From ~10 Failed Apps to GlowUp — How a 24-Year-Old Solo Canadian Builder Used $0 in TikTok Ad Spend to Reach $800K in a Year and $1.2M ARR
GlowUp is a makeup app: upload a look you love, AI applies it to your own face and suggests the products to recreate it. After ~10 flops, Louis-David Paul-Hus launched it with $0 ad spend via TikTok and reached $800K in a year and $1.2M ARR.
Louis-David Paul-Hus@LouisDavidPHThe pain point, and how they found it
“That makeup is cute — but how would it look on me?” Scrolling social, people see a look they love but stall: will it suit my face, and what do I buy to recreate it? That uncertainty is the core friction in cosmetics buying. GlowUp targets exactly that: upload one photo of a look you like, and AI previews it on *your* face and suggests the products to recreate it. The pain (will it suit me / what do I buy) is itself the product’s entry point.
GlowUp lets you upload a photo of a makeup look you love; AI then previews that look on your own face and suggests the cosmetics you’d need to recreate it for real. The core is virtual try-on, makeup steps tailored to your features, and product recommendations based on skin tone. It’s operated by Viral Tech Inc., bills itself on the App Store as the “#1 AI makeup assistant,” and reports 500,000+ users.
It was built by solo Canadian developer Louis-David Paul-Hus (@LouisDavidPH). As he states on X, he’s a 24-year-old solopreneur, and before GlowUp he had shipped roughly ten apps that all failed. What changed wasn’t his tech but his *order of operations*: decide the distribution wedge (TikTok) first, then build the product — on the premise that ‘ship it to the store and they’ll find you’ no longer works.
Technically he built the app in the no-code/low-code tool FlutterFlow. He first tried RevenueCat for subscriptions but hit friction with FlutterFlow and migrated to Adapty. Rather than training a flashy in-house model, he ‘translated’ existing AI into a makeup experience and shipped fast with no-code to validate — a stack an indie builder can copy directly.
Growth was relentlessly ‘TikTok at $0 ad spend.’ The title of his own case-study video is the summary: “My mobile app made $800K in 365 days.” Multiply that by paywall optimization that plugged the leaks in monetization (below), and ARR climbed to $1.2M.
From the founder (primary source)
The repeatable playbook
- 1Invert the order: stop building-then-distributing; pick the channel (TikTok) and the angle first, then build for it
- 2Make the core an instant before/after (snap a look → AI applies it to your face) — designed for short-form video
- 3Go all-in on $0 organic TikTok at first, learning which angle goes viral yourself at zero CAC
- 4To scale, pay creators in equity, not cash — enlist many promoters with no upfront ad spend
- 5Measure paywall reach first (GlowUp started at ~20%), then A/B the placement (MRR $30K → ~$100K)
- 6Don’t offer too many price options; converge on a single middle price to cut hesitation and lift conversion
- 7Translate, don’t build: ship existing AI fast with no-code (FlutterFlow) and validate in small, fast loops
GlowUp wasn’t a first-try hit. By his own account, Louis shipped roughly ten apps before it and all failed. What made the difference wasn’t talent but abandoning the ‘build-then-distribute’ order — the pile of failures became the foundation for learning to design distribution first.
Deep dive
【Deep dive】The essence of GlowUp is *sequencing*. A builder with ~10 failures behind him won not by changing his tech but by changing the order in which he did things. Here is the reproducible breakdown.
■ He inverted the common thread of his failures: stop building-then-distributing. His prior ~10 apps followed the ‘build something good and they’ll find it’ order. But shipping to the store finds no one. Facing that plainly, Louis inverted the sequence: *first* pick the place you can distribute (TikTok) and the angle that lands, *then* build a product optimized for it. GlowUp’s core experience — snap a look you love, AI applies it to your own face — is itself designed to be instantly legible and imitable on TikTok. Same engineering ability, opposite outcome, just from reordering.
■ Distribution went all-in on $0 organic TikTok. He started not with paid ads but with organic TikTok posts. ‘Your ideal makeup, applied to your own face, just by taking a photo’ reads as an instant before/after — perfect for short-form video. At zero CAC he could read demand and personally learn *which angle goes viral*, then hand that winning formula to the next stage (creators). The title of his own video, “$800K in 365 days,” rests on this $0 acquisition base.
■ He paid creators in equity, not cash. To scale distribution, Louis is reported to have partnered with TikTok/Instagram creators using equity rather than cash payments. For a cash-poor solo founder this works on two levels: (1) he could enlist many creators without fronting ad spend, and (2) creators gained ‘the more it grows, the more I gain’ incentives, making their posts more committed than one-off paid ads. He built the early growth engine on *sharing the upside*, not on a marketing budget.
■ Plug the leaks: paywall optimization more than tripled MRR. Behind the growth, Louis tightened monetization. Per Adapty’s official case study, only ~20% of users were even *reaching* the paywall at first. After A/B testing placement and picking the optimum, MRR jumped from $30K to nearly $100K. No matter how hard you drive traffic, a paywall users never see earns nothing — the same inflow can yield a 3×+ difference purely from how it’s shown.
■ Pricing: ‘don’t offer too many choices’ was the answer. He also found that presenting multiple price plans up front actually *lowered* conversion (his hypothesis: too many options overwhelm users). After several experiments, converging on a single not-too-high, not-too-low middle price produced the best results. Pricing isn’t ‘charge more to earn more’ or ‘charge less to sell more’ — it’s a design problem of reducing hesitation. Stacked together, these optimizations grew ARR from $400K to $1.2M.
■ Tech: ‘translate,’ don’t ‘build’ — and iterate fast with no-code. GlowUp was built in FlutterFlow; subscriptions moved to Adapty after RevenueCat proved a poor fit. Instead of training a giant model, he translated existing AI into a makeup experience and shipped fast with no-code to validate. That ‘small, fast loops’ operating mode is also why ~10 failures became fuel. The indie moat isn’t frontier tech itself — it’s the product of (sequencing × $0 distribution × monetization tuning).
