168 example ads for 54 brands, and 195 YouTube videos and shorts breaking down ads.
A lot of brands have a cost per acquisition that's higher than they want. It's never truly one thing. So we ask what has to be true for your customer to buy.
We look at your ads and what your customers are saying. For example on Jope, a lot of owners say their dog had a limp or weak, shaky back legs, and no top ad is made for these people.
230 of those 877 owners tried another product first, so they have to believe this one is different. So this ad uses what university researchers found and a real customer's review.
We put money behind them and see what the data says.
We leave it on and make more.
We check the hook first, then the copy, and go again.
Baudie, Deodorant enhancer spray
Other ads I made for Baudie
Humann, SuperBeets heart chews
Other ads I made for Humann
Jope, Hip and joint chews for dogs
Other ads I made for Jope
Health y Sol, Beef tallow soap
Other ads I made for Health y Sol
Plunge, Cold plunge tubs
Other ads I made for Plunge
WildBird, Baby carriers
Other ads I made for WildBird
Happy reviews means 4 and 5 stars. Top ads means their Ad Library sorted by most impressions.
This is how I think about ads, before, while and after I make them. Click any one.
What has to be true for the shopper to buy? First off, they actually gotta believe it works. Reviews help, but a lot of brands are already doing that.
So what else? If they understand their problem, why it's happening and how the product actually helps, then more likely than not, they will buy.
Then we reverse engineer it. So we make ads that teach the things that actually matter.
The same idea is in Roger Martin and Dr. Benjamin Hardy.
People buy things for reasons. No one cares about getting a full day of iodine unless it helps them do something.
And most of the time, the reason why they're not buying is because they just don't believe it. So every claim you make, you must support it with evidence, very similar to a lawyer.
I got a lot of this from Eugene Schwartz, Drew Eric Whitman and Alex Hormozi.
It's never truly one thing. Sometimes it's the hook, the proof, the offer or the page. Sometimes it's an objection or a belief the market already has, and the ad has to handle that.
In the ad account, this is where I look first.
| If this is happening | I would look at | Who I learned it from |
|---|---|---|
| New ads get no spend | The creative | Spencer Pawliw |
| ROAS drops every time the budget goes up, or AOV is under $40 | The offer | Chase Chappell |
| Under 1 to 2% of visitors buy | The page | Chase Chappell |
| Meta shows a good ROAS but the business is not making money | The data and the margin | Blue Sense Digital |
| Total revenue holds but new customer revenue falls | Getting new customers | Blue Sense Digital |
Fixing the biggest problem first is from Eliyahu Goldratt and Alex Hormozi.
I collect the ads, the site and the reviews, and I sort who buys, the problems they feel and what they already tried.
It's not even me saying it. What is the data saying? What is the market already saying? And then we just translate that to ads.
For example, Canopy's sleep ads sell the sounds, but more of their buyers credit the moist air (163) than the sounds (135).
| Stage | What they already heard | What wins |
|---|---|---|
| 1 | Nothing like it, and the brand is first | Say the need or the claim simply, then prove it works |
| 2 | The claim, from others too | The same claim, bigger and more exact |
| 3 | Every claim at its biggest, and many tried a rival | A new mechanism, so the ad leads with how it works |
| 4 | Rivals copied the new mechanism | The mechanism made easier, quicker, surer or freed of an old limit |
| 5 | They believe no claim and no mechanism | Who the buyer is, more than the promise or how it works |
If you only target people who are already educated, it's a very small market pool. So I like to go very top of funnel, these skeptical buyers hearing about the brand for the first time. Spencer Pawliw's default is product aware first, and I go the other way.
And it's better to attack a problem they're dealing with right now than a potential problem.
237 live ads. Their top ad, "Stairs used to be nothing", runs as 54 ads.
877 happy reviews. 277 name a big dog, 230 tried another product first, 189 name a limp or weak back legs and 56 name stairs.
10,432 reviews from 9 other joint brands. In the 1,432 that say what came before, back legs or hips is 425 and stairs is 127.
Their top ads talk to the older dog on the stairs. No top ad is made for weak back legs or big dogs.
"I already tried joint chews. They do not work for my dog."
University researchers found glucosamine and chondroitin helped in only 1 of 9 dog and cat trials.
One buyer, one belief to change, and a real customer's image and review.
A fresh agent grades it blind, then I approve it or send it back.
I would run one CBO beside their stairs ads for 7 days. If it works, it works. If it does not, now you know.
Yes the concept is important, but the angle is what connects to it. Spencer Pawliw's example is an us versus them ad that can fail or spend $20,000 a day, and what changed was the angle.
A new angle can unlock a different market, but it's not always the fix. Gruns ran one greens gummy with ads for GLP-1 users, gut trouble, hair, fertility, kids and more.
I learned this from Spencer Pawliw.
1234Names one want they were born with, as a problem in their day.
"Dry and tight skin after showering?" That is comfort, in a moment they live every day.
They can see themselves using it and getting the want.
A real person in her bathroom with the bar in her hand, so you can see yourself there.
Why it works, and proof the seller didn't make.
Many body washes use sulfates, the same cleaner in laundry detergent. Then a real customer says the result.
The easy next step and the risk taken away.
Save up to 40% on bundles, and a lifetime guarantee.
I sorted these from Drew Eric Whitman's Cashvertising.
A lot of people have good headlines, but then they mess up on the actual explaining. Here's how I did it on Health y Sol.
| The line | The copy |
|---|---|
| The moment | "Dry and tight skin after showering?" |
| The cause they never heard | "Regular soaps & body washes use harsh chemicals like sulfates. The same cleaner in laundry detergent." |
| What it does to them | "Sulfates strip your skin's oil, leaving your skin dry & tight after you shower." |
| The fix, as the opposite | "That's why Health y Sol uses beef tallow! A natural fat that cleans without stripping your skin's oil." |
| The result, from a buyer | "My skin is not dry after the shower. I dont even need to put lotion on anymore." |
And I want one fact they did not know, so they say "huh, I never thought about that". If they would say "I already know that", I go back to the research.
| Field | What it means | Jope Ad 1 |
|---|---|---|
| Product | Hip and joint dog chews with UC-II | |
| Desire | The outcome they want, as "I want...", then what it looks like in their day | I want my dog to get up the stairs and go on long walks again |
| Mental movie | The scene the ad plays in their head | Your dog's back legs stop shaking, and your dog goes up the stairs and wants a longer walk |
| Sub-avatar | The person, by experiences and beliefs | An owner who gave joint chews for years and saw nothing |
| Angle | The reason this person buys | The old chews did not work, and this one uses something different |
| Mechanism | Why this can work when the last thing did not | Glucosamine and chondroitin helped in only 1 of 9 dog and cat trials. UC-II is the same type of collagen in a dog's joints |
| Authority | Who makes the claim believable | University researchers, and Paola G. with her 12 year old boxer |
| Awareness | What they know before the ad | Solution aware, because they already buy joint chews |
| Market stage | How many claims they already heard | Stage 4 of 5, because rivals already sell this mechanism, so the proof makes it surer |
| Positioning | Who the enemy is, never the customer | Glucosamine chews |
| Urgency | The reason to act today, or the risk taken away | 90-day money-back guarantee |
| Concept | The big idea being tested | A teaching ad with a real customer's image and review |
What does it look like in their day? "I want to feel like me again" doesn't say, but being tired all day when nothing much happened does. And "25 to 35 male in California" changes no line of copy, so I look at what they went through.
Most of this is from Spencer Pawliw.
A hook that obviously will not work is "lose 23 dB at the club", because how can someone really picture that?
Out of the headlines I liked, 17 of 23 name the buyer, their problem or the result. The most common reason I rejected one was that it did not say how, 7 of 19.
If no one in the reviews uses a word, I swap it for one they use. A new fact they never heard stays.
"Peaceful" is in 6 of Canopy's happy reviews and sleeping better is in about 70, and I still picked "sleep peacefully", because you can see it.
The crossed out line is Claude's first draft, and the white line is what it became after my notes.
Good for a product that is easy to understand.
The old way next to this one, so you can see the difference.
The headline says the problem the way they say it.
Speaks only to one group with its own reason to buy.
If the market heard every claim, lead with what's different inside.
A real customer already says it better than I can.
One real number can do the selling.
It feels like content and not an ad, like a Notes screen.
Founder ads work so well because of the authenticity. So I start from what the brand already has, and that picks the framework.
| If the brand has | I start with |
|---|---|
| A founder who had the buyer's problem | The founder tells how they had the problem |
| A founder who is not the buyer | The founder tells who they built it for |
| A niche buyer you can name in one line | A short origin story, then the product |
| A cause the buyer already holds | The founder says what they stand for |
| Buyers who tried the category and felt little | Why the others fall short, and what's different here |
| A harmful material in the category | Name the material, then show the swap |
| Cheap copies | Show the copies failing on camera |
| A price that makes buyers stop | The founder answers the price objection |
A winning ad continues to bring your brand profitable sales consistently, so it's getting significant ad spend and hitting your KPIs. Most impressions does not mean the highest ROAS, because views does not equal cash.
| Type | What it is | What I do |
|---|---|---|
| Breakthrough | Big spend and hits KPI | Leave it on and make more on that angle, because it is the only one that raises how much the account can spend |
| Spend winner | A big share of spend, misses KPI | Leave it on unless far under KPI |
| KPI winner | Small spend, ROAS at or above the campaign's | Leave it on, because it can become a breakthrough in week 2 or 3 |
| Loser | No spend, no KPI | Move it out after a test of at least 7 days |
I learned this mostly from Spencer Pawliw and Blue Sense Digital.
If someone is cold, they need a lot more education than a product page gives them, so why not add a simple listicle.
I look at the offer too. On Ryker, I don't believe buy one get 30% off is the best they can do, so I would test a triple bundle.
I thought about what I would manually do to make a winning ad, then I built that on Claude Code. I tried to one-shot it and it just never works, so I work with it the whole time.
$ py review_words.py "Humann" "blood moving" "circulation" "more energy" "afternoon" 3463 different happy reviews 0 blood moving <- 0: our word, not theirs 66 circulation 234 more energy 57 afternoon
The ad said "blood moving", but 0 people say that and 66 say "circulation". So I changed the line.
| What Claude does | What I do |
|---|---|
| Pulls the ads and the reviews, and counts what people say. | I decide which gap matters and who the ad is for. |
| Writes first drafts of the copy. | I rewrite the lines that do not sound human. |
| Builds the ad in code from the brand's real images. | I tell it what to change until it's exactly what I would launch. |
| A fresh agent grades each ad blind. | I approve it or send it back. |
And when the output is weak, I fix the system and test it again.
No before and after AI, because that is essentially lying and illegal, and no UGC avatars or fake UGC.
I stopped using AI images of a product, a customer or a result. I still use one for the problem itself, like a red cheek.
Every number and review is real. Each claim has a source, and weak ones get flagged for the brand.
I break down ads in public, so you can understand my way of thinking.
Why every claim in your ad needs hard proof
Why your headline should name your buyer and their problem
Why your ads should make one claim and prove it
Why your ad should teach the cause, then show the easy fix
Why your ads need to teach why your product works
Why your hook should name what your buyer already seesIf I was hiring someone, one of the main things I would look for is who do you learn from.
Helping businesses get more customers cheaper from their ads
Every single time I do new batches, I'm always improving.
If you think I'm wrong, say it. We work together on this and it's not a one-man army.
I can show you what your customers say, what your ads say and the first ad I'd test.
johnsontransales@gmail.comHope you guys have a wonderful day.
Hey, my name is Johnson Tran.
YouTube
Instagram
X
LinkedIn