Ad Creative
Why the ad that embarrasses you usually outperforms the polished one
Short answerThe ad that embarrasses you is usually plainer, more specific and closer to how your buyer actually talks. Polishing tends to sand those things off. Facebook decides who sees an ad largely by how accurately it reflects a real person’s situation, so the rough, accurate ad often gets shown to better people than the beautiful, vague one, which is a big part of why ugly ads convert. Polish multiplies a good idea. It can’t replace one.
You know the ad I mean. You wrote it fast, the image is just a photo with a line of text on it, and the copy says the thing a little too bluntly. You almost didn’t launch it.
Then it beats the one you spent a week on.
Not every time. But often enough that I’ve stopped treating my own embarrassment as a sign an ad is bad. If anything, it’s usually a sign I said something specific, and specific is what most ugly ads that convert have in common.
What polishing actually removes
Think about what happens to an ad as it gets “cleaned up”. The blunt claim gets softened so nobody can object to it. The customer’s clumsy phrasing gets replaced with nicer words. The plain phone photo gets swapped for a designed graphic. Each step makes it look more professional.
Each step also makes it a little more general.
And general is the one thing that hurts you most right now. The way I read delivery today, when your ad is considered, Facebook is weighing whether the context of the message matches the lived context of the person seeing it. What decides that isn’t your production budget. It’s what the ad says, how it says it, and what it assumes about the reader.
That’s why I’m comfortable saying an ugly ad soaked in accurate lived context will often get shown to exactly the right people, while a beautiful ad with vague signals struggles to get distributed at all. I went into the image side of this in why your ad image is doing more targeting than your targeting is.
The raw material that makes an ad accurate is usually the stuff people say at 2 a.m. and never on a sales call. It’s not pretty language. Clean it up enough and it stops being theirs.
Even Meta’s polish can backfire
I learned this one the annoying way.
I had an ad account that was doing exceptionally well, and then one day it just flopped. A friend asked if I had “text improvements” turned off. That’s the setting that lets Meta’s AI adjust your ad text for you. Sure enough, the box was on. On the live post it can end up replacing your body copy with a sort of headline-looking thing.
I flipped it off. CTRs went through the roof, cost per click got better, and I got a better quality customer at a lower price.
So even when the polishing is done by the platform itself, with all the data in the world behind it, taking your plain words and “improving” them can turn ugly ads that convert into polished ones that don’t. These days I turn all of those AI enhancement features off by default and only test them later, on purpose. (There are some accounts where turning them all on is the only thing that works, which is exactly why I test instead of assuming.)
Why simple images keep winning for me
When I start a new account, I usually lead with images, and they’re about as unglamorous as it gets. A simple image showing the spokesperson, the founder or the ideal prospect, with a plain headline right on it.
That’s not a design philosophy. It’s that the image has one job, which is to tell the right person in half a second that this is for them. A designed poster can do that job worse, because it looks like every other ad and says less.
I’m also a noted video ad hater, and I’ll admit that bias bites me sometimes. Reels I’ve made myself and run on Instagram have done well. The common thread isn’t the format. It’s that each one says one thing to one kind of person.
If you want the practical version of this for images specifically, read why text on an image outperforms the beautiful photo you paid for.
Embarrassing isn’t the same as sloppy
This is where people take the idea too far, so let me name the exceptions.
Ugly isn’t magic, and plenty of ugly ads never convert. A rough ad with no idea behind it is just a rough ad. What makes the embarrassing one win is that it’s clear and specific, and those are the qualities you should protect, not the roughness.
Some markets also genuinely want to see the thing done well. In e-commerce, demonstration is how people buy, so a clear, well-shot product demonstration isn’t vanity. And sophisticated buyers want to understand how something works, so a sloppy explanation costs you there.
And once an idea is proven, production can help. Polish multiplies a working concept. What it can’t do is rescue a concept that was never going to land, and a nice-looking ad can hide a dead idea for weeks while you blame everything else. So test the idea cheaply first, find the one with a pulse, then spend the effort. If you’re fuzzy on what counts as an idea versus a variation, the difference between a concept and a hook is the place to start.
What to do this week
Take your best-looking ad that isn’t performing and make the embarrassing version of it. Same concept. Plain image of a real person or the product, one line of text on it that states the outcome in words your customer would use, and copy with the softening taken back out. Run it in the same ad set alongside the polished one for at least a full week before you decide whether ugly ads convert better for you.
Do ugly ads really work better on Facebook?
Not because they’re ugly. They tend to work because they’re plainer and more specific, and they don’t look like every other ad in the feed. An ad that accurately reflects the buyer’s situation often gets shown to better people. A rough ad without a clear idea behind it will still lose.
Should I turn off Meta’s AI creative enhancements like text improvements?
My default is to turn the AI enhancement features off at launch, then test them deliberately later. I had an account flop with text improvements switched on and recover when I turned it off. Some accounts do respond to having them on, so treat it as a test, not a rule.
Is it worth paying for professional ad creative?
Once you know which concept works, sometimes yes. Production can multiply a proven idea, and some markets, like e-commerce products, sell through clear demonstration. Before that, cheap and fast is better. Test several plain versions of different ideas and only spend on the one that has already shown it can sell.
P.S. When you catch yourself softening a line because it feels too blunt, ask whether it’s untrue or just uncomfortable. Fix the untrue ones. Leave the uncomfortable ones alone.
