Source: https://youtu.be/Aq0TfaAdSmY
Compiled: 2026-03-24
These are the subtle, easy-to-miss details buried in the conversation — not the headline takeaways, but the little things that can save you months of trial and error.
1. Test hooks BEFORE you test images.
Don't waste money on fancy AI-generated images until you know which hooks work. Use ugly red squares with yellow text. If the hook doesn't work as plain text, a prettier image won't save it.
2. Split test primary text in two separate phases.
- Phase 1: Test 10 variations of the above the fold (the first ~3 sentences people see before clicking "see more")
- Phase 2: AFTER you have a winning hook, test 10 variations of the below the fold text
- Most people test the whole primary text at once — that's messy. Isolating above vs. below the fold gives you cleaner data.
3. Don't bother split-testing headlines.
Jordan explicitly said he doesn't test headlines because "it doesn't matter very much." Focus your testing budget on hooks, images, and landing pages instead.
4. You've become a "copy chief," not a copywriter.
Even Jordan — a copywriter by background — now mostly edits AI copy rather than writing from scratch. The skill shift is: come up with the big idea, let AI execute, then tweak. The human value is in the idea and the editing, not the first draft.
5. Jordan can still beat his own AI on copy, but NOT on images.
This tells you where to focus your personal effort vs. where to let AI take over completely.
6. Use MULTIPLE AI models for the same prompt — diversity is the strategy.
Different AI models "think like different people." Claude gives different prompts than Gemini. Running the same brief through multiple models gives you more diverse creative angles, which increases your chances of finding a signal.
7. Gemini got aggressive with images after its latest update.
Jordan specifically called out that Gemini used to be bad at generating ad image prompts but now generates "very aggressive" prompts — which is exactly what you want for direct response. Possibly not compliant with Meta/Google policy, but it performs.
8. Close-up images of hands, feet, skin, or body parts spike your CPMs.
Meta has a weird policy where close-up body part images raise CPMs significantly. If your AI is generating these, filter them out. This one small thing can tank your ROAS silently.
9. Build compliance rules INTO your image generation prompts.
Jordan doesn't check compliance after images are made. Instead, the prompt itself includes instructions like "no violence, no nudity, no pills, no disease names, no guns, no drugs." Bake it in at generation time so you don't waste creative output.
10. Have AI models critique each other's work.
They have a feature where Claude analyzes Gemini's creatives and vice versa. Useful especially when starting a new campaign — cross-model feedback catches blind spots and refines prompts.
11. Never feed a full VSSL script to AI — compress to 5-7 "fragments" first.
This is maybe the most actionable AI tip in the entire interview. Full scripts cause hallucination. Jordan's fragment list:
1. Explanation of the product mechanism
2. Target demographic
3. Their nightmare scenario (deepest fear)
4. Their ideal state (what life looks like if the problem is solved)
5. ~4 more key data points about the offer
Every single AI prompt in the entire workflow runs off these fragments only. This is what makes his output quality high despite massive volume.
12. The nightmare scenario drives the best ad angles.
Jordan's biggest creative wins came from deeply understanding the target's worst fear — not the product features. The "grandma in a senior home" angle for memory loss came directly from identifying that fear through the fragments system.
13. Your best ideas often come from OUTSIDE the VSSL.
The ad doesn't have to match the VSSL mechanism at all. Jordan's vision offer angle came from a Bible story about Jesus healing a blind man. Completely unrelated to the VSSL's actual mechanism — but so captivating people couldn't stop clicking. The lesson: your ad just needs to get the click. The VSSL does the selling.
14. The listicle structure — how it actually works:
- "Top 5 at-home remedies for [problem]"
- Items 2-5: generic real tips (put feet in salt water, drink more tea, etc.)
- Item #1 (the "best"): is the offer's VSSL
- This positions the product as the top recommendation alongside real advice
- This is NOT a comparison page — it's a value-first page that funnels to the offer
15. Quiz funnel has 3 specific psychological stages:
1. Problem state: Ask questions about their problem (activates pain)
2. Solution state: Ask "how would your life change if you solved this?" (activates desire)
3. Result: "Based on your answers, we've determined the best solution for you — watch this video"
- The key insight: you're not qualifying them. You're creating an emotional arc from pain → desire → hope → action.
16. The "granny blog" — what makes it work:
- Looks like it was written by an actual older person (not a marketer)
- Personal story, very emotional, raw, unpolished-looking
- People trust it MORE because it looks authentic
- Lower click-through rate BUT higher conversion when they do click
- "My grandma was in a wheelchair, now she can walk again"
17. Scientific advertorials get very qualified clicks but low CTR.
Format: "Researchers at [prestigious institution] figured out the real root cause of [problem]"
- Works amazingly on some offers, completely fails on others
- The traffic that does click through is highly qualified
- Best for offers with strong mechanism stories
18. Stop trying to match specific ads to specific landing pages.
Jordan used to try to find which advertorial works best with which specific ad creative. He stopped because:
- It added enormous complexity
- The juice wasn't worth the squeeze
- Usually ONE landing page wins across almost ALL ads
- Now he looks at landing page data in aggregate — not per-ad
- This simplification was a "huge unlock"
19. Don't put landing pages in separate campaigns.
Use your tracker (RedTrack, etc.) to randomly split traffic across landing pages from the SAME campaign. This gives you cleaner data and lets the algorithm optimize for purchases without landing page variables messing up your campaign structure.
20. The 10/50 landing page testing method:
- Winner gets 90% of traffic
- Test new variation with 10%
- If the 10% holds up → move to 50/50 split
- 50/50 gives you enough data in a single day to confirm the winner
- Then loop: new winner gets 90%, test another variant at 10%
21. A landing page with terrible CTR can still be your winner.
Jordan's nerve offer example: a "granny blog" landing page had terrible click-through rate (nobody was clicking through to the VSSL), BUT the people who DID click through converted at 10%. That made it the overall winner. He then just tweaked the page to improve CTR — conversion rate dipped slightly but overall performance improved. Always look at end-to-end purchases, not intermediate metrics.
22. Initiate checkouts — the metric nobody talks about.
- On ClickBank, ~25% of people who initiate checkout actually purchase
- This means cost per initiate checkout gives you 4x more data than waiting for purchases
- There's a very high correlation between cost per initiate checkout and whether an ad becomes a long-term winner
- He got this tip from Jason Katasi (major affiliate, now in agency space)
- He still optimizes campaigns for purchases — but uses initiate checkout data for his own faster decision-making
23. CPC and CTR have ZERO correlation with winning ads.
Jordan ran an AI analysis on his entire dataset looking at ads with 100+ sales. Found:
- No CPC correlation with winners
- No CTR correlation with winners
- Only CPMs showed slight correlation (high CPMs = worse performance)
- High CPMs likely mean Meta doesn't like your image/text
- Stop looking at CPC and CTR to evaluate ads. Only look at cost per purchase (or cost per initiate checkout for faster reads).
24. KPI targets differ dramatically by offer type.
- ClickBank affiliate offers: target 1.5 ROAS
- Recurring app offers: target 6.0 ROAS (because they're optimizing for break-even knowing LTV will pay off)
- Set different KPIs per campaign — don't use one universal target
25. The bid cap transition method:
1. Launch new ads with max conversion (let Facebook optimize freely)
2. Once you have data, split test different bid amounts
3. Everything that's scaling transitions to bid cap
4. Bid caps self-regulate spend — you don't have to manually babysit budgets
5. Most affiliates aren't doing this
26. Budget scheduling — the actual mechanics:
- EVERY campaign starts at a $2,000/day base budget
- They may have 100 campaigns running at this base
- Check performance in the morning and again in the evening
- For campaigns performing well: use Facebook's budget scheduling feature to ADD more budget that day
- This is reactive scaling, not set-it-and-forget-it
- "One of the best ways to scale"
27. The Andromeda CBO goldmine:
- Load a CBO campaign with many ad creatives
- Each individual ad takes very little spend
- Some ads get cheap sales every few days at nearly infinite ROAS
- Don't try to scale those ads individually — they won't scale
- But running MANY of these low-spend, high-ROAS ads together prints money
- Facebook's algorithm has gotten incredible at matching specific ads to specific micro-segments of your audience
- This is the "lots of ads in a CBO" strategy everyone should be running
28. Volume IS a cloaking strategy.
When you have 1,000+ ads live, competitors using spy tools can't figure out which ones to steal. They'd have to rip all of them — and they're too lazy for that.
29. Redirect chain for Facebook Ad Library snoopers:
When someone clicks your ad from the Facebook Ad Library (not from their feed), it carries different URL parameters. Jordan detects that and redirects those clicks to a completely different offer — a real, legitimate ClickBank offer at the top of the charts, but NOT the one he's actually running. So spies think he's promoting Offer A when he's actually promoting Offer B.
30. Design your ad library destination to look legitimate.
Don't redirect to a blank page or something suspicious. Redirect to an actual hardcore direct-response page running a real offer. It has to look like a real campaign — just not YOUR real campaign.
31. How Jordan approaches a brand new offer (step by step):
1. Start with the vendor's VSSL (don't make your own yet)
2. Test with simple splash/bridge pages first
3. Run the 3 initial test campaigns (inspired variations, red square hooks, AI images)
4. Find 2-3 winning ads
5. Test primary texts (above fold first, then below fold)
6. Test 5 landing page types (bridge, listicle, quiz, scientific, granny blog)
7. Find a winning landing page → send 90% traffic there
8. Once baseline is established: start making your own VSSL opens
9. Try to beat different sections of the vendor VSSL
10. Add complexity layers: pop-unders, push notification collection
11. Trademark your custom VSSL
32. Add pop-unders AFTER you're at scale — not before.
Pop-unders (background tabs with a different offer) are an advanced monetization layer. They're basically free traffic — someone who didn't buy your main offer closes the tab and lands on a complementary offer. Jordan's doing "a couple thousand a day" from pop-unders alone. But this is added complexity — nail the basics first.
33. Choose complementary pop-under offers strategically.
Jordan was running a supplement offer at $200K/day spend. His pop-under was a digital product in the same niche. The logic: if someone can't afford the supplement, maybe they'll buy a cheaper digital product. The vendor of the digital product was "blown away" at the free traffic coming through.
34. Collecting emails barely hurt conversion but delivery was the problem.
Jordan put an email form right before the checkout button. People had to enter email to proceed to checkout. Impact on conversion rate: "very little difference." But he couldn't get the emails to deliver reliably. Email marketing is its own complex beast.
35. Push notifications are the easier alternative.
Instead of emails, he collects push notification subscriptions via browser popups ("This page would like to send you push notifications"). Then sends push notifications for other offers with urgency hooks: "You have to see this — XYZ just happened. Click here." No deliverability issues, no email infrastructure needed.
36. Newsbreak — the specifics for newcomers:
- Was $0.05 CPCs when Jordan first got on
- Now flooded with ClickBank affiliates
- Still potentially viable for non-ClickBank offers
- Works for: weight loss, make money online, trading, mass market stuff
- Does NOT work for: high-ticket offers targeting sophisticated audiences
- Newsbreak did an official partnership with ClickBank → brought the flood
37. Rumble — cheap but niche-specific.
Jordan had success with a survival offer on Rumble. The audience skews conservative/prepper. Still "super cheap" traffic. Worth testing if your offer matches that audience.
38. Images for mass market, videos for sophisticated audiences.
- Health supplements, weight loss, nerve pain → images work better on Meta
- High-ticket coaching, real estate, trading for sophisticated buyers → videos only
- The dividing line: curiosity-based offers (images) vs. explanation-needed offers (videos)
- Jordan did his first couple million on ClickBank with AI avatars (video) but switched to images-only after Meta's algorithm update
39. If you're running high-ticket, always have a mid-ticket front end.
Jordan's real estate offer: $300 front end, $5-10K back end. They need to break even on the front end because the back-end cost of goods is high. Going straight to "book a call" had quality issues. A mid-ticket buffer qualifies the buyer AND gives you front-end revenue.
40. Facebook API ad launching — the setup:
- Create an app inside Meta's developer tools
- Get it approved (usually pretty easy)
- Connect it to your profile
- Launch ads through the API directly — zero manual labor
- Jordan's partner built this; took about a month to figure out
- Result: 100 ads up in 5 minutes with zero human involvement
- Note: profiles can get locked, which is annoying
41. Don't build your own tracker.
Jordan tried, it blew up, and he lost a bunch of money. He uses RedTrack (an off-the-shelf tracker). Building a custom tracker is one of the hardest technical challenges in the space. Use what works.
42. Spy before bed — let your subconscious work overnight.
This isn't just a habit tip. Jordan says when he does spy research before bed, his subconscious processes it overnight and he wakes up with ideas. He specifically schedules spy time for evenings.
43. Residential proxy + real feed scrolling is #1.
Better than any spy tool. Set a residential proxy for a US IP, then just scroll Facebook as a regular consumer. "That's where I find the best stuff." You see what the algorithm is actually serving to real people — not what a spy tool's database thinks is running.
44. YouTube review channels = demand indicator.
If YouTube channels are posting lots of reviews for a product, that product's offer is in demand and scaling. High view counts on review videos = high search volume = the offer is big.
45. Scam-busting YouTube channels = offer validation + inspiration.
Channels like CoffeZilla that "expose" products:
- The product name is in the video title → view count tells you search demand
- The content itself shows you what the offer does → "Can I do that, but less shady?"
- These channels are essentially free offer research
46. $500 ad account spy services — worth it.
You can pay ~$500 to look inside someone's actual ad account. Jordan found unique creative formats from a top affiliate this way that he then transferred to his own campaigns. Before he was friends with some of his biggest competitors, he was paying to spy on their ad accounts.
47. Be transparent with vendors about custom VSSLs.
Jordan always tells the vendor what he's doing: "I think I can beat your VSSL. Do you mind if I run my own landing page direct to your checkout?" He's not sneaking around. Transparency builds trust and gets you cooperation.
48. Send traffic directly to the vendor's checkout page.
When running a custom VSSL, you skip the vendor's entire sales page. Your VSSL → vendor's checkout page directly. This means you control the entire customer journey except the checkout.
49. The vendor relationship is "the most important thing."
Jordan stressed this repeatedly. Good vendor relationships = you can test more offers, get better terms, get cooperation on custom VSSLs, and learn from each other. It's the #1 relationship in the affiliate space.
50. There's no "making $10K/month" as an affiliate.
Jordan's exact words: "I don't know any affiliates making 10K a month. If you find a campaign that works, you're going to make a lot more than 10K. If you don't find a campaign that works, you're losing money. There's no in between." It's binary: you're either crushing it or bleeding.
51. Pay media buyers on performance with no/low base fee.
Jordan's Google Ads guy got 30% of profit with zero base salary. When the channel worked, it was a great deal for both sides. When it stopped working, no ongoing cost to the business. Risk is shared.
52. When a media buyer's channel dies, invest in their pivot.
Jordan's Google guy couldn't make YouTube work anymore. Instead of firing him, Jordan paid for native ads coaching so the guy could pivot to a new traffic source. Investing in your team's growth = loyalty + future value.
53. Recurring apps are where the smart money is going.
Lithuanian affiliates are doing 9+ figures/year on fitness apps, hypnosis apps, quit smoking, quit vaping, sleep better apps. All recurring subscriptions at 100% margin. The model: use your affiliate marketing skills to drive installs, then the app's recurring revenue builds long-term value. This is what makes a company exitable.
54. Take an ad angle that already works and build a product around it.
Jordan's best-performing ad angle was "try the Sudoku trick to improve your memory." So he built an actual brain games app around that exact angle. The ad already proved the market demand — now the product matches the proven hook perfectly.
55. The multi-face publishing model.
Big publishers run 10+ different "faces" (spokespersons/avatars) in the same niche, each with slightly different mechanisms. When one face's audience fatigues, the next one is already running. This is how you build a durable publishing business vs. riding one wave.
56. Start simple, add complexity only after baseline is proven.
Every part of Jordan's system started simple: vendor VSSL → simple splash pages → basic images. Complexity (custom VSSLs, pop-unders, push notifications, quiz funnels) only gets added after the basics are working.
57. Simplify ruthlessly when complexity doesn't pay.
The biggest example: Jordan stopped trying to match specific ads to specific landing pages. The complexity multiplied variables without meaningful lift. One winning landing page across all ads = simpler AND more profitable.
58. The role of the human is shifting from executor to "copy chief."
AI writes the draft. You come up with the big idea, guide the direction, and edit the output. This applies to copy, images, landing pages, and even VSSLs. Knowing WHAT to create matters more than knowing HOW to create it.
59. Leverage skills across business lines.
Jordan chose affiliates specifically because it builds skills (media buying, copywriting, audience understanding) that transfer to his exit-vehicle business (recurring apps). Every business decision feeds a larger strategic plan.
60. The best learning happens in private conversations, not courses or presentations.
Jordan's most valuable insights came from casual chats at side events — guys with no Instagram pages doing $300K/day sharing tips over dinner. Not from paid courses, not from conference stage presentations.
Compiled by Bloop 🫧 — 2026-03-24