
Fable 5.2 made its own version of our VSL edit and I liked the Opus 5.5 version more. For planning bigger builds though, Fable catches way more of the little details.

Saving businesses from ad agencies and junior talent 1 Unicorn Marketer at a time. Click link to get matched with a Unicorn Marketer today!

Fable 5.2 made its own version of our VSL edit and I liked the Opus 5.5 version more. For planning bigger builds though, Fable catches way more of the little details.

In case you weren't aware there are still business owners who refuse to use Slack in 2026...

Burger King once got 82,771 people to delete 233,906 Facebook friends for a free Whopper in less than a week and the high IQ move wasn't the burger. Every friend you dumped posted to your News Feed as "Caroline sacrificed Josh Lowensohn for a free Whopper." and the dumped friend got a notification telling them they'd been traded for a free hamburger, back when Facebook never told anyone they'd been unfriended. So every redemption was also an impression and the people taking the deal were doing all of the distribution for free. Facebook said the app "ran counter to user privacy" and asked them to change it, so Burger King killed the campaign instead and later that year it won a Titanium Lion at Cannes. Jonah Berger calls this Public in his book Contagious and the basic idea is that anything that leaves a visible trace every time it's used ends up advertising itself. If you run a referral or share-to-unlock offer, test a version where finishing it automatically creates something the customer's friends will see against your current optional share and watch new visitors per redemption and blended CAC. TLDR: a lot of referral programs make sharing an optional last step, which is the opposite of what made this work.

The biggest unlock from using 70% of a week of Claude in 24 hours wasn't a better prompt, it was running a bunch of sessions at the same time and bouncing between them while they worked.

A creative format you should 100% have in your mix is 7-15 second short video clips/GIF style ads. Here’s 2 recent ones we are running for @UniMarktersHQ 👇

Cracker Jack put a prize in every box for over 100 years and then in April 2016 Frito-Lay swapped it for a sticker you had to scan with an app. The prize was the sales strategy all along. Their 1915 counter card told store owners that once fans heard about the baseball cards inside "they will start making collections at once" and "that means you will sell more Cracker Jack." By 1916 the box had PRIZE printed on the side, so for a nickel you were buying the chance to find out what was inside. The 2016 version asked you to download the Blippar app and scan a sticker to unlock four baseball games and one fan summed it up on their Facebook page with "An app is not a prize." This is called the motivating-uncertainty effect and the basic idea is that people will often work harder for a reward they aren't sure about than for a guaranteed one that's worth more (in one study more people finished drinking 1.4 liters of water in two minutes for a coin flip between $1 and $2 than for a sure $2). High IQ move for DTC brands is to put a small unannounced surprise in a random half of first orders and compare repeat purchase rate at 60 and 90 days against the other half. And if your bonus needs an app download or a code to redeem, measure every step because each one is a place people quit.

Most people use AI like a search bar, so it only knows what they typed into it a few minutes ago. I flipped that and started letting Claude interview me and the deeper questions about how I make decisions have been way more useful than the informational ones.

A screenshot of a great ROAS doesn't count for anything when a marketer applies to become a Unicorn Marketer. Anybody can crop a dashboard down to their best week (or the one client where everything went right), so we only count results we can confirm. Stage 3 of our vetting is a 79-point test project our leadership team designs and it's where most applicants wash out. That's a big part of why fewer than 1% of the marketers who apply make it onto our roster. It's also worth borrowing even if you never work with us. Next time someone shows you a screenshot of their results, ask them to share their screen and walk you through the account it came from. You'll find out pretty quickly how much of the story holds up.

If you want your ads to do the selling without a wall of copy, study the Morton Salt girl. Most people think she's just a cute mascot, but the original 1914 ad is really a product demo. According to Morton's own history they tried "Flows Freely", "Runs Freely" and just "Pours" before anyone brought up the proverb, so the line everybody knows was something like the fifth attempt. The copy under the picture says "No need to shake it. No need to bake it. It Pours." and it only takes about a dozen words because the picture already did the explaining. Most brands explain what their product does when they could just show it doing the thing. If your software is faster, put a timer on the screen and run both because "Report built in 40 seconds" beats "save time on reporting" pretty much every time.

Timex strapped a watch to an outboard motor propeller on live TV and the propeller came back up without it, which is still one of my favorite lessons on proof. Jewelers didn't want a cheap watch with a 50% markup (vs 100% on other watches) so Timex went to drugstores and cigar stands with a counter machine that dunked a ticking watch in water and hit it with a hammer. Then they put the same idea in their ads: Mickey Mantle hitting a baseball 50 times with a Timex taped to his bat, a watch towed more than 27 miles behind a boat and later John Cameron Swayze freezing watches in ice and strapping them to cliff divers on TV. When the outboard test failed live, Swayze apologized, said it had worked in rehearsals and promised they'd fasten it better next time. That failure is the spot people remember because a watch that could break on live TV made every watch that didn't break way more believable. AKA costly signaling: when buyers can't see quality before they buy, the best proof is a test a worse product couldn't afford to take in public. By 1962 one out of every three of the 23M watches sold in the U.S. was a Timex. High IQ move for anyone running ads: if your product's quality is hard to see, test a creative that opens on a real stress test with the numbers on screen vs your claim-led ad and compare hook rate and CPA at the same spend.

The biggest time saver editing our new VSL with Claude had nothing to do with my prompts. And once the style was locked in, it turned that same VSL into 6 shorter ad versions the next day and it went way quicker.

Juicero's founder said their $699 press had four tons of force, "enough to lift two Teslas," but that's not why anyone buys juice. The packs that went in it cost $5 to $8 each, so I'd argue the business was really the packs and the machine was just what you had to own to buy them. Even the CEO's defense of the connected press was that it let them remotely disable packs during a spinach recall, which helps the company more than the person making juice. Before your next expensive build, write down every feature you're paying for and what the customer gets that they couldn't get without it. If the answer is a slightly better version of what they can already do by hand, your price can't stand on that feature. The truth is your buyer doesn't care how hard the thing was to build, they care what they'd have to do without it.

If you hired a marketer to run your ads and they still need your sign-off to pause a losing ad, test a new headline or move budget between ad sets, you didn't really hand anything off. You just put yourself between them and the account. A lot of owners end up here without meaning to. They've been burned before (or they're trusting someone new with real money) so they want to see everything first, which feels like the responsible thing to do. The problem is most of the decisions that make an ad account better are small, frequent and time sensitive. An ad that's losing money on Tuesday keeps losing money while it sits in your inbox until Friday. And a marketer who has to wait three days for every yes is going to make a lot fewer ads. What I like to do instead is write down which decisions they make on their own, which ones they make and tell you about after and which ones need your yes first. Half a page is plenty. Here's roughly how I'd split it for most businesses: Just do it: pausing ads that are clearly losing, testing new versions of ads that use claims and offers you've already approved and shifting spend between ad sets inside the budget you agreed on. Do it and tell me: raising or lowering the daily budget within a range you set ahead of time, launching a new campaign or turning off an audience or placement. Ask me first: any new offer, discount or guarantee, any claim about your product you haven't approved, any change to your landing page or checkout and anything that pushes spend past that range. The ask-first list is where your real risk lives. Those are the decisions that can cost you money you can't get back, a promise to customers you can't keep or a problem with your brand. Pretty much everything else is the job you hired them to do. If you're working with someone new it's fine to start with a longer ask-first list and move things into just do it as they earn your trust. What does not work is leaving it undefined, because then every decision either waits on you or gets made without you. Neither one is what you're paying for.


One of the most expensive numbers in a lot of ad accounts is a blended ROAS that looks totally fine. Say your account did $60k in sales last month on $20k in spend. That's a 3x and nobody's worried. But break the month down by campaign and you might find one did $45k on $10k, another did $9k on $4k and the third spent $6k to bring back about $6k. Once you take out product costs, shipping and fees that third campaign lost money on pretty much every order. The average hid it because the best campaign was paying for the worst one. The same thing happens one level down. A campaign that looks healthy can be hiding a placement, an age range or a single product that's dragging it down. In Ads Manager you can see it in a couple minutes by clicking Breakdown. The way I'd check it is to pull the last 30 days and put every campaign next to the ROAS you actually need to break even. Look for anything spending real money under that line. Then do the same with placements and products inside your biggest campaign. I wouldn't cut anything off a week of data or a handful of sales because small segments swing all over the place. Some prospecting spend also runs at a lower ROAS on purpose because it feeds the retargeting that closes the sale. But a big chunk of spend that's sat under break-even for a month with real volume is usually the first place I'd look for profit.

If you're about to send the same Black Friday discount email as everyone else, look at what Patagonia did in 2011 first. The part most people miss is that it wasn't a wild swing. Years earlier they used "Don't Buy This Shirt" in a catalog essay and said it got a strong response, so the Times ad was a proven idea on a bigger stage. They were honest about it too. Their own blog post the day it ran said "we are a growing business", which is way more believable than a brand pretending it doesn't want your money. Two more ways to steal it on top of the plays in the video: give customers a way to get more out of what they already bought (Patagonia offered to list your used gear on their site for free when you sold it on eBay) and test your contrarian line on a subject line or one ad set before you put real money behind it. The truth is the loudest day of the year is also the easiest day to get ignored.