Solved: Creative Fatigue - Should I Test Different Audiences?

Hi there,

Thanks for reaching out! Happy to give you some of my initial thoughts and guidance on your situation. It's a really common problem to run into, so don't feel like you've done something wrong. A campaign that works well for a few months and then starts to see costs creep up is something we see all the time.

I've had a look at what you've described and put together some detailed thoughts for you below. Hopefully it gives you a clear path forward.

Okay, let's look at why you're costs are rising...

First off, what you're experiencing is completely normal. It’s called ad fatigue, or sometimes audience fatigue. You've been running the same video to the same interest-based audience since December. That's a good run, and getting a cost per purchase of $15-25 at a $450/day spend is a solid result, you should be pleased with that initial performance.

The problem is, no campaign can run forever without changes. Your audience, even with Advantage+ on, isn't infinite. After a few months, a large chunk of the most likely buyers within that audience have seen your ad. Many have probably seen it multiple times. At first, seeing an ad a few times can be good – it builds recognition. But after a point, people just start ignoring it. They become 'blind' to it. This is classic creative fatigue. The ad just doesn't grab attention like it used to.

At the same time, the platform's algorithm has already converted the 'low-hanging fruit' – the people who were easiest and cheapest to convert. To get more conversions, it now has to work harder, showing the ad to people who are less likely to buy, or who need more convincing. This combination of your creative losing its impact and the algorithm having to search in less fruitful parts of the audience is exactly why your cost per purchase has climbed from the $25 mark up to $40. Lowering the budget was a sensible immediate reaction, but it's a temporary fix, not a long-term solution to get you back to scaling profitably.

You asked if it's worth testing more audiences because you've heard targeting doesn't matter anymore. This is a bit of a myth, or at least an oversimplification. While Meta's AI is powerful, it's not magic. Giving it a single, broad interest and letting it run is a decent starting point, but it's not an advanced scaling strategy. It's like telling a taxi driver to just "drive west". You might get somewhere interesting, but it's not an efficient way to get to a specific destination. You absolutely need to be testing new audiences. But just as important, you need to be testing new creative, too. The two go hand-in-hand.

You'll need to get some new creative sorted...

Before you even touch your audience targetting, I’d look at your creative. You've been running one successful video. That's great, it's your control, your proven winner. But now it's fatiguing, it’s time to test new things against it. Relying on a single creative is a massive risk, as you're now discovering. When it stops working, your whole campaign grinds to a halt.

The goal isn't just to make one new ad, but to build a system of creative testing so you always have new ads in the pipeline, ready to swap in when performance on an older one starts to dip. This stops you from ever being in this position again.

A few thoughts on what to test:

-> Different Ad Formats: You've used a video. That's brilliant for telling a story and demonstrating a product. But what about other formats?

• Image Ads: A really strong, high-quality image can get a message across much faster than a video. It can be great for impulse buys and for capturing attention quickly in a busy feed. Test different product shots, lifestyle images, or even graphics with a strong headline.
• Carousel Ads: These are fantastic if you want to showcase multiple products, different features of a single product, or tell a step-by-step story. You could show different angles, different use cases, or even customer testimonials, one per card. They are interactive and can get really good engagement.
• UGC-style Video: User-generated content is huge right now. A video that looks like it was filmed by a happy customer on their phone can often feel more authentic and trustworthy than a polished, professional ad. I remember one client, a SaaS company, saw really good results with UGC videos. It's a completely different angle to test.

-> Different Hooks & Angles: Even with your existing video, you could test different opening hooks. The first 3 seconds are everything. Could you start with a question? A surprising statistic? A shot of the end result your product delivers? You could edit your existing video to create 3-4 new versions with different openings and test them against each other. Also think about the overall message. Are you selling based on price? Quality? A specific problem you solve? Each of these is a different angle you can build a new creative around.

The key is to split test properly. Don't just throw a new ad into your existing ad set. Ideally, you'd have a specific campaign or ad set dedicated to testing new creative, so you can clearly see what's working and what isn't without disrupting your main, proven campaign (your 'control'). Once you find a new winner that beats your old video, it becomes the new control, and you test future creative against that.

I'd say you need a proper audience testing structure...

Alright, now for your main question. Yes, it is 100% worth testing more audiences. Anyone who says targeting doesn't matter anymore is either only running massive budgets where broad targeting can work, or they've not had to scale a campaign from a tricky spot. For most advertisers, smart audience selection and testing is what seperates a breakeven campaign from a highly profitable one.

The way to think about this isn't just "what interest should I try next?". It's about building a full-funnel strategy. You need to target people at different stages of their awareness and journey with your brand. We usually break this down into three parts: Top of Funnel (ToFu), Middle of Funnel (MoFu), and Bottom of Funnel (BoFu).

-> ToFu (Top of Funnel): This is your cold traffic. People who've likely never heard of you before. This is where your interest-based targeting sits, along with other prospecting audiences like Lookalikes.

-> MoFu (Middle of Funnel): This is warm traffic. People who have shown some interest but aren't ready to buy yet. Think of people who've watched your videos or engaged with your social media page. They know you exist, but need more convincing.

-> BoFu (Bottom of Funnel): This is your hot traffic. People who are very close to buying. They've visited your website, maybe even added a product to their cart. They just need that final nudge to get them over the line.

Your current campaign is only focused on ToFu. By building out MoFu and BoFu campaigns, you create a complete system. You attract new people at the top, and then you have seperate, dedicated campaigns to nurture them and convert them at the middle and bottom. This is almost always more effective and efficient than trying to do everything with one ad set.

As a consultant, specialiseing in paid ads, when I audit client accounts, one of the most common issues is a messy or non-existent audience structure. They're often testing random audiences that don't align with their goals. Here’s a list of audiences I would usually prioritise, in order. The further down the funnel an audience is, the better it will usually perform because they are more qualified.

META ADS AUDIENCE PRIORITISATION

ToFu (Prospecting):

1. Detailed targeting (Interests, behaviours, demographics that you haven't tested yet)
2. Broad targeting (Only worth it once your pixel has thousands of purchases and you have a big budget)
3. Lookalike audiences of your best customers. I've listed these in order of value:
   • Lookalike of highest value previous customers (if you can upload that list)
   • Lookalike of all previous customers (from your pixel or a list)
   • Lookalike of people who Purchased
   • Lookalike of people who Added Payment Info
   • Lookalike of people who Initiated Checkout
   • Lookalike of people who Added to Cart
   • Lookalike of all Website Visitors
   • Lookalike of people who watched 50% of your videos

• People who visited a landing or product page (but excluded purchasers)
• All website visitors (excluded purchasers)
• People who watched 50% of your videos (excluded purchasers)
• People who engaged with your Facebook or Instagram page (excluded purchasers)

• People who Added Payment Info (in the last 7-14 days)
• People who Initiated Checkout (in the last 7-14 days)
• People who Added to Cart (in the last 7-14 days)

This structure gives you a roadmap. You can see there are loads of audiences you probably haven't tried yet. You don't have to test them all at once, but you should be methodically working your way through them.

We'll need to look at you're prospecting audiences first...

This is where you'll find your scale. Your single interest campaign is getting tired, so you need to find new ponds to fish in. The mistake most people make here is picking interests that are too broad or not specific enough. You need to think deeply about your ideal customer.

For example, if you were selling high-end camera equipment, targeting the interest "Photography" is way too broad. You'll get millions of people who just like taking photos on their phone. It's a poor quality audience. A much better approach would be to target interests like specific camera brands (e.g., Canon EOS R5, Sony Alpha), famous professional photographers, high-end photography magazines, or software like Adobe Lightroom. You're trying to find interests that your ideal customer would have, but that a casual amateur would not. This is what we mean by getting the targetting right.

Make a list of 5-10 of these highly specific interest 'themes'. Group related interests into a single ad set and test them. For example, one ad set could target a bundle of professional camera body interests. Another could target a bundle of photography magazine interests. Run them with a small budget and see which one delivers the best cost per purchase. Double down on the winners and turn off the losers.

Once you have more data, you should immediately start building and testing Lookalike audiences. These are often the best-performing cold audiences. Start with the highest-value one you can. If you have at least 100-200 purchases tracked on your pixel, create a 1% Lookalike audience of purchasers in your target country. This tells Meta to "go and find me more people who look and behave exactly like the people who have already bought from me". This is infinitely more powerful than just targeting a generic interest. As you get more data, you can test other Lookalikes from the priority list above (like Add to Carts, or Website Visitors). I remember working on campaigns where a simple switch from interest targeting to a purchaser Lookalike has halved the cost per acquisition.

You'll need a clear campaign structure...

Okay, so you have ideas for new creative and new audiences. How do you put it all together without creating a huge mess in your ad account? You need a simple, scalable structure.

I would suggest moving away from a single campaign and setting up at least two, maybe three, long-term campaigns:

1. A Prospecting (ToFu) Campaign: The goal here is to find new customers. This campaign will use a CBO (Campaign Budget Optimisation) budget. Inside this campaign, you'll have multiple ad sets. Each ad set will target a different cold audience.
   • Ad Set 1: Your original winning interest (to keep what's working, working)
   • Ad Set 2: New Interest Bundle A
   • Ad Set 3: New Interest Bundle B
   • Ad Set 4: 1% Lookalike of Purchasers
   • Ad Set 5: 1% Lookalike of Add to Carts
   Within each of these ad sets, you'd run your 2-3 best ads (your proven video and your new test creatives). The CBO will automatically push more of the budget to the best-performing ad set. You monitor the performance and turn off any ad sets that are too expensive after a few days.
2. A Retargeting (MoFu/BoFu) Campaign: The goal here is to convert the warm and hot traffic that your prospecting campaign generates. This should have its own, smaller budget.
   • Ad Set 1 (BoFu): Target people who Added to Cart or Initiated Checkout in the last 14 days (but didn't buy). The ad copy here can be more direct, maybe mentioning a discount or reminding them their cart is waiting.
   • Ad Set 2 (MoFu): Target people who visited your website or engaged with your social pages in the last 30 days (but didn't add to cart). The ads here could show testimonials, different product features, or answer common questions.

This kind of structure gives you so much more control and clarity. You can see exactly which part of your funnel is working and which isn't. If your prospecting CPA is good but sales are low, you know your retargeting needs work. If your retargeting is converting well but you're not getting enough volume, you know you need to find better prospecting audiences. I've sketched out a basic demonstration of how this might look in your ads manager.

This is the main advice I have for you:

This is a lot to take in, I know. It moves you from just 'running an ad' to actively managing a paid advertising system. It takes more work, but the results are far more consistent and scalable. To make it a bit clearer, I've broken down my main recommendations into a simple action plan for you.

I know this seems like a big shift, but tackling campaign fatigue requires a strategic response, not just a small tweak. By implementing a robust testing framework for both your creatives and your audiences, you'll not only solve your current issue of rising costs but also build a much more resilient and scalable advertising machine for the future.

This is exactly the kind of process we take our clients through. It’s about moving from guesswork to a data-led system. I remember one eCommerce client achieving a 1000% Return On Ad Spend for a subscription box, by implementing these exact kinds of structured testing and full-funnel strategies. It's a proven model.

If you'd like to chat through this in more detail and have us take a look at your account to give you some more specific pointers, we offer a free initial consultation. It might be helpful to have an expert eye on it to point you in the right direction.

Hope this helps!

Regards,

Team @ Lukas Holschuh