Solved: How to Improve Mediocre Facebook and TikTok Ads?

Hi there,

Thanks for reaching out. I saw your post and thought I could offer some thoughts. It’s a really common situation to be in, so don't feel like you're the only one hitting this wall. That feeling of having something that kinda works, only for it to fizzle out the moment you try and give it a push, is incredibly frustrating. The good news is, it's usually not about the ad being "lucky" for a day or two; it's almost always a sign of deeper issues in the overall strategy and structure. The decay you're seeing isn't just random, it's a symptom.

I'm happy to give you some initial guidance based on my experience. What you're describing points to a few core areas that likely need a proper rethink: your testing framework, your audience strategy, and how you're measuring success and approaching scale. Let's break it down a bit.

I'd say you need to overhaul your creative testing framework...

First off, the idea that an ad just "works" or "doesn't" is a bit of a trap. Most winning ads aren't born, they're built through iteration. A 2.0 ROAS on cold traffic for a few days at a low spend is a signal. It's not a winner yet, but it's a signal that something in that ad is resonating. The job isn't to just crank up the budget on that exact ad, but to figure out *what* part is resonating and build on it. When you double the budget from £100 to £200, you're not just showing the ad to more people; you're forcing the algorithm to find conversions in more expensive, less receptive pockets of your audience, and any weakness in the ad gets magnified instantly. That's why it dies.

A mediocre ad is just a starting point. Your goal should be to create a system for improving it. This means moving away from just launching an ad and hoping, and towards a methodical process. What you need is a proper creative testing system.

Isolating Variables

When an ad shows a glimmer of hope, you need to become a bit of a scientist. Break the ad down into its core components:

-> The Hook (first 3 seconds of a video, the headline, the main image)
-> The Angle / Messaging (the core problem you're solving or desire you're fulfilling)
-> The Creative Format (Image, Carousel, UGC Video, Polished Video)
-> The Copy (the main body of text)
-> The Call to Action (the button and the text leading to it)

Your "mediocre" ad has one or two of these elements right, and the others are probably weak. The next step isn't to scale, but to test. For instance:

Original Ad (Ad A): Image of Product, Headline "New Summer Collection", Copy about features, 2.0 ROAS.
Test 1 (Ad B): Keep the SAME Image. Change Headline to "Tired of Boring Summer Outfits?". Test if a problem-focused hook works better.
Test 2 (Ad C): Keep the SAME Headline. Change the Image to a User-Generated-Content (UGC) style photo of someone wearing the product. Test if a different creative style is more persuasive.
Test 3 (Ad D): Keep the SAME Image and Headline. Re-write the copy to focus on benefits and feelings, not just features. Maybe get a professional copywriter to look at it. We've seen with some of our B2B SaaS clients that getting a specialist copywriter onboard completely changes the game. The same principle applies here.

This proccess is methodical. You're not just throwing stuff at the wall. You're learning with every test. You're trying to find a winning combination of hook, message, and format that can actually sustain a higher spend. An ad that dies at £200 a day does so because it's not persuasive enough for the wider audience you're forcing Meta to find. Your job is to make it more persausive.

We've had several clients, particularly in the SaaS space, achieve great results with UGC videos. For an eCommerce brand, this could be massive. Think about getting videos from customers, or even just shooting simple, authentic-looking videos on your phone showing the product in use. This often out-performs slick, professional ads because it feels more real and trustworthy.

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

This is probably just as, if not more, important than the creative itself. Who you show your ads to determines everything. The quick decay you're seeing is a classic sign that your audience targeting might be too narrow, or not structured correctly for scaling.

I often see accounts where people are just testing random interests in one campaign. A much better approach, and what we implement for our clients, is a structured funnel approach: Top of Funnel (ToFu), Middle of Funnel (MoFu), and Bottom of Funnel (BoFu). Your "cold traffic" campaigns are your ToFu.

Top of Funnel (ToFu) - Prospecting:

This is where you find new customers. The problem many people have here is picking interests that are way too broad. For example, if you sell high-quality kitchen knives, targeting an interest like "Cooking" is probably a waste of money. Millions of people who just watch cooking shows on TV are in that audience; they aren't necessarily passionate home chefs willing to spend £150 on a knife. You need to be more specific.

Think about the niche interests your ideal customer has. What brands do they follow (e.g., specific high-end cookware brands)? What magazines or influencers do they engage with? Who are their heroes? Targeting "Gordon Ramsay" is broad. Targeting a specific, high-end culinary school or a niche food blogger is much sharper. You need to find interests where the concentration of your ideal customer is high.

Once you get enough data (and you really need at least 100 purchases, ideally a lot more), you start building Lookalike audiences. But you have to be strategic here. A Lookalike of "All Website Visitors" is weak. A Lookalike of "Added to Cart" is better. A Lookalike of "Purchasers" is good. A Lookalike of your "Highest Value Previous Customers" is the best. You should prioritise testing them in that order. The audience quality gets better the further down the funnel the source event is.

Here's a rough prioritisation I'd use for ToFu audiences:

1. Detailed Targeting: Start with your best-guess, highly specific interests. Group them into themes.
2. High-Intent Lookalikes: Once you have enough data, build LALs of Purchasers, Initiated Checkouts, etc. (1% first, then expand to 3%, 5%).
3. Broad Targeting: Only test this once your pixel is very seasoned with thousands of conversion events. You can't start with this.

Middle & Bottom of Funnel (MoFu & BoFu) - Retargeting:

This is where the real money is often made, and it's what props up your overall ROAS. A 1.0 ROAS on cold traffic is fine, even expected sometimes, if your retargeting is converting at 5.0+ ROAS. You need to look at the blended ROAS across the whole account. It sounds like you might be judging your success solely on cold traffic performance, which is a recipe for disappointment.

You need seperate, always-on campaigns to retarget people who've engaged but not bought. Structure it based on intent:

-> MoFu (Warm): People who visited your website, watched 50% of your video ad, or engaged with your social profiles. Show them different ads - maybe customer testimonials, reviews, or a different angle on the product. They're aware of you, now you need to build trust and desire.

-> BoFu (Hot): People who Added to Cart or Initiated Checkout but didn't buy. This is your most valuable audience. Hit them with ads that overcome final objections. A reminder, a small discount code ("Complete your order and get 10% off"), or messaging about fast shipping can work wonders here. These people were *seconds* away from giving you money. It's madness not to have a dedicated campaign talking directly to them.

I remember one campaign we worked on for a women's apparel brand where we saw a 691% return using Meta Ads and Pinterest Ads. A huge part of that success was a robust retargeting structure that captured abandonned carts and nurtured website visitors who were just browsing. Your cold ads bring people to the shop; your retargeting ads close the deal.

You probably should rethink what 'success' looks like...

This ties into the previous point. Scaling isn't linear. You can't just expect to double your budget and get double the sales at the same ROAS. It never works that way. Every time you increase spend, you are asking the platform to find more buyers. It will always find the easiest, cheapest conversions first. As you spend more, it has to work harder, show the ad more times, and reach less-interested people to find the next conversion. Therefore, your CPA will rise and your ROAS will fall. This is the law of diminishing returns in advertising.

Success isn't about avoiding this; it's about managing it. The goal is to scale to a new, higher level of spend while keeping the ROAS at a level that is still profitable for your business. Maybe a 4.0 ROAS at £50/day becomes a 2.5 ROAS at £500/day. If your margins support a 2.5 ROAS, that's a huge win. You have to know your numbers: what is your break-even ROAS?

Furthermore, you should be thinking about Customer Lifetime Value (LTV). If you acquire a customer at a 1.0 ROAS today, but they come back and make two more full-price purchases in the next six months, your actual return on that initial ad spend is much higher. Relying solely on the immediate ROAS from the first purchase can be very misleading. If you can increase your back-end value (through email marketing, repeat purchase offers etc.), you can afford to pay more to acquire a customer upfront, which gives you a massive advantage over competitors who are only focused on day-one profitability.

This is a major part of scaling. I remember one software client who was stuck, unable to scale using Meta Ads and Google Ads. A big part of the solution was improving their funnel to increase the LTV of each user. Once each user was worth more, we could afford to bid more aggressively and scale their ad spend significantly, unlocking a whole new level of growth for them.

You'll need a concrete plan to move forward...

Alright, that was a lot of theory. So what should you actually *do*? You need to stop the cycle of launching, hoping, and getting frustrated. You need to build a machine.

I've detailed my main recommendations for you in a table below. This is the kind of foundational thinking that needs to happen before you can expect to scale spend consistently and profitably.

As you can probably tell, this is a lot more involved than just boosting posts or launching a few ads. Building a truly scalable advertising engine that consistently delivers results takes a deep understanding of the platform, a disciplined methodology, and a lot of time spent analysing data and iterating. It's not just a set-and-forget task; it's a constant process of optimisation.

This is where expert help can make a huge difference. Instead of spending the next six months learning these lessons the hard way (and spending thousands on ads that don't work), you could work with a team that has already built these systems for dozens of businesses. We've seen these patterns play out time and again. For example, we helped an eLearning client generate $115k in course sales using Meta Ads, and we helped a subscription box achieve a 1000% ROAS also using Meta Ads. The underlying principles of structure, testing, and strategic scaling are universal.

If you'd like to go over your ad account and strategy in more detail, we offer a free, no-obligation initial consultation. We could have a look together at what you're doing now and give you some more specific, actionable advice on how to implement the kinds of structures I've talked about here. It might be really helpful to get a fresh pair of expert eyes on it.

Hope this helps give you a clearer path forward!

Regards,

Team @ Lukas Holschuh