I study what happens when hundreds of creators attack the same advertising problem. Writing about the patterns behind what wins.
NYC
Joined June 2026
- Tweets1.1K
- Following314
- Followers199
- Likes137
Most ads get killed or shipped on taste. That's not a review.
The ship-or-kill check I run:
1. What was this supposed to prove? If no one can say it, you're reviewing decoration.
2. Are you killing it for being unlike last month's winner, or for failing a test?
3. Who is this for that your last ad wasn't for? If you can't name them, it's a remake.
4. Name the risk it takes. If there isn't one, you're shipping comfort.
5. If it ran for two weeks, what would you learn? "It looks better" is not a lesson.
The room's favorite is often the safest cut. Ship the one that teaches you something.
Save it for Monday's review, and send it to the colleague who keeps asking for "just one more version."
Most ads spend all their craft on the open and throw the ending away.
The ending check I run before I ship:
1. The last line restates the idea in fewer words than the open.
2. Something from second 1 comes back. A callback beats a logo card.
3. The viewer leaves with one next step, not a feature list.
4. Mute it. The last frame still makes sense without the voiceover.
5. Cut the last 3 seconds. If nothing breaks, the ending wasn't doing work.
A strong open gets the thumb to stop. A strong ending gets the ad remembered.
Keep this one. If you know someone stuck on a "fine but forgettable" ad, send it their way.
Most ads script the product. The ones people remember script the buyer's next line.
The reaction test I run on every cut:
1. Write the line the viewer says out loud after second 3. If you can't, the ad hasn't landed.
2. That line is envy, relief, or "I need that." Never a feature list.
3. Put a person on screen who says it, or make the viewer say it in their head.
4. The product shows up as the reason for the reaction, not as its own segment.
5. End on the reaction getting clearer, not on a logo card.
If the only reaction you can write is "cool product," you don't have an ad yet.
Bookmark it. Then send it to the one person on your team who signs off on ads.
Nobody would have shortlisted a self-taught working mom for this job.
That's the whole case against hiring one name per brief. The winner is usually someone you'd never have picked.
Open the brief to many and let the work decide.
Send this to whoever still picks creators by résumé.
20 new edits of the same idea is not 20 new ads. It's one ad in 20 outfits.
What counts as a new creative:
1. A new person on screen
2. A new first scene
3. A new promise or reason to buy
4. A new format: skit, demo, testimonial, POV
5. A new objection it answers
What's just packaging:
1. A new headline
2. New music
3. A new color grade
4. A new CTA
5. The same script, read faster
Packaging squeezes a winner. It won't find you the next one.
Count your last 20 ads against this list. Then save it, and send it to whoever plans your tests.
Same math in ad creative.
One genius hire on a brief is one guess at the hook.
Fifty good creators on the same brief is fifty guesses, and the audience picks the winner, not the meeting.
Send this to whoever still hires one genius per brief.
Poker Face got 84 versions from 59 creators on the same brief.
The winning ad ran 21 seconds. #2 and #3 ran about 45.
Same product, same brief, and the shortest cut won by doing less.
One polished ad would never have shown that. Seeing many takes on the same brief did.
Send this to whoever always asks for one more round of polish.
Micro Drama has 40 paid spots and 33 entries so far. $5,000, open until Nov 19.
Base44 closes Thursday: $6,000, 50 winners, 137 entries.
Next: Reichman University $7,000 (Nov 8), then a mystery cybersecurity brand at $12,500 (Nov 10).
adarena.app
Most bad ads were bad briefs first.
6 lines I want in every creative brief before anyone opens an editor:
1. Who exactly is watching. One person, not a demographic.
2. What they believe right now, in their own words.
3. What they should believe 15 seconds later.
4. The one proof that makes that switch believable.
5. The one rule they're allowed to break. The winners I keep seeing broke one line of the brief on purpose.
6. What not to make. The obvious ad everyone in your category already ran.
If you can't fill line 2, no edit will save the ad.
Save this for your next brief, and send it to the person who writes yours.
The detail that matters: he made every ad himself for 1.5 years before switching.
One person, however good, runs out of angles. A bench of creators hitting the same brief finds the hooks one maker never would.
B2B is about to learn what e-com already knows.
My biggest takeaways from @hectorserrranoo, Head of Content at @WisprFlow:
1. Paid UGC is one of the highest-alpha opportunities for B2B SaaS right now. Paid ads are nothing new, but Hector has discovered there’s still a huge opportunity in doing them differently. E-com brands have long hired creators to produce ads, compensating them based on performance. Now platforms like @TrybeUGC allow B2B SaaS companies to do the same. Hector calls it “the TikTok Shop method, but for Meta Ads.” Wispr Flow was one of the first companies to hop on, and paid is now one of its main growth pillars.
2. Turn on paid early, but only after PMF. Hector credits Wispr Flow’s growth primarily to the quality of the product. “There’s really no hack besides having a good product.” But once you have organic pull, don’t wait to start paid UGC. Hector argues that paid is the fastest way to learn whether your messaging lands and whether cold traffic converts.
3. The key is to incentivize creators by giving them a % of GMV or ad spend. Wispr Flow pays its creators a percentage of the spend behind their ads. E-com brands pay a percentage of GMV. Earnings are also uncapped, so some of Wispr Flow’s creators make over $13,000 a month. “They’re making more money than full-time influencers.”
4. A “smart retainer” keeps new creators around long enough to get good. Pure commission is risky early for new creators, since their initial ads won’t perform well, so Wispr Flow sets a floor. Hit a quota of, say, 25 videos and you’re guaranteed $1,500. If your commissions only reach $500, they top you up by $1,000. If they reach $3,000, you keep the commissions and the retainer doesn’t apply. Three or four months in, the best creators break out and earn purely on commission.
5. Recruit your first creators off-platform, by hand. A new brand may struggle to attract great creators, so Hector manually searches Instagram for niche keywords (“tech,” “nine-to-five,” “lawyer,” “doctor”), then searches the follower lists of creators he likes, “because typically good creators know other good creators.” He sends 150 to 250 targeted DMs over a week, with no mass-messaging software, and usually lands five or six really good creators.
6. Copycat ads die fast, so push every creator to experiment. Creators can see which ads get the most spend, and the instinct is to copy them. Meta’s Andromeda system scans each ad’s transcript, frames, and on-screen text, and groups ads that are more than 60% similar, so lookalikes fatigue quickly. Hector reminds creators that the top earners got there by experimenting.
7. Recruit niche creators to target your ICPs. Each creator should speak to a different audience you want to target. Wispr’s creators range from moms, lawyers, engineers, and sales reps to a 45-year-old founder of an eight-figure e-commerce business. People “mistake UGC for thinking it’s just college kids making content,” but a dev tool’s user is probably an older technical person. For those audiences, “the more boring and the uglier the ads are, the better they will perform.”
8. You need two people to run a paid-UGC program: a creative lead and a media buyer. Hector runs the creator side: comms, concepts, and approving videos. Eric, the media buyer, lives in Meta Ads Manager. For the creator role, look for an online following, not a resume. Hector had a 2.5 GPA. Tobin, Wispr Flow’s early marketing intern, had a 4.0 from Waterloo. Both became great marketers. Hector’s bar is someone hardworking and talented who has grown an account from zero to 10,000 or 15,000 followers, ideally on Instagram.
9. “The best marketers are allergic to AI.” Hector isn’t against automation behind the scenes. His point is about the final output people see. Even the top AI labs use teams of humans to produce their videos and designs. For an AI company like Wispr Flow, AI influencers and AI-generated static ads would land badly and look off. Matt, Wispr Flow’s former head of growth, runs more automations than anyone Hector knows, but “that last 10% that’s the most important always has a human on the endpoint.”
10. Develop young talent like a soccer academy. Hector played youth soccer in Spain and the U.S. before Division I, and he wants startups to copy the model: scout young talent, give them money and room to experiment, let them fail, and coach them until they become A players. His first seven months at Wispr were “all over the place,” and it took about a year to feel comfortable. “A 30- or 40-year-old founder doesn’t understand what an 18-year-old kid who’s in these Discord chats” sees every day, and the people you develop stay loyal.
Watch the full conversation: youtube.com/watch?v=KzS5Qs7v…
The fastest way to spot a weak ad: try to say it in one sentence.
If you need the word "and," you have two ads fighting over the same 15 seconds.
The one-idea test:
1. Write the ad as one sentence. Cut until there's no "and."
2. Every shot has to serve that sentence. Cut the one you love that doesn't.
3. The product shows up as proof of the sentence, not as its own segment.
4. The ending says the sentence again in fewer words.
5. Show it to someone once. If they repeat a different sentence, that's your ad.
Most ads don't die from bad craft. They die from two good ideas sharing one slot.
Save it for your next script review, and send it to whoever keeps adding "just one more message."
ts do the same thing. Fifty tweaks of one concept split the budget and none breaks out, while five genuinely different ideas usually hand you a winner.
Most ads are decided before the viewer knows what they're watching.
The 2-second check I run on every ad, sound off:
1. Something is already happening. Mid action, mid sentence, mid problem. No logo, no intro.
2. There's one thing to look at. If the eye has to choose, the thumb chooses for it.
3. The viewer is in it. Their problem, their excuse, their line. Not your product's.
4. It opens a question the next 3 seconds answer.
5. It works on mute.
Fail 2 of these and the rest of the edit doesn't matter.
Save this for your next edit, and send it to whoever cuts your ads.
We're #18 on this list, and the whole thing is worth a bookmark.
The shift nobody's pricing in yet: the best creator opportunities stopped asking for your follower count. They just want to see the work.
Unlimited Creator opportunities for everyone
1. Whop
-> whop.com
-> Digital products, memberships & communities
2. Superteam Earn
-> superteam.fun/earn
-> Web3 gigs, bounties & projects (Solana)
3. Kaito
-> kaito.ai
-> Creator rewards & InfoFi
4. Gibwork
-> gib.work
-> Onchain work marketplace & gigs
5. Poidh
-> poidh.xyz
-> Social / community bounties
6. Spade Clipping
-> spadeclipping.co
-> Video clipping (agency-style)
7. Promote fun
-> promote.fun
-> PPV / promotion campaigns
8. Wurk
-> wurk.fun
-> Web3 microtasks & jobs
9. Clipster
-> clipster.gg
-> Video clipping
10. Starlist
-> starlist.world
-> Brand partnerships & early creator campaigns
11. Rarelist
-> rarelist.io
-> Creator opportunities & early rewards
12. Nucleus
-> nucleus.codes
-> Web3 opportunities
13. Stallion
-> stallion.so
-> Web3 bounties
14. Clipstake
-> clipstake.com
-> Video clipping
15. Liveframe
-> liveframe.ai
-> Video content
16. Scribble Network
-> creators.scribble.network
-> Creator campaigns
17. Gimi
-> app.gimi.co
-> Creator campaigns
18. Adarena
-> adarena.app
-> Competitions
19. Zero Authority DAO
-> zeroauthoritydao.com
-> Bounties
20. LP Army
-> lparmy.com
-> Bounties
21. Rally
-> app.rally.fun
-> Creator campaigns
22. Shillz
-> shillz.app
-> Social promotion
23. EarnRoyalties
-> earnroyalties.fun
-> Bounties
24. FirstDollar
-> app.firstdollar.money
-> Creator campaigns
25. Wetakeover
-> creator.wetakeover.xyz
-> Creator campaigns & payments
26. Darebay
-> darebay.com
-> Video clipping
27. CreatorWire
-> creatorwire.xyz
-> Creator opportunities
28. WizzHQ
-> app.wizzhq.xyz
-> Bounties
29. 1W UGC
-> 1w-ugc.pro
-> Creator opportunities / UGC
30. Cre8core
-> cre8core.fun
-> Bounties
31. AP Collective
-> hub.apcollective.io
-> Creator opportunities
32. TokenLockr
-> tokenlockr.io
-> Creator campaigns
33. Virality
-> virality.gg
-> Content campaigns
34. ClipersHQ
-> clipershq.com
-> Video clipping
35. Dopamyn
-> app.dopamyn.ai
-> Creator campaigns
36. Ledge
-> ledge.io
-> Creator opportunities
37. Clipify
-> clipify.com
-> Video clipping & short-form content
38. Vues
-> vues.app
-> Video clipping network
39. Clipping io
-> clipping.io
-> Video clipping campaigns
40. Content Rewards (Whop)
-> whop.com
-> PPV clipping & creator rewards
41. FindClout
-> findclout.com
-> Curated clipping & creator marketing
42. Immunefi
-> immunefi.com
-> Web3 bug bounties
43. Code4rena
-> code4rena.com
-> Web3 audit contests & bounties
44. Sherlock
-> sherlock.xyz
-> Web3 security bounties
45. Dework
-> dework.xyz
-> Web3 bounties & DAO tasks
46. Layer3
-> layer3.xyz
-> Web3 quests & bounties
47. Galxe
-> galxe.com
-> Web3 quests, campaigns & rewards
48. Zealy
-> zealy.io
-> Community quests & creator campaigns
49. TaskOn
-> taskon.xyz
-> Web3 tasks & bounties
50. RabbitHole
-> rabbithole.gg
-> Web3 quests & skill-based rewards
save this for no need to find by search
hope will this post help everyone <3
The Snips brief said open on a cliffhanger. The #1 ad out of 84 opened on yours.
"I'll finish it some day."
3 things it did that you can steal for your next hook:
1. Start on the viewer's unfinished business, not the product's plot.
2. Turn what they abandoned into the cast. A detective in the closet "begging for the end."
3. Make the tagline the payoff. "Full story, half the time."
The best cliffhanger is the one your customer left hanging. Save this for your next brief.
Everyone this week is saying fire the agency, hire 50 creators. Half right.
We put one Poker Face brief in front of 59 creators. 84 ads came back. The $1,500 winner ran 21 seconds. #2 and #3 ran about 45.
5 things the short one did that you can copy tomorrow:
1. Open on a feeling, not the app. "Miss Poker Night with the guys?"
2. Name the product by second 6. "Bring it back with Poker Face."
3. Kill the objection in 7 words. "No bots. No strangers. Just your crew."
4. End on the action. "Play Now."
5. Give everyone the same brief. You only learn 21 beats 45 when both sit side by side.
Save it for your next brief. How long was your last winner?
Grandma just out hooked half the ads in your feed. 5 things to steal from this NEWSKOOL reel:
1. Start mid fight. "You stole my card!" is line one, no setup.
2. Chaos in the back, calm in the front. The brawl never stops behind her, so your eyes never leave.
3. Cast against the claim. Nobody sells "anyone can" like a grandma with a six figure website.
4. Pay off every name. "Who's Charles?" Cut to Charles, chilling in a chair.
5. Make the product the punchline. The tagline is her last line, then the logo.
Save it for your next brief.
The Arena is Open.
Base44 wants to see the moment someone realizes they can build an app, website, or super agent using just their words. Create a 20–45s vertical ad capturing that spark. $6,000 total prize value across 50 winning places. Read the brief:
adarena.app/challenges/base4…
