Nicole Cheung - she makes $400k/mo from ai apps (while in college)

Faceless TikTok slideshow: girl without makeup -> girl with makeup with GlamUp branding -> app screenshot of the personalized routine. Hundreds to thousands of posts used the identical on-screen hook ('never saw myself with the / right makeup routine' + two crying-face emojis on a third line), identical caption ('I will be unstoppable' + exhausted emoji), same hashtags and one of ~10 proven audios. On-screen text kept small and off-center, 'Gen Z Insta-girly' style. Hook wording, line count and text size were tested; other audios did not go viral. Earlier formats that failed: Latina makeup slideshows, color theory, celebrities, makeup tutorials. Posted across ~20 UGC creators' accounts.

Found photos of attractive women on Pinterest, ran them ~10 times through a separate makeup-removal app until plain, then enhanced with GlamUp's makeup to create maximum before/after contrast. Used as app demos rather than real users' photos (privacy).

Each co-founder ran ~3 TikTok accounts, warmed them up and tested formats. Only after a format went viral did they hire creators (30-40 at GlamUp's peak), trained through a course covering account creation, warm-up and exact content details (text size, placement, etc.).

After the (blurred) scan result, the paywall offers two buttons: 'View results' (-> paywall) or 'Invite 3 friends'. Inviting leads to a share-your-invite-code screen; results unlock free once 3 friends download, finish onboarding and redeem the code. Friends get no benefit beyond helping. Target users are 13-25 and don't want to pay, so they post codes in TikTok comment sections. Branch was used for attribution/referral links. Repeated on Sprout and the newest app.

The copy button doesn't copy the bare code; it copies a template like 'Download GlamUp and use my referral code ...'. Users paste it into TikTok comments, so the keyword 'GlamUp' ranks high in TikTok search and searches surface the app's videos.

Raised the weekly subscription from $3/week to $9/week (weekly and monthly plans only, no annual).

New app sells only monthly and annual plans. Raised both by 20%; now testing a further 20% increase.

Hard paywall with weekly tiers $8.99/$10.99/$12.99. On close, immediately show a discounted paywall at 50% off ($4.49/$5.49/$6.49). Not used on Sprout.

Tiers include e.g. 2 or 4 scans per week. Many scan types (color analysis, glow-up guide with six makeup guides, etc.) make users want more scans; running out triggers an upgrade paywall to a higher tier.

User picks a scan, gets photo instructions, takes a selfie or video, sees an 'Analyzing' screen, then all results are blurred behind the paywall. The scan is fake; the AI API is called only after payment to avoid wasted API cost.

After onboarding questions, a 'Leave a rating' screen triggers the native rating pop-up; to continue, users must tap an 'I rated' button (whether or not they rated). Same screen used on Sprout. Newest app dropped the forced button (co-founder uncomfortable) and uses a plain 'Leave a rating' + continue, which still works. Similar approach for the notification permission page.

Onboarding is for priming, not analytics: ask about goals/concerns (beauty goals, skin concerns; job goals) so users feel they need the app, then tailor a plan. On Sprout, added 'fun fact' pages after questions: e.g. 'Have you tried other job search apps?' is followed by a Sprout-vs-other-apps comparison; 'What is your goal?' is followed by 'you can reach that goal with Sprout way more than without'. Then rating, notifications, referral code.

Sprout sells weekly and monthly plans only (up to ~$80/month), no annual, because a job search shouldn't take a year. No close-out discount paywall. Payments go through Stripe in an in-app sheet instead of App Store IAP.

Previously users finished onboarding and entered the app (home screen, pick a scan / swipe on jobs) and were paywalled at the first valuable action. Tested a complete hard paywall straight after onboarding, with no access to the app. Sprout also switched to a hard paywall and saw better conversion.

Applicants come in via a Typeform linked to a CRM. Course modules: getting started, warm-up playbook, creating content, content bank examples, content policies, account management, next steps, resources, bonuses. Each module is an engagingly edited video of the founder talking with examples (how to edit, how to introduce the app so it converts). A quiz ends every module to quality-check completion. Creators start brand-new accounts. 1,000+ creators onboarded; 200+ active at once.

Inbound: post the application form in Discords of influencers who coach people to go viral (students already semi-trained and interested in content), and on Handshake, where students apply thinking of a social-media internship. Also outbound. Every candidate gets a ~15-minute interview to vibe-check, judge fit with the content and seriousness, and is told the course plus quizzes are mandatory. Prior viral experience is a bonus, not required.

Flat $10-30 per post. Bonuses at view thresholds; Sprout uses a decreasing CPM: ~10K views $40, 50K $50, 100K $80, 1M $800 (vs flat CPM of 10K $10 / 100K $100 / 1M $1,000).

For every viral format: idea, details, example videos, audios, and a line-by-line script where each line has the emotion to portray (sarcastic tone, eye roll) and the shot (long, medium, close-up). Founder films example videos herself for creators to copy. Formats include fake recruiter-call skits and talking heads.

Faceless 6.7-second video: text 'War is over. Finally landed my first full-time job' over a Google Sheet job tracker; a dropdown is switched to 'Accepted'. Length tested at 5, 6 and 7 seconds; 6.7s won. Filmed at a tilted (not 90-degree) angle; show the sheet, zoom in, click the dropdown slowly so options are visible but not clear, click Accepted, zoom again, zoom out. Creators get ready-made Google Sheet templates and a tutorial video.

Took Umax (face-rating / looksmaxxing app for men, seen early via a co-founder's HF0 cohort) and reframed it for women around makeup and skincare products rather than face ratings; checked the App Store and found few such apps. Built a barely working MVP in 3-4 weeks and worked on distribution in parallel; v2 redesign took another month.