TikTok

Asking for bank details after creators saw potential earnings lifted enrollment 25%. TikTok's creator revenue-sharing program required KYC, including linking a bank account, before creators could earn. TikTok moved the bank account step later so creators could make content and see potential earnings first, and enrollment rose 25%.

The original flow lost many creators because they would not hand over banking details before they had any reason to trust the program. The difficulty of the task was not the problem. The step stayed required for compliance and was only resequenced, so look for steps where users drop off over trust or risk and test placing them after value is shown.

abtest.design

BoldVoice saw huge humorous virality but lower-intent, less loyal subscribers. A TikTok trend of people crying at low Accent Oracle scores gave one creator about 300 million views in a few months with no ad spend. As a paid channel, TikTok produced flimsier, lower-loyalty subscribers than others.

Anada Lakra treats humorous virality as a companion to paid ads, not a replacement. Defaulting users to annual plans let BoldVoice see per-channel cash payback right after the 7-day trial, which is how the TikTok gap showed up.

Sub Club▶ 47:30

Pick micro-creators for content and point of view, not followers. Pok Pok's partnerships manager hunts on TikTok first, then checks Instagram, for creators in niches like Montessori homeschool moms, crunchy moms or day-in-the-life accounts. Many have around 2,000 followers.

TikTok is a core channel for reaching millennial moms. Consumers stopped buying from big influencers about a year earlier, and creators began asking Pok Pok for deals. One person with a good eye finds and cuts creators quickly. Quote: "on Tik Tok it's not about followers, it's about content."

Subversive▶ 33:10

Remix proven competitor formats 10 to 20% instead of inventing new ones. Before launch, Nick Lawton searches TikTok and Instagram for competitors, picks formats to remix about 10 to 20%, and plugs the product in. Flame's team, after studying TikTok 4 hours a day for two months, likewise copied formats working for others rather than inventing.

Lawton warns that copying a format exactly from a warmed-up account onto a day-zero account often fails. Flame keeps its account names private because formats stop working once everyone copies them.

Superwall▶ 6:47

Spark ads buy cheap reach, but conversions are measured on Meta. Spark ads push organic videos into the For You page but have no anchor or button, so only CPM is attributable. Meta ads have a button and conversion attribution, so real scale comes from Meta.

Spark ads suit videos that already carry a strong organic CTA. Quote: "Yeah, CPMs are cheaper and that's really the only thing you can measure from it."

Superwall▶ 34:14

Exploration-based ranking lets TikTok break small and niche creators. Adam Mosseri says TikTok's recommender is best at breaking small talent. Exploitation ranking shows what many people already like; exploration tests content on people who might like something they don't yet know, like showing Afrojack to a Bieber fan.

Inspired by TikTok, Instagram invested in originality, breakout content and recency, and believes it is now ahead in a couple of these areas. Quote: "it's much harder to go and figure out how to essentially test content."

Lenny's Podcast▶ 49:00

Instagram's Reels lost ground by launching on low-read-through Stories. The first Reels, launched in 2019, was built on top of Stories, where read-through is low, so most Reels went unseen and disappeared. When TikTok exploded in 2020, as pandemic users wanted joy and accepted sound on, Instagram was out of position.

Mosseri calls it a big fork in the road for the business. The lesson is not to build a new format on a surface that few people read through.

Lenny's Podcast▶ 1:01:40

Health-adjacent apps may need Meta because TikTok rejects their ads. Cedric Roberge's peptide app gets Spark ads rejected and videos taken down because the content reads as medical advice. Rival GLP apps in the Health and Fitness top 100 run only Meta ads.

His app ranks around 250. TikTok is still his main source of creator content, though creators also steer followers to their own communities and link in bio.

Superwall▶ 50:40

Hire TikTok Shop creators because affiliate pay forces them to sell. StudyFetch prefers TikTok Shop creators, who earn on commission and must make videos that convert, plus creators with past brand or UGC work. It instructs them to sell the outcome, such as better grades.

Quote: "if you don't sell the thing well on your video, you're not going to get sales."

Superwall▶ 24:53

A skilled marketing generalist can be the first TikTok hire. Kieran had never made a TikTok before joining StudyFetch 2.5 years earlier. He argues short-form is a formula of trial, adaptation and decisions from test results, so marketing skill transfers.

A TikTok Shop background adds selling skill and current format knowledge, which avoids recycling year-old trends. Quote: "TikTok and short form is just another channel."

Superwall▶ 45:15

StudyFetch finds TikTok viewers convert better than Instagram's. Kieran says TikTok viewers are higher intent and more US-based, while Reels can push videos abroad where they convert near zero. Some Instagram virals converted close to nothing while TikTok virals converted well.

For US creators he aims for over 80% US viewers; for non-US creators he prefers TikTok over Reels. Quote: "Tik Tok's audience is way higher intent than Instagram's."

Superwall▶ 19:03

TikTok bought distribution by subsidizing both viewers and creators with billions. Evan Spiegel names TikTok and Threads as the only two recent consumer hits besides Snapchat, and says both solved distribution: TikTok with money, Threads by riding Meta's existing products.

His point is that consumer teams spend too much time on whether the product is right and too little on how people will find it, so plan distribution as early as the product. Lenny notes TikTok is more media than social. Quote: "People don't spend nearly enough time thinking about distribution and figuring out distribution."

Lenny's Podcast▶ 3:45

Drop a format that keeps landing under about 5K views. Jay finds a good TikTok format consistently gets at least 4,000 to 5,000 views per video, which signals it will go viral in time; consistently below that, the format is probably dead.

In his niche, a format with around 70% of viewers past 3 seconds yields at least 4,000 to 5,000 views. He also sees the same video blow up on Instagram yet stall on TikTok. These are his own observations.

Superwall▶ 29:15

TikTok's day-30 retention is recalled near 40%, above Instagram and X. Flame's founder describes TikTok as one of the most addictive apps, recalling day-30 retention near 40% against about 35% for Instagram and 30% for X.

These are figures he recalls, not published data.

Superwall

Under 300 views is an account problem; above it, iterate content. Flame diagnoses by tier: under 300 views means a shadow flag, ban or bad setup; 300 to 1,000 means iterate content, formats and hooks; 1,000 to 5,000 means keep the format and tweak the copy until it crosses 100k.

One meme format iterated this way reached 340,000 views. Ryan Thorpe's team similarly treated breaking out of the "0 to 300 view jail" into 5,000 to 10,000 views as the signal a video was worth scaling.

Superwall▶ 29:38

Total watch time lifts views more than the 3-second hook. Across Flame's viral videos, extending total watch time helped most. A meme video edited to hold attention longer reached 340,000 views.

The hook still matters; they extended watch time by curating the memes in the video. Quote: "The part that helps the most is the total watch time."

Superwall▶ 30:40

Remix proven competitor formats 10 to 20% instead of inventing new ones. Before launch, Nick Lawton searches TikTok and Instagram for competitors, picks formats to remix about 10 to 20%, and plugs the product in. Flame's team, after studying TikTok 4 hours a day for two months, likewise copied formats working for others rather than inventing.

Lawton warns that copying a format exactly from a warmed-up account onto a day-zero account often fails. Flame keeps its account names private because formats stop working once everyone copies them.

Superwall▶ 24:00

Flame multiplied one working account into dozens to win on volume. Flame ran around 20 to 30 accounts on roughly 10 phones, 3 to 4 accounts each, reaching 50 million views and 150,000 users.

Quote: "Tik Tok was a volume game for us at that point."

Superwall▶ 24:00

Give every account its own clean device identity. Flame factory resets each iPhone and gives it its own Apple ID and regional proxy. Reusing one Apple ID let TikTok fingerprint and link accounts, costing a week, about 120 hours across three people.

Nick and Ivan found TikTok allows roughly 10 accounts per device before views and engagement drop, so they built a custom Android that swaps the whole ROM to give each account a unique phone, with human-like input patterns. That yields 100 to 200 TikTok accounts per device per day.

Superwall▶ 26:40

Under 300 views is an account problem; above it, iterate content. Flame diagnoses by tier: under 300 views means a shadow flag, ban or bad setup; 300 to 1,000 means iterate content, formats and hooks; 1,000 to 5,000 means keep the format and tweak the copy until it crosses 100k.

One meme format iterated this way reached 340,000 views. Ryan Thorpe's team similarly treated breaking out of the "0 to 300 view jail" into 5,000 to 10,000 views as the signal a video was worth scaling.

David Senra▶ 47:06

Post organically first, then boost winners and spark ad the best. Ryan Thorpe's portfolio posted every creator video on brand accounts, boosted any with traction (even around 10K views or many comments), and ran real breakouts as Spark Ads from the creator's own account.

This gave four or five ways to push one video and saved paid testing costs in quiet months. He calls 2020 to 2025 open season on TikTok for the portfolio.

David Senra▶ 47:06

Give every account its own clean device identity. Flame factory resets each iPhone and gives it its own Apple ID and regional proxy. Reusing one Apple ID let TikTok fingerprint and link accounts, costing a week, about 120 hours across three people.

Nick and Ivan found TikTok allows roughly 10 accounts per device before views and engagement drop, so they built a custom Android that swaps the whole ROM to give each account a unique phone, with human-like input patterns. That yields 100 to 200 TikTok accounts per device per day.

Superwall▶ 39:27

Automating distribution, not content, produced about 100 million views. Nick and Ivan's first automation target was TikTok, which produced about 100 million views, mostly promoting clients' music; one static image reached 2 million views.

Making content for clients was too hard, so clients supply content while they automate distribution, warm-up and comments, labor that is costly with VAs. They moved to Instagram after the TikTok ban.

Superwall▶ 21:21