Solved: Extreme Fluctuations in Meta Ads Performance

Hi there,

Thanks for reaching out. I had a look over the situation you described with your Meta ads, and I'm happy to give you some initial thoughts and guidance. It's a really common problem, those daily performance swings can be incredibly frustrating, especially when you're trying to scale your ad spend. It feels like you're riding a rollercoaster you can't control.

The good news is that while some fluctuation is normal in paid advertising, the extreme swings you're seeing from a 7.5% conversion rate down to 3%, and CPAs jumping from $16 to $25, usually points to some underlying issues in the campaign structure and strategy. It's not just random bad luck, and it's definitely something that can be managed and stabilised. The fact you've had to pull back your spend from $5k/day to $1k/day tells me we need to build a more resilient foundation for your campaigns before you can scale back up profitably.

Let's get into it.

I'd say you need to understand the 'why' behind the fluctuations...

First off, lets talk about why this happens. Meta's ad delivery is an auction. Every day, you're competing against a different set of advertisers for the same eyeballs. The costs can change based on who else is bidding, what time of day it is, and even what day of the week it is. Your "Monday problem" is a classic example of this. Mondays are often a big day for eCommerce brands launching weekly promotions, so competition can spike, driving up your costs if your campaign isn't set up to handle it.

The algorithm itself also plays a part. When you're using campaign budget optimisation (CBO) or advantage+ shopping (ASC), Meta is constantly trying to find the cheapest conversions. Sometimes it has a good day and finds a pocket of users who are ready to buy. The next day, that pocket might be exhausted, and it has to search wider, which costs more. Your job isn't to eliminate these fluctuations entirely – that's impossible – but to build a system that minimises the downs and maximises the ups, creating a more stable average over time.

The fact you've tested CBO, ABO, and ASC tells me you're doing the right things in terms of testing Meta's tools. But these tools are only as good as the inputs you give them. The problem likely isn't the campaign type, but the strategy underneath it: your audience targeting, your creative, and how you're measuring success. If we can get those bits right, the fluctuations will become much more manageable and your profitable days will become far more frequent, whatever day of the week it is.

We'll need to look at your audience strategy...

This is probably the single biggest area for improvement for most ad accounts I see, and I suspect it's a major cause of your issues. When performance is inconsistent, it's often because the audiences are either too broad, not structured properly, or you're not effectivly moving customers through the different stages of awareness.

You need to think about your advertising in terms of a funnel. Right now, it sounds like you might be lumping everyone together. We need to seperate them out. Think of it in three stages:

• Top of Funnel (ToFu): These are new people who have never heard of your beauty brand. They are your cold audiences. The goal here is to introduce them to your products and get them to your website.
• Middle of Funnel (MoFu): These are people who have shown some interest. They've visited your website, watched a video, or engaged with an ad, but they haven't added anything to their cart yet. The goal is to get them back to the site and encourage them to browse more deeply.
• Bottom of Funnel (BoFu): These are your hottest prospects. They've added a product to their cart, or even started the checkout process, but didn't complete the purchase. The goal here is to push them over the finish line.

You need seperate campaigns (or at least ad sets) for each of these stages, with different messaging and creatives for each. By lumping them together, Meta might spend your budget on easier-to-reach (but less likely to convert) cold audiences one day, and then find a pocket of high-intent cart abandoners the next, causing those massive swings in CPA and conversion rate. By separating them, you gain control and can allocate budget where it's performing best.

For an eCommerce brand like yours, here's how I would prioritise testing your audiences. The further down the list, the "colder" the audience is. You should always prioritise the audiences closest to the purchase action because they'll almost always give you the best return.

You should have dedicated campaigns for these. A simple structure would be one campaign for Prospecting (ToFu) and one for Retargeting (MoFu/BoFu combined, if your budget is smaller). This ensures a portion of your budget is always going towards warming up your hottest audiences. With the kind of spend you were doing, I'd even suggest three seperate campaigns. This structure provides stability. Your retargeting campaign should be your consistent performer, generating a stable ROAS, while your prospecting campaign finds new customers. The performance of the prospecting campaign might fluctuate more, but the retargeting campaign acts as a solid, profitable base.

I remember working with a women's apparel brand, a similar space to beauty, and implementing a proper funnel structure on Meta was a huge part of how we helped them achieve a 691% return on their ad spend. It's all about control and directing your money to the right place at the right time.

You probably should rethink your creative approach...

Your ads themselves are the other huge variable. If your performance is dropping off, it might be because people are getting tired of seeing the same ads. This is called 'ad fatigue', and it's a very real thing. If your frequency metric is climbing (say, above 3-4 in your prospecting campaigns), it's a sure sign you need to introduce new creatives.

You can't just run one or two winning ads and expect them to last forever, especially at a $5k/day spend. You need a system for continuously testing and rotating your ad creatives. For a beauty brand, the possibilities are endless. You should be testing:

• Different formats: Static images vs. carousels vs. video ads. Video is massive for beauty – tutorials, before-and-afters, user-generated content (UGC) style reviews. These feel more authentic and often perform much better than polished studio shots.
• Different hooks: The first 3 seconds of your video or the headline of your image ad are everything. You should be testing different opening lines. One could focus on a problem ("Tired of makeup that doesn't last?"), another on a benefit ("Get that all-day glow"), and another on a unique feature ("Made with all-natural ingredients").
• Different angles: Don't just show the product. Show the experience. Show someone feeling confident wearing it. Show the texture. Show how easy it is to apply. I remember generating over 10 million views for a luxury brand launch by focusing on the lifestyle and feeling around the product, not just the product itself.
• User-Generated Content (UGC): This is gold for beauty brands. Get real customers (or pay creators) to make simple, phone-shot videos of them using and loving your product. It builds massive trust and social proof, and often outperforms expensive, professional ads.

When you're testing, be methodical. In your prospecting campaign, create multiple ads within each ad set. Let Meta's algorithm find the winners. After a few days, turn off the losers and replace them with new tests. It should be a constant cycle of testing, learning, and iterating. This proactive approach to creative testing stops your performance from suddenly falling off a cliff because an ad has gone stale. It smooths out the bumps because you always have new, fresh ads being introduced into the auction.

For example, here's how you could frame a simple UGC-style ad with different copy angles:

Testing these simple variations can uncover what messaging truly resonates with your audience, leading to a lower, more stable CPA.

You'll need a better way to measure success...

You mentioned your AOV is $33 and your CPA fluctuates between $16 and $25. It's easy to get obsessed with the daily CPA, but it can be a misleading metric to focus on so tightly. Instead, you should be focusing on Return On Ad Spend (ROAS). Are you getting more money out than you're putting in?

Let's do some quick maths.
-> On a good day: CPA is $16, AOV is $33. Your ROAS is $33 / $16 = 2.06x.
-> On a bad day: CPA is $25, AOV is $33. Your ROAS is $33 / $25 = 1.32x.

A 1.32x ROAS is tight. After accounting for the cost of your goods, shipping, and other overheads, you're likely losing money on those days. A 2.06x ROAS is better, but might still be close to break-even depending on your margins. The goal of the structured approach we've discussed is to push that average ROAS up, so even on a 'bad' day, you're still profitable.

I'd also check your attribution settings. Are you looking at a 1-day click or a 7-day click window? Many people buy a day or two after seeing an ad. A 7-day click window gives you a more accurate picture of your true CPA and ROAS. Your "bad" days might not be as bad as they seem when you account for delayed conversions.

For eCommerce clients, we often aim for a much higher ROAS. For example, we achieved a 633% return for a cleaning products eCommerce store using Meta Ads. This isn't to say you should expect that overnight, but it shows what's possible when your funnel, creative, and targeting are all working together. The focus shifts from "what's my CPA today?" to "what's my blended ROAS over the last 7 or 14 days?". This is a much healthier and more scalable way to look at performance.

Finally, you need to consider increasing your Average Order Value. With an AOV of $33, you have very little room for error on your CPA. If you could increase your AOV to $50, suddenly even a $25 CPA gives you a 2.0x ROAS. Can you create product bundles? Offer a small discount for buying two items? A post-purchase upsell? Any small increase in AOV gives you much more breathing room with your ad spend and makes the daily fluctuations less scary.

This is a lot to take in, I know. But the core idea is to move from a reactive approach (panicking at daily fluctuations) to a proactive one (building a resilient system). I've detailed my main recommendations for you in a table below to make it clearer.

Implementing a full system like this takes time and a fair bit of expertise, especially the ongoing management of creative testing and budget allocation between campaigns. While the advice above should give you a solid plan to start turning things around, doing it all yourself can be a full-time job.

This is where getting expert help can make a huge difference. An experienced eye can spot opportunities you might miss, implement these complex structures far more quickly, and manage the day-to-day optimisation process to ensure you're getting the best possible return from every pound you spend. You get to focus on running your beauty brand, safe in the knowlege that your advertising is in expert hands.

If you'd like to chat through this in more detail and have us take a proper look at your ad account, we offer a free, no-obligation initial consultation. We could walk through your current setup together and give you an even clearer picture of the path to stable, scalable growth.

Regards,

Team @ Lukas Holschuh