Solved: Should I Update a Winning Ad or Replicate It?

Hi there,

Thanks for reaching out! Happy to give you some initial thoughts and guidance on your question. It's a really common dillema people face when they've got something that's working well.

The short answer is: you should definitly replicate the ad set. Don't edit the winning one.

Editing a live, performing ad set, especially changing the targeting, is a surefire way to reset the learning phase. When you do that, you're basically telling Facebook's algorithm to start from scratch. All the data and optimisation it's built up over the past two weeks gets thrown out the window. It might recover and perform even better, but it could just as easily perform worse, and you've lost your proven winner. It's a gamble, and usually not one worth taking.

By duplicating it, you create a new ad set that will, yes, have to go through its own learning phase. But your original, high-performing ad set keeps running untouched. It's your control, your baseline. This way, you can test your new refined targeting (the lower age and extra interest) against your known winner in a proper A/B test. After a few days, maybe a week, you'll have clear data to compare the two. If the new one is clearly outperforming the original, brilliant. You can then pause the old one and you've successfully improved your campaign. If it performs worse, no problem – you just pause the new one and you've lost nothing, your original winner is still ticking along nicely. It's the safest way to iterate and improve without risking what you've already built.

But this question actually brings up a much bigger, more important topic about how to structure and scale your campaigns for long-term success. Relying on a single winning ad is a fragile strategy. What happens if it stops working tomorrow? You're back to square one. A more robust approach involves building a proper campaign structur that targets people at different stages of their buying journey.

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

Right now, it sounds like you have one campaign with one ad set that's found a pocket of customers and is doing well. That's a great start, better than most people manage. But to really grow, you want to think like a funnel. In paid ads, we often break this down into three stages: Top of Funnel (ToFu), Middle of Funnel (MoFu), and Bottom of Funnel (BoFu).

Don't worry about the jargon, the concept is simple:

-> Top of Funnel (ToFu): This is your cold audience. People who've never heard of you before. This is where your current 'winning ad' likely sits. You're trying to find new customers. The goal here is broad reach and introducing your brand or product to people who might be interested based on their interests, demographics, or behaviours.

-> Middle of Funnel (MoFu): These are people who've shown some interest. They've engaged with your brand in some way but haven't taken that final step. Maybe they watched a part of your video ad, visited your website, or liked your Facebook page. They know who you are. The goal here is to re-engage them, build more trust, and nudge them further down the path to converting.

-> Bottom of Funnel (BoFu): This is your warmest audience. These people are on the verge of buying. They might have added a product to their cart but got distracted, or initiated the checkout process but didn't complete it. The goal here is to give them that final push to convert. This is often where you see the highest return on ad spend (ROAS).

A solid Meta Ads account should have separate campaigns targeting each of these funnel stages. Your ToFu campaign is for prospecting new customers (like what you're doing now). Your MoFu and BoFu campaigns are for retargeting. This structure is more stable because you're not just relying on finding cold customers; you're also actively working to convert the interest you've already generated. I remember working with an eCommerce client selling subscription boxes, and when we implemented this funnel structure for them on Meta Ads, they achieved a 1000% Return On Ad Spend. This strategy works because you're showing the right message to the right person at the right time, instead of showing everyone the same ad.

So, instead of just thinking about refining one ad set, you could think about building out this structure. Your winning ad can be the foundation of your ToFu campaign. Then, you can build a new retargeting campaign for MoFu/BoFu audiences. This is how you go from having one winning ad to building a predictable, scalable advertising machine.

I'd say you need a solid testing methodology...

This brings us back to your original question about testing. Duplicating your ad set is the first step in a good testing process. But what you're really doing is creating a controlled experiment. You need to be methodical about it. When you duplicate that ad set to test the new targeting, that's the *only* thing you should change. Don't change the ad creative, the copy, the headline, or the landing page. If you change more than one variable, you won't know which change caused the difference in perfomance.

Your current test is about the audience. But once you've found your best performing audiences, the next step is to test the creative. This is often where the biggest wins are found. You could test:

-> Different Ad Formats: Single Image vs. Video vs. Carousel.

-> Different Copy & Headlines: Test different angles. Does a benefit-led headline work better than one that creates urgency? Does long-form copy work better than short, punchy text?

-> Different Images/Videos: Even small changes in the visual can have a big impact. A different product shot, a different background colour, a different first 3 seconds of a video.

You should run these tests within dedicated testing campaigns or within your existing ad sets using Facebook's A/B test feature. The key is to be systematic. Let tests run long enough to get statistically significant results – don't make decisions after just a day or two and a handful of clicks. A good rule of thumb is to let an ad set spend at least 2-3 times your target cost per conversion before you decide if it's a winner or a loser. If an audience has spent £60 without a sale and your average cost per sale is £20, it's probably time to cut it.

This constant process of testing and iterating is what separates accounts that get lucky for a couple of weeks from those that deliver consistent results month after month. It's not about finding one 'winning ad' and being afraid to touch it; it's about building a system to constantly find the *next* winning ad.

You probably should think about your audiences...

Let's go a bit deeper on your plan to refine your targeting. Lowering the age and adding an interest is a good, specific hypothesis to test. But how do you decide which audiences to test in the first place? When I look at new client accounts, a common mistake is testing random interests that seem like they might fit, without a clear strategy or priority.

Here's how I'd typically prioritise audiences for an account, especially an eCommerce one, moving from the warmest (and likely best performing) to the coldest:

When you look at it this way, you can see there's a huge number of potential audiences to target beyond the one you're currently using. For new accounts with little data, you have to start at the bottom with Detailed Targeting, which is exactly what you've done. But as soon as you have enough data (you generally need at least 100 people in a source audience), you should be building out your retargeting (BoFu/MoFu) and Lookalike audiences.

When you're choosing interests for your ToFu campaigns, you need to be really specific. Let's say you're selling software for eCommerce store owners. Targeting a broad interest like "Amazon" is a bad idea. Why? Because millions of people who just *shop* on Amazon are in that audience. The vast majority are not your target customer. You're better off targeting more niche interests that your ideal customer is much more likely to have, like "Shopify", "WooCommerce", or pages for well-known eCommerce marketing gurus. The goal is to pick interests that have a high concentration of your target persona.

The real power comes from Lookalike Audiences. This is where you tell Facebook, "Here is a list of my 500 best customers. Go and find me millions of other people in the UK who look just like them." These almost always outperform interest-based targeting once you have enough quality source data. I remember one client selling online courses who generated over $115k in revenue in just a month and a half using Meta Ads, and a huge part of that success was systematically testing lookalikes of their highest-value students. It's a powerfull tool.

You'll need to understand your numbers...

One final thing to consider is cost. You've got a winning ad, which means your cost per conversion is probably good. But what *is* a good cost? It's impossible to say without knowing your business, but I can give you some general benchmarks from my experience.

The cost per result depends massively on what you're optimising for, who you're targeting, and where in the world they are. For something like a simple lead or a signup, here's a rough idea:

As you can see, costs can vary wildly. Traffic from developing countries is cheaper, but the quality of the lead or customer is often much lower. For actual sales, the numbers are different again because conversion rates are much lower.

These are just ballpark figures, but they give you a framwork for evaluating your own performance. If your winning ad is getting sales for £15 each in the UK, you're doing really well. If it's costing you £80, there's probably room for improvement. The key is knowing your numbers: your average order value, your customer lifetime value. As long as your CPA is comfortably below what a customer is worth to you, you have a profitable campaign you can scale.

When you start testing new audiences and creatives, you can use these numbers as a guide. If a new ad set is spending money and the CPA is trending way above your average, you know to cut it loose. If it's coming in below, you might have a new winner on your hands.

I know this is a lot of information to take in, especially when you started with a simple question. But hopefully it shows that there's a much bigger, more strategic world beyond just having one ad that works. To summarise, here's what I would be thinking about if I were in your shoes:

Managing all of this – the funnel structure, the constant testing of audiences and creatives, the data analysis – can feel like a full-time job, because it is. This is where getting expert help can make a huge difference. An experienced hand can build this structure out for you, implement a rigorous testing plan, and interpret the data to make the right decisions, ultimately scaling your results far beyond what a single winning ad can achieve.

We've taken clients from a position just like yours, with one or two things working, to having fully-fledged, scalable advertising systems that drive predictable growth. For example, we're currently working with an HVAC company seeing about $60/lead in a tough market, and I remember a home cleaning campaign we ran that got leads for as little as £5. It all comes down to applying these kinds of structured principles.

If you'd like to chat more about how we could apply this thinking to your business and help you build on the success you've already had, we offer a free initial consultation. We could have a proper look at your account together and map out a growth strategy. No obligation at all, of course.

Either way, you're in a great position with a winning ad. Just be sure to protect it while you explore how to get your next ten winners.

Regards,

Team @ Lukas Holschuh