Getting users
Video hook tactics
Hooks founders tested in TikTok and Reels videos and ads, and which ones held viewers and drove downloads. 62 tactics from 42 founder interviews.
Most founders found one hook that worked, often borrowed from another niche or a competitor and tested on their own account first, then posted it hundreds of times with small changes. A few words in the opening often decided both who watched and whether they downloaded. Big view counts did not always turn into installs: clickbait and debate-bait hooks drew views that converted poorly when the video did not lead clearly to the app.
- Halo AI gave every creator one format, a shocking AI prank image then an app demo and a text-message story, and reported 1.2B views in 120 days, with about 1% of videos passing 1M.Halo AI
- Small wording changes moved results: Massive's stop-using-LinkedIn hook converted far better than an Indeed version with 11M views, and a scratch-map app's exam-procrastination caption brought downloads where a breakup caption got views but almost none.MassiveA scratch map app
- Pushcroll spent about 80% of its effort on the hook and switched formats when about 30% of viewers swiped away in three seconds; its breakout video had a skip rate of about 19%.Pushcroll (1)Pushcroll (2)
- Founders often tested hooks on their own accounts before handing them to creators: Nomadtable's founder said a hook that worked for him worked for nearly every creator, and GlamUp's first fixed-hook slideshow got about 10M views.NomadtableGlamUp
- Views without a clear path to the app underdelivered: Nomadtable's clickbait hostel hook got 300K views and 174 likes, and a Jenni AI debate video reached 30M views on X but plugged the app too weakly to convert.NomadtableJenni AI
- Cloning proven posts with one change was common; one agency said clones get a median of about 20% of the original's views, with some reaching 2-6x, while Nomadtable found adding a physical action to the opening made no clear difference.Practitioner pattern · Matt Espinoza (Founder of Clover, which sells the cloning, multi-country account and AI search/Reddit seeding services he recommends)Nomadtable
Summary written 2026-09-26 from the tactics below. Figures are as founders stated them.
What worked, with the best evidence
What didn't work, or only partly
- Nomadtable: Film one-take hostel videos with clickbait hooks like 'never staying in a hostel again'Up to multi-million views, but poor engagement (300K views, 174 likes); dropped
- Reflective Apps: Put a famous person's voice behind the video as the hookOne video hit 50-60M views, then a cease and desist; AI-voice copies partly worked
- Nomadtable: Always be doing something in the opening, like drinking or tying a shoeNo clear impact; dropped
Companies in this topic
More tactics (47)
- Symmetry: Post ~80K creator videos a year from 500+ accounts; track which formats win~1B views in a year; main driver of ~3M downloads
- Change the caption from 'post-breakup activity' to 'procrastinating for exams'Same video: millions of views and no downloads became App Store jumps in several countries
- Pingo AI: Film yourselves talking to the app on TikTok, one account per languageReached $20K MRR in about 3 months with almost no outside creators
- Pingo AI: Tie videos to beginner lessons and culture moments, like K-dramas or the TikTok banA 'basic Korean quiz' got 77K views; a TikTok-ban/RedNote video got 1M
- Coconote: Post short videos that show the app solving a real student problemSeveral ~20M-view videos, ~99% organic, converting far better than viral novelty content
- DaGame: Post 'Day 4 of making a DaBaby game' videos and add what commenters suggestTop video 4M views; first videos drove 300-400K downloads
- Put eggs in the first image of healthy-eating slideshowsHigher-view slideshows consistently had eggs in the hook image; no numbers
- Conch: Stitch an already-viral classroom clip with better audio, then demo the app15-16M TikTok views in a day, 100M+ across platforms; drove a $500K month
- Pushcroll: Put 80% of effort into the hook; aim for under ~18% skippingTheir breakout video had a ~19% skip rate; 30% means switch format
- NGL: Post TikToks of someone reacting to a shocking anonymous message they got~1M views fast; first 1,000 users in a day, then global in a week
- Tolan: Make videos of people using the app while studying, cooking or getting readyStudying ('up at 3am') took off first; a cooking video got ~10M views
- Noise: Test slideshow hooks on your own TikTok account, adding text inside TikTokFirst post 887K views; one hook image passed 5M views
- Fitted: Keep testing new video formats and post each one on every accountAbout one viral video a month for 18 months; top videos hit 136M and 60M views
- Post calm, sideways talking videos while everyone else posts loud vertical ones35K views turned into 200-250 new followers
- Cluely: Hand one winning video format to under 20 creators, then move on to new formats0 to 100M views in the first few weeks at under $1 CPM
- Reflective Apps: Give creators strict briefs: 10-15 second videos, 25-27 cuts, POV phone shots, trending soundsSet a quality floor across ~250 creators; results still varied
- Doublespeed: Open with a pain point, list harder fixes, then show your app as the easy oneRefined over 3-4 rounds; one 50-image bank can feed ~300 slideshows
- Career Hound: Keep reposting one video: a list of companies hiring remote, ending with your siteRun for two years; part of 11.7M views in 30 days
- Welltory: Hook every ad on something scary happening inside your bodyBeat long-term 'feel less tired' hooks on Meta, per guest; no numbers
- Cal AI: Have creators show the photo-scan naturally in the first 15 seconds, with no scriptCalled one of the biggest rules; no numbers
- Stella: Start videos with 'Repeat these for 7 days' so people save themFounder says saves lift engagement and make it go viral 'every time'; no numbers
- Flame: Post up to 150 videos a day to test one hook, then tweak the winners dailyFounder credits volume for the growth; no per-hook numbers given
- Boost App Social: Give creators a one-page list of hooks that work, then let them improviseFounder says it balanced freedom and quality; low-effort hooks were rejected
- Boost App Social: Caption every ad and film the creator over a green-screen demo of the appDistilled from ~$1.5M of spend; the format behind $100K MRR
- Boost App Social: Open ads with one exact frustration, like 'stuck on 1,000 views'Among the final best ads; smaller reach but works really well
- Photoroom: Show the AI background remover in the first 6 seconds of every adLower CAC, higher conversion; opened Mexico, Brazil and Indonesia
- Nomadtable: Test every new hook on your own account; give creators only the ones that hitA good hook works for nearly every creator; founder is the top-viewed poster for 2-3 months
- Flamme: Scroll TikTok 4 hours a day for 2 months to learn what worksGot the first account getting views, then copied the approach across accounts
- ElevenLabs: Test ad formats with agencies, then hire motion designers to make variants in-houseMotion-design ads are the bulk of ads and work really well; no numbers
- SteelAlive: Comment a short relatable story that ends 'now I use an app for this'Viewers ask which app or visit the profile; no numbers
- Toxic Traits: Show a text that's only arguably a red flag so viewers argue in commentsFounder says the formats worked; 330 videos ready, ads never tested
- Clear 30: Film your own funny ads that call out a very specific momentSpecific callouts grab attention; calm ads with an older speaker convert best
- YOGABODY: Run ads that name a very specific pain, like 'run 10+ miles with tight hips?'Converted Dan after 6 ads despite a painful checkout; ads live ~2 years
- Sizzle AI: Open with a silent shocked face, demo the app, end on a goofy twist79K likes; this ending beat the creator's other endings
- Run your viral organic videos as ads, and have top creators film adsOrganic margins 50-80% vs 10-30% for ads; host says reuse beats typical ad UGC
- Built With Science: Hire clippers to post long-video clips on Reels and TikTok, then reuse winning hooksSome clips go viral; their hooks now shape new long videos
- BetterMe: Run ugly ads that show the product and the pain in the first secondOne ran for years; called 'the best ad ever'; no numbers
- Coconote: Mix a known game (2048) with studying and promote a catchy URL in videosOne video reached ~2M views; no conversion numbers
- Praycreen: Post shock-value TikToks across 10-12 accounts to get people arguing in commentsEarly comments were angry, then turned to 'I needed this'; main driver of growth
- Clear 30: Study which ads Meta and TikTok spend most on, and reuse their hooksAd with a woman beat one with a man; camera-shot look beat polished
- Stronger: Adapt viral video formats from other app niches, then test hooks by watch timeHow they find each new format; no numbers given
- Sorce: Film a friend 'discovering' the app, with two faces and a moving cameraGoes viral consistently with one or two formats; widely copied
- Suno: Lean into people hating your category to drive commentsCalled the easiest way to go viral; no numbers
- When a creator video passes a view mark, swap its hook or demo and repost elsewhereNo results given; pitched as 5x testing volume per creator
- Stronger: Copy a winning video across many accounts, changing a few words or colors each timeAdopted after a rival did it with ~10 accounts; no separate numbers
- Give creators a short brief with a Loom, target persona and proven videosSeen across SideShift customers; long briefs get skipped
- Coconote: Aim videos at the parent who pays, not just the studentMany viral Instagram videos used this angle; later scaled back
New tactics every Monday
The week's founder interviews, boiled down to what they did and what happened.