Solved: Adset Overspending After Duplicating a Winner

Hi there,

Thanks for reaching out!

Saw your issue with the new ad set overspending and driving some rubbish traffic. It’s a really common problem, and tbh, it’s one of the most frustrating things to see happen in an ad account. You think you've cracked it with a 'winner', you duplicate it to scale up, and then it just burns through your cash. I'm happy to give you some of my initial thoughts on why this is probably happening and, more importantly, a better way to think about structuring your campaigns so this doesn't keep happening.

The short answer is that the problem isn't really about this one misbehaving ad set. It's usually a sign of a deeper issue in the overall strategy, particularly in how you're defining your audience and what you're asking the algorithm to do for you. Let's get into it.

Why duplicating a 'winner' is often a losing game...

Right, first things first. Let's tackle this idea of a "winning interest adset". I see this all the time. Someone finds an audience that works, gets a few good days of results, and then assumes it's a magic bullet they can fire over and over again. Unfortunately, it just doesn't work like that. Duplicating an ad set, even one that was performing brilliantly yesterday, doesn't guarentee success today.

Think about it like this: the digital advertising space isn't a static pond; it's a fast-flowing river. Here's a few reasons why your 'winner' clone fell flat on its face:

• -> Audience Saturation: Even if the interest group is millions of people, Meta's algorithm is designed to find the lowest-hanging fruit first—the people most likely to convert within that audience. Your original ad set might have already picked off the best prospects. When you launch a duplicate, you're often forcing the algorithm to dig deeper into the less-interested parts of that same audience, or you're just competing against yourself for the same people, driving up costs.


• -> The Algorithm Forgets: Every time you create a new ad set, even a direct copy, you're wiping the slate clean. It enters a brand new 'learning phase'. It has none of the data or momentum of the original. It has to start from scratch, figuring out who to show the ad to, what time of day works, which placements are best. This initial period is always volatile and expensive.


• -> Competitor Mayhem: You're not advertising in a vacuum. The auction is a live environment. A new competitor could have just launched a massive campaign targeting the exact same interest, flooding the auction with high bids and driving your costs through the roof. What worked when you were one of three companies bidding for that audience's attention won't work when you're one of thirty.


• -> Creative Decay: People get sick of seeing the same ad. Even the best creative has a shelf life. The ad that was fresh and compelling last week might be old and annoying this week. Duplicating the ad set just means you're showing the same tired creative to a potentially tired audience.

So, the big takeaway here is to stop thinking of ad sets as fixed assets. They are temporary vehicles. The underlying targeting might be sound, but you can't just set it and forget it, and you certainly can't expect a simple copy-paste to work every time. It’s a signal that you need a more robust system for testing and scaling, not just cloning.

You'll need to understand how Meta spends your money...

Now let's talk about that horrible first hour. The overspend and the "horrible traffic" are actually two symptoms of the same thing: the Meta delivery algorithm doing exactly what you told it to do.

I’m going to bet you're using the default bid strategy, which is 'Lowest Cost' (or 'Highest Volume' as it's sometimes called). When you tell Meta to get you the lowest cost results, you're giving it one primary command: "Spend my entire daily budget and get as many of the objective I've chosen as possible."

The algorithm doesn't care about pacing. It's not trying to spend your £40 budget gently over 24 hours. If it thinks it can find a pocket of cheap impressions or clicks at 9 AM, it will dump your entire budget in that first hour to hit its goal. This is what causes that initial surge of spend. In the learning phase, this is even more pronounced because the algorithm is in a frantic state of exploration. It's testing all sorts of user segments, placements, and times of day, and that testing costs money—your money.

This aggressive exploration is also why the traffic feels "horrible". The algorithm is testing the fringes of your audience. It's showing your ad to people who are *maybe* a good fit, just to gather data. So you get a flood of low-quality impressions and maybe some mis-clicks from people who aren't your ideal customer. It's the price of data.

So what can you do? You have a few options:

• -> Ride it out: Sometimes, you just have to let the learning phase run its course. It typically needs about 50 conversion events within a 7-day period to stabilise. That first hour or even first day is never representative of how the ad set will perform long-term. Panicking and switching it off after an hour is one of the biggest mistakes people make. You've just paid for the most expensive data and then thrown it away without letting the algorithm use it.


• -> Use a different bid strategy: If you need more control, you can switch to a 'Cost Cap' or 'Target Cost'. With a Cost Cap, you tell Meta, "Don't bid more than £X for a conversion." This prevents those wild swings in cost but comes with a risk: if your cap is too low, your ad might not get delivered at all. It's a way of telling the algorithm you prioritise cost stability over volume. This can be a good way to manage things if you know your numbers inside and out.

The key here is to understand that the system is designed for volatility at the start. The overspend isn't a bug; it's a feature of the learning phase on a Lowest Cost bid strategy. But this leads to a much bigger, more important question. Is the traffic really horrible, or is your offer just not resonating with them?

I'd say your ICP is a Nightmare, Not a Demographic...

This is probably the most important part of this whole letter. You said you know the "targeting was fine" because it was a winning *interest*. Tbh, this is where most campaigns fail before they even launch. Targeting an interest like "Shopify" or "Small Business Owners" is lazy. It's a terrible proxy for intent.

Why? Because the "Shopify" interest includes store owners, yes, but it also includes developers who build Shopify apps, marketing agencies who serve Shopify clients, journalists who write about e-commerce, and millions of people who just read an article about Shopify once. You're paying to reach a huge, unfocused blob of people in the hope that a tiny fraction of them are your actual customers.

You need to stop thinking in demographics and interests and start thinking in *pain*. Your Ideal Customer Profile (ICP) isn't a collection of data points; it's a person wrestling with a specific, urgent, expensive, career-threatening nightmare. Your job isn't to find people who *like* Shopify; it's to find people whose Shopify store is giving them sleepless nights.

Let's make this real.
Instead of "Companies in the finance sector with 50-200 employees"...
Your ICP is: "A Head of Finance at a Series B tech company who is terrified she's going to have to explain to the board why their AWS bill has unexpectedly doubled this quarter, putting their runway at risk."

Instead of "People interested in handcrafted jewelry"...
Your ICP is: "A 35-year-old professional woman who is frustrated that all her friends seem to have unique, story-driven accessories while she's stuck wearing the same mass-produced stuff from the high street. She wants to feel unique and express her personality but doesn't have time to trawl through Etsy."

See the difference? The nightmare is tangible. It's emotional. And once you have defined that nightmare, your entire advertising strategy becomes clear. You're not just selling a product; you're selling the solution to their specific hell. This clarity allows you to craft a message they simply can't ignore.

Crafting a message that hits the nerve

When you know their pain, your ad copy writes itself. You stop talking about features and you start talking about feelings. You use frameworks that work.

For a service business, you use Problem-Agitate-Solve.

• Problem: Are your cash flow projections just a shot in the dark?
• Agitate: Are you one bad month away from a payroll crisis while your competitors are confidently raising their next round?
• Solve: Get expert financial strategy for a fraction of a full-time hire. We build dashboards that turn uncertainty into predictable growth.

For a B2B SaaS product, you use Before-After-Bridge.

• Before: Your AWS bill just arrived. It’s 30% higher than last month, and your engineers have no idea why. Another fire to put out.
• After: Imagine opening your cloud bill and smiling. You see where every dollar is going and waste is automatically eliminated.
• Bridge: Our platform is the bridge that gets you there. Start a free trial and find your first £1,000 in savings today.

This is what separates the campaigns that get ignored from the campaigns that build empires. When your ad speaks directly to someone's nightmare, the targeting almost becomes secondary. The right people will self-select because you're the only one in their feed who seems to truly understand their problem.

We'll need to look at how you pay Facebook to find non-customers...

This next point is critical and it's another massive mistake I see people make constantly. What is the objective of your campaign? Is it "Traffic"? "Engagement"? "Brand Awareness"?

If it is, I can tell you right now you are actively, deliberately paying Facebook to find you the worst possible audience for your product. That sounds harsh, but it's the literal truth of how the algorithm works.

When you set your campaign objective to "Traffic," you command the algorithm: "Find me the people inside my targeting who are most likely to click a link, for the lowest possible price." The algorithm, being the obedient machine it is, goes out and finds the compulsive clickers. The people who click on everything but buy nothing. Their attention is cheap because no other advertiser who wants to make a sale is bidding for them. They are low-value users, and you've just asked to be served a plate full of them.

The same goes for "Reach" or "Brand Awareness." You're telling the algorithm to find the largest number of people for the lowest cost. It will seek out users who are cheap to show ads to precisely because they never engage, click, or buy anything. You are optimising for non-customers.

The only way to find actual customers is to optimise for the action you ultimately want them to take. If you want leads, your objective must be "Leads". If you want sales, your objective must be "Sales". This is non-negotiable.

"But my pixel doesn't have enough data!" I hear you cry. I don't care. Even with zero data, you must tell the algorithm what your goal is. It's infinitely better to get 2 expensive, high-intent conversions than 200 cheap, worthless clicks. The conversion objective tells the algorithm what kind of people to look for. Over time, it will get better and better at finding them. If you start with a "Traffic" objective, you are permanently poisoning your pixel data with the behaviour of non-buyers.

Brand awareness is a byproduct of making sales and having a great product that solves a real problem. It is not a prerequisite for making a sale. Switch your campaign to optimise for a final conversion event, and I promise you the quality of your traffic will improve dramatically, even if the volume drops at first.

You probably should structure your account for scaling...

Okay, so let's tie this all together into a structure that actually works. Instead of randomly duplicating ad sets that have a good day, you need a methodical system that separates your audiences based on their relationship with your brand. The classic way to do this is with a ToFu/MoFu/BoFu structure.

• ToFu (Top of Funnel): This is your cold traffic. People who have never heard of you before.
• MoFu (Middle of Funnel): This is your warm traffic. People who have engaged with you in some way (visited your site, watched a video) but haven't bought yet.
• BoFu (Bottom of Funnel): This is your hot traffic. People who have shown strong buying intent (added to cart, initiated checkout).

You should have separate, always-on campaigns for each stage of this funnel. Here’s how you’d populate them:

META (Facebook/Instagram) ADS AUDIENCE PRIORITISATION

ToFu Campaign (Objective: Sales/Leads)

This is where you do your testing. You'd have multiple ad sets, each targeting a different cold audience. But instead of just "interests", you test smarter:

1. Pain-Point Targeting: Based on your ICP Nightmare work. Target interests related to their problems. If you sell an accounting SaaS, don't target "Accounting", target interests like "QuickBooks", "Xero", or followers of financial publications that talk about cash flow problems. These are much more specific.
2. Lookalike Audiences: Once you have data, this is your goldmine. You test lookalikes in order of value.
   • -> Lookalike of your highest value previous customers
   • -> Lookalike of all previous customers
   • -> Lookalike of people who added payment info
   • -> Lookalike of people who initiated checkout
   • -> Lookalike of all website visitors

MoFu Campaign (Objective: Sales/Leads)

This is your retargeting campaign for people who've shown some interest. You show them different ads, maybe testimonials or case studies, to build trust.

• -> All website visitors in the last 30 days (exclude purchasers)
• -> People who watched 50% of your video ads (exclude purchasers)
• -> People who engaged with your Facebook or Instagram page (exclude purchasers)

BoFu Campaign (Objective: Sales/Leads)

This is for closing the deal. These people are on the verge of buying. You might show them an ad with a special offer or a reminder about what they left in their cart.

• -> People who added to cart in the last 7 days (exclude purchasers)
• -> People who initiated checkout in the last 7 days (exclude purchasers)

This structure is robust. It allows you to test new ideas at the top of the funnel without messing with your high-performing retargeting campaigns. It ensures you're showing the right message to the right person at the right time. It's how you build a predictable, scalable advertising machine instead of just getting lucky with one-off ad sets.

You'll need to know what a 'good' cost actually looks like...

One final piece of the puzzle. How do you know if your traffic is "horrible" or if your costs are "too high"? It's all relative. A £50 cost per lead might be a disaster for a business that sells £100 products, but it's an incredible bargain for a company with a customer lifetime value of £10,000.

You're focusing on impressions and immediate spend, which are vanity metrics. You need to focus on the metrics that actually matter: Cost Per Acquisition (CPA) and Lifetime Value (LTV). If you don't know your numbers, you're flying blind.

The real question isn't "How low can my CPL go?" but "How high a CPL can I afford to acquire a truly great customer?" The answer lies in its counterpart: Lifetime Value (LTV).

Let's calculate yours. This is a simplified model, but it's a hell of a lot better than guessing.

Once you know your LTV, everything changes. If your LTV is £10,000, you can comfortably spend up to £3,333 to acquire a customer and still have a very healthy business. Suddenly that £250 lead from a CTO on LinkedIn doesn't seem so expensive, does it? It looks like a bargain. This is the maths that unlocks aggressive, intelligent growth and frees you from the tyranny of cheap clicks and worrying about an ad set overspending by £15.

This is the main advice I have for you:

This is a lot to take in, I know. But the path out of your current frustration isn't about finding a magic button to fix one ad set. It's about building a professional, systematic approach to your advertising. I've detailed my main recommendations for you below:

Paid advertising, especially on platforms as complex as Meta, is far more than just boosting posts or duplicating what worked yesterday. It requires a deep understanding of auction dynamics, marketing psychology, and financial modelling. What I've outlined here is the foundation of a strategy that can deliver predictable, scalable results.

Implementing this is a lot of work, and it requires expertise to get it right. If you'd like to go through your account and build out a proper, robust strategy like this one, we offer a free, no-obligation initial consultation. We can take a look at your specific setup and give you a clear roadmap for what needs to be done.

Hope this helps!

Regards,

Team @ Lukas Holschuh