach is great for finding your initial winning combination. But for long-term, sustainable success, you need to think in terms of a funnel. This is how all successful eCommerce brands structure their advertising. The basic idea is to treat customers differently depending on how familiar they are with your brand.

We usually structure this into three stages: Top of Funnel (ToFu), Middle of Funnel (MoFu), and Bottom of Funnel (BoFu).

Top of Funnel (ToFu): Prospecting
This is for reaching cold audiences – people who have never heard of you. The goal is to introduce your brand and products.
-> Audiences: This is where your Detailed Targeting (interests, behaviours) and Lookalike Audiences live. Once you get at least 100 purchases, you should create a Lookalike audience of your buyers. This is often the most powerful prospecting audience you can have. You'd prioritise them like this:

1. Lookalike of highest value previous customers
2. Lookalike of all previous customers
3. Lookalike of people who Added to Cart
4. Your winning Detailed Targeting audiences from Phase 1

Middle of Funnel (MoFu): Engagement Retargeting
This is for retargeting people who have shown some interest but haven't visited your website yet. They might have watched one of your video ads or engaged with your Facebook/Instagram page.
-> Audiences:

• People who watched 50% of your video ads
• People who engaged with your Instagram/Facebook page

Bottom of Funnel (BoFu): Website Retargeting
This is the money-maker. These are people who have visited your website but didn't buy. They are warm leads and are much more likely to convert. Your messaging here should be more direct, maybe featuring a testimonial, a reminder of your value proposition, or even a small discount to get them over the line.
-> Audiences (in order of priority):

• Added to Cart (last 7-14 days) but didn't purchase
• Initiated Checkout (last 7-14 days) but didn't purchase
• Viewed a Product Page (last 14-30 days) but didn't add to cart
• All Website Visitors (last 30 days) but didn't purchase

You'd have separate campaigns for each stage of this funnel, with budgets allocated accordingly. Most of your budget will go to ToFu, with smaller amounts dedicated to MoFu and BoFu, as those audiences are smaller but convert at a much higher rate.

This is the main advice I have for you:

I know this is a lot to take in, especially when you're just starting out. It's a far cry from just setting up one CBO with a few ad sets. But this kind of structured, methodical approach is what separates the advertisers who get consistent results from those who just burn through cash. I've detailed my main recommendations for you in a table below to make it clearer.

As you can probably tell, geting this right involves a lot more than just pressing a few buttons in Ads Manager. It requires a deep understanding of strategy, auction dynamics, audience psychology, and methodical testing. It's a full-time job, and doing it wrong can be a very expensive learning experience.

This is where working with a specialist can make a huge difference. Instead of spending months and thousands of dollars on trial and error, you can lean on the experience of a team that has built and scaled these kinds of funnels many times before. We can help you sidestep the common pitfalls, identify the right audiences and creatives from day one, and build a robust advertising system that becomes a reliable engine for growth for your business.

If you'd like to chat through this in more detail and see how we could apply these principles specifically to your new product, we offer a free initial consultation. We can review your setup and give you a clear roadmap for what needs to be done. There's no obligation at all, of course.

Hope this helps give you a bit of clarity!

Regards,

Team @ Lukas Holschuh