How She Built Two AI Apps to $300K MRR

First version: scan your face, get text-based makeup tutorials. It did not work on social media. Changed the core feature so users scan their face and see a before/after of how they would look with specific makeup looks, a visual result that suits TikTok content. The team usually tests a few candidate features; once a feature is viral it works ~10x better and you know within the first week.

For 3-4 months marketed the text-based first version via influencers, faceless accounts, the founder filming her own UGC, Reddit marketing and Discord. Nothing worked until the feature changed. Her takeaway: if several variations and methods fail, the feature is not viral and you learn that quickly.

With a creator network already in place (~800 active creators on the stealth app), new feature ideas are built as a Figma mockup or in the team's own content creation portal first. Creators post content about it; only features whose content goes viral get built into the app.

Base rate $10-30 per post; her own rule is a minimum of 500 views before the base is paid. Bonuses start at 10K or 50K views on a decreasing-CPM ladder, e.g. $50 at 50K views (~$1 CPM) up to ~$500 at 1M views (~$0.50 CPM). Base depends on content type and creator experience: creators working with five apps won't take $10, but new creators (college kids) with easy-to-film content will.

Early on: post a campaign on SideShift or similar UGC platforms. Now: inbound applications from creator Discord chats/communities run by former creators, personal referrals from friends, and own outreach to micro-influencers averaging ~10K views per video who go viral but have no way to monetize.

Every creator opens a brand-new account (easier to steer the algorithm to the target audience). Warm-up takes ~3 days: follow and repost targeted micro-accounts in the niche, not macro accounts (following Messi confuses the algorithm); scroll and comment in the niche every day; don't switch accounts often; wait until after international travel. Full warm-up playbook is posted as an article on her X account.

When starting out, her whole day was blocked with interviews of UGC creators. She accepts roughly 50% or fewer, favouring people who take it seriously, so the program's foundation is good.

For each app she films a tailored course (her face / video tutorials, 30 min to 1h+), written assuming the creator knows nothing: 1) app intro and setup, including any codes needed to use features; 2) account creation and warm-up with example usernames, bios and profile pictures; 3) content creation with a content bank per format (talking head, reaction, B-roll, Snapchat-style): idea, example videos (filmed by her, swapped for a creator's video once one goes viral), scripts, emotions, on-screen and out-of-video captions, hashtags. Scripts are written like screenplays with an emotion per line and long/medium/short shots to hold watch time; 4) account management: keep warming up, work the comment section, post stories. Creator questions are answered within a couple of hours.

With no money, Nicole or her co-founder made all the content until something went viral; only then did they scale by hiring creators. Finding viral formats is the founder's job (hours and days of research), not the creator's. She still films the example videos in each app's content bank.

Warm up an account in the niche so the For You page shows current niche content. Ignore influencers (their views come from fans). Look for accounts with under 100 followers whose videos average 200-1,000 views but have one video at ~1M views; that format is what's viral. Copy it. A viral format is one that 5 of 10 creators can make go viral.

For one content format she tests a short script (~10s), a medium one (~25s) and a longer one (~30s), plus variations in hook, on-screen text, filming style and emotional delivery, since these can decide whether a video goes viral.

For consumer apps the priority of onboarding is to prime users to convert, and only second to learn about them. Use 10+ pages. Skip questions that don't drive conversion (email, phone). GlamUp asked: what's your beauty goal, skincare concern, what do you always struggle with, what products do you use. She studied ~100 apps' onboardings, learning only from high-MRR apps.

Place social-proof screens between onboarding questions, tailored to the previous answer: if the user picks acne as their skin concern, next screen says '80% of our users have acne as their top concern'. In the stealth app, after asking 'Where are you based?', it shows how many users the app has in that location.

Sprout launched with a hard paywall on the core action: users must pay to swipe right (apply). They tested whether users could see jobs before paying; currently users can browse jobs but not apply. Her newest app is fully hard-paywalled. She prefers hard paywalls for single-player apps to maximize revenue per download; freemium for social/two-player apps.

Across her apps she has raised prices by ~20-40% and seen the conversion rate go up. Treats price as something to test until you find the point that works, with tiers A/B tested.

Tested pricing and paywall ~2 months ago on the stealth app and is testing again now, because changes in distribution mix or App Store rating count shift perceived value.

On the stealth app the team deliberately removed features to focus on improving the ones users care about; she advises starting with one feature and seeing what users demand.

GlamUp started with $10-20K as its entire runway. No paid acquisition: growth came from viral organic content (UGC and faceless accounts). All money is reinvested in the business or new apps.

Ideas come from watching the market, not personal problems. Go on TikTok and Instagram, find a niche with many passionate creators/viewers but few apps (e.g. cooking); if a few apps there make $50-100K MRR it's a good niche. Avoid saturated categories like health and fitness. Study viral hooks/keywords in the niche and what apps with the same audience do for content.

Day 1: research the niche, market and competitor apps. Day 2: in Figma, draw variations of the screen meant to go viral on social (not the onboarding), borrowing design components from apps in and outside the niche. Day 3+: distribution or build. Build a simple app before posting so early viral content doesn't waste users; onboarding/paywall can be optimized later.

On the stealth app the team focused on making a good product for users; people began finding the app through family and friends.

On one of the host's AI apps, weekly started at $6.99; A/B testing $9.99/week against it with a $50 yearly option. Sample is still small.

AI headshot app built in 2024 with an intro offer ($1.99 first month then $6.99). Nobody paid, so the developer lowered the price. Host and guest advised raising the price and fixing the plain-looking paywall instead.