If your underperforming Meta ads are burning money, the bug usually isn’t your targeting, your budget, or your pixel — it’s that you’re not entering the conversation already running in your customer’s head. Here are the seven fixes that move ads from leaking cash to printing it.
Most “fix your ads” advice you’ll find online is technical hygiene — audiences, budgets, pixel setup, creative refresh frequency. All useful. However, none of it matters if the message is wrong.
When we audit underperforming Meta ads, the same pattern shows up roughly nine times out of ten: the technical layer is fine. The ad just isn’t speaking to the question the customer is actually asking themselves at 11pm. They scroll past, because nothing on screen feels like it’s about them.
Fix that — really fix it, not just tweak headlines — and the same audiences, same budget and same product start converting. (If you want the broader view of where stores leak revenue without spending more on ads, start with our homepage thesis.)
Here are the seven fixes, in the order we’d run them.
Fix #1 — Stop writing ads from your buyer persona doc. Listen on Reddit instead.
Your “ideal customer profile” document tells you they’re “women 30–45 who care about sustainability.” That doesn’t help you write an ad. What helps you write an ad is the exact words your customer used last week when they were ranting about a product that didn’t work.
That’s not in your persona doc. Instead, it’s on Reddit.
So pick two or three subreddits where your customers actually hang out — not where you wish they did, where they actually go. For a sensitive-skin skincare brand, it’s r/SkincareAddiction. For a BBQ store, it’s r/smoking. For an eBike brand, it’s r/ebikes. However, if you’re not sure of yours, run your product term through findquestions.com — it scans Reddit for the real questions being asked, and gives you the subreddits to monitor.
Then read. Don’t skim. Read the rants, the 1am posts, the “I’ve tried everything and nothing works” threads. Copy every frustrated phrase into a spreadsheet, word for word.
The principle, and you’ll have seen it in every serious copy book: enter the conversation already taking place in your customer’s mind. When your ad uses their exact words back at them, they don’t think “this is an ad.” They think “it’s like they’re reading my mind.” Scepticism drops. Clicks rise.
Fix #2 — Boil it down to one burning question.
After Fix #1 you’ll have a spreadsheet of fifty-plus phrases and complaints. That’s too many for an ad. So compress it.
First, group the phrases into themes. Then find the five-to-ten questions your customer asks over and over. Finally — and this is the step most people skip — find the one that comes up the most, the one they’re most frustrated about, the one nobody has answered properly.
- For a sensitive-skin brand it might be: “Why does every ‘natural’ moisturiser still make my skin worse?”
- A BBQ buyer might land on: “How do I stop wasting weekends on dry, ruined briskets?”
- And an eBike customer’s nightmare: “Will the battery actually last my commute, or will I be stranded again?”
That single question becomes the foundation of every ad you write, every headline you test, every landing page you build. Most underperforming accounts don’t have one. That’s why nothing in the account feels connected to a real human problem.
Fix #3 — Write ads that answer their question, not advertise your product.
This is where most ads die. They list features. “3 ingredients. Made in Adelaide. Cruelty-free.”
Features don’t sell. Instead, the translation of a feature into the answer to your customer’s burning question — that’s what sells.
- “3 ingredients” is a feature. “Nothing in here to trigger your sensitive skin” is the answer.
- “500Wh battery” is a feature. Translate it to “You’ll get home tonight without pedalling” — that’s the answer.
- “Australian-made tallow” is a feature. But “your skin can finally absorb something it actually recognises as food” is the answer.
Every line in your ad copy should pass the same test: does this answer their question, or am I just describing the product? If it’s the second, rewrite it.
Fix #4 — Match the awareness stage. (This is the single biggest mistake we see.)
Eugene Schwartz, the godfather of direct-response copy, mapped customers into five stages of awareness:
- Unaware — doesn’t know they have a problem.
- Problem-aware — knows they’ve got a problem; doesn’t know there’s a solution.
- Solution-aware — knows there are solutions; doesn’t know about yours.
- Product-aware — knows you exist; hasn’t bought yet.
- Most-aware — ready to buy; just needs the nudge or the offer.
Cold Meta traffic — the people you’re paying to reach for the first time — almost all sit at unaware or problem-aware. They are not searching for you. They are not ready to be sold to. In fact, they’re scrolling past photos of someone’s lunch.
Show Image
Where underperforming Meta ads break this rule
The most common cause of underperforming ads is this: pitching your product to someone who doesn’t yet know they have the problem your product solves. It feels like shouting “BUY MY THING” at a stranger on the street. They scroll.
Instead, lead with the problem. Agitate it until they go “oh, that’s me.” Then introduce that there’s a solution. Then tell them you’re it. That’s the sequence. Get the stage wrong and nothing else in the account can save you.
Fix #5 — Build a landing page that walks them through the rest of the journey.
Your ad got them to click. Now your landing page has to carry them from problem-aware → solution-aware → product-aware → ready-to-buy in a single page. But if your ad sends them to your homepage, or to a generic product page, you’ve broken the chain — they have to do the work themselves, and they won’t.
A landing page built for cold Meta traffic should follow this skeleton — it’s the same skeleton we walk through in our 90-day path:
- Headline — answers their burning question, in their words.
- Agitate — show you understand the frustration. The failed past attempts. The wasted money. The “I’d given up.”
- Introduce the solution category — frame the kind of fix that actually works.
- Introduce your product as that fix — translate features into answers, the way Fix #3 said.
- Proof — reviews, before/afters (framed as customer-shared experiences for compliance), real photos.
- The offer and one clear call to action.
One page. One job. One CTA. If your ads point anywhere more scattered than this, you’re paying for clicks that bounce.
Fix #6 — Mine the Meta Ad Library. Model winners. Don’t copy them.
You don’t have to guess what hooks will work in your category. Your competitors and the best brands in the world have already spent millions finding out — and the data is public.
So open the Meta Ad Library. Search your category. Look beyond Australia — the best operators in your niche worldwide. Then look for the ads that have been running the longest. This is the whole trick: brands kill ads that lose money, fast. Therefore, if an ad has been running for six months, it’s because it’s making them money. Longest-running = proven winner.
Show Image
Find three winners. A mix of video and static.
Then break each one down:
- What’s the hook in the first three seconds?
- Which problem does it lead with?
- How does it bridge to the product?
- Where’s the offer?
Now rebuild that structure for your product, your customer, your burning question. You’re not copying their words — instead, you’re modelling their architecture. Same skeleton, your meat.
A fast cheat: paste the winning ad’s transcript or copy into ChatGPT and ask it to rewrite the same structure for your product and your burning question. Then make it sound like a real human wrote it, not a chatbot.
Fix #7 — One concept per asset. One concept per ad set. (Mess this up and Meta can’t help you.)
A “concept” is built from four elements:
Concept = Angle × Persona × Format × Offer
- Angle — the hook (e.g. “3 ingredients”).
- Persona — who it’s for (“the eczema veteran who’s tried everything”).
- Format — video, static, carousel, UGC.
- Offer — what they get, and at what price.
Show Image
Change one of those four and you have a new concept. Same four, same concept.
Here’s the rule almost everyone gets wrong since Meta’s Andromeda update: one concept per ad set. If you put two different concepts in the same ad set, Meta’s targeting can’t tell which audience each one is for. So it tries to serve both to everyone, both underperform, and you blame the creative.
Therefore, each ad set holds three creatives that are variations of the same concept. A genuinely different concept? That’s a new ad set, not a new creative. This single fix cleans up more underperforming Meta ads than anything else we do.
Why most underperforming Meta ads share the same root cause
If you read back through, the technical fixes sit at the bottom of the list (one concept per ad set, Meta Ad Library). The fixes that move the needle most sit at the top — listen, find the question, answer it, match the stage.
Underperforming Meta ads almost never have a tactical problem. Instead, they have a message-to-customer mismatch. You’re talking; the customer isn’t listening, because nothing you’re saying feels like it’s about them.
So: listen first. Write second. The technical layer carries you the last 20%.
Want us to find the leak in your underperforming Meta ads?
If your ads are running and not converting, the bottleneck is in one of these seven places — usually one or two, specifically. We’d rather show you which than have you guess.
So book a free 20-minute revenue audit. We’ll look at your ads, your tracking and your landing page in front of you, and tell you which of these seven is costing you the most right now.
cortexlabs.digital — turning the traffic you already have into more sales.