Building distribution before code: how Sway AI hit $2M ARR in 6 months

Executive overview

Most founders build the product first and figure out distribution later. Daniel built an Instagram following, charged strangers $15 for manual profile reviews, and had thousands of paying customers before writing a single line of code.

The result: Sway AI, a dating profile analyzer, went from zero to $2M ARR in six months. He now applies the same principles as head of product at Cal AI, acquired by MyFitnessPal.

Build and validate distribution before you build the product.

Validating the idea before building

• Spotted demand by browsing the Hinge subreddit — people publicly posting profiles asking for feedback
• Hypothesised a larger private audience unwilling to ask friends but willing to pay an app
• Started an Instagram page posting dating tips ("how I cook on Hinge") before any app existed
• Added a pre-order link to the bio after a few weeks of organic traction
• Charged $15 for manual profile reviews via DM — people paid before an app existed
• Proof of willingness to pay from real strangers de-risked the build

The TikTok content system

• Studied high-performing slideshows from a "men's improvement" TikTok account and repurposed formats
• Designed backgrounds in Figma; used an Excel sheet to auto-populate slide text in seconds
• Batched a full month of content in one to two hours
• Launched with fully unbranded content to warm the algorithm, added branding only after traction
• CTAs initially only on the last slide; once views were reliable, embedded branding throughout
• One Christmas Day post hit 4M views and drove $5–6K in single-day revenue

Scaling from organic to paid

• Reached $30–40K MRR purely through organic social and influencer collaborations
• Partnered with influencer "Defund Simping" — ads ran to his existing audience, not cold traffic
• Controversial ad creatives drove high comment engagement and strong conversion
• Key constraint: relying on one influencer limited creative variety — build a pipeline early
• Dating niche is difficult for influencer deals because creators sell competing coaching products
• Paid ads were the main growth lever once organic proved the model

Paywall and monetisation design

• Onboarding structured as a sales funnel: clarify why the user is there → label the problem → show the dream result
• Videos explaining the app's value precede any questions — removes confusion before the paywall
• Questions use multiple choice to keep friction low and momentum high
• First paywall: carousel of eight value slides conveying subscription benefits, then a single unlock CTA
• Defaulted to annual plan to collect more revenue upfront and fund growth
• Added a credit upsell for users who re-ran the analysis — contributed ~11% of total revenue
• AI photo packs (10 photos for $49.99, single for $40.99) became the biggest upsell by enabling "whale" spending with no cap
• Aha moment (profile rating and feedback) placed after the paywall to prevent drop-off mid-onboarding

App idea framework

• Three criteria for a strong app idea: novel mechanic, easy to demonstrate in a short video, large existing market
• Novel mechanic in an established category beats a blank-slate idea — reduces the variables you're testing
• Find a content niche that already goes viral on social; build a product that plugs into it naturally
• Influencers are cheaper and more available in some niches (fitness) than others (beauty, dating coaching)
• Assess creator economics: if promoting your app cannibalises the creator's own product, they won't convert

Growth strategy lessons from Cal AI

• Influencer marketing offers guaranteed views; organic UGC depends on going viral — influencers are more consistent
• Influencers let you test one variable (does the product convert?) rather than two (can I go viral AND convert?)
• Cal AI's three key onboarding screens — camera, loading, food detail — are optimised to be immediately legible in creator videos
• Groups, streaks, progress photo sharing, and leaderboards added to build retention and product-led growth
• Milestone badges (e.g. 365-day streak shareable to Instagram) drive social sharing without paid spend
• Speed matters: a two-week window of motivation is enough to ship a major feature end-to-end

Design principles

• Build a mood board of great apps; revisit regularly to understand what makes them work
• Prioritise distinctiveness — products that look unlike anything else attract curiosity and sharing
• Simplicity in marketing-facing screens trumps aesthetic complexity: if the app can't be understood in three seconds of a creator video, redesign it
• Apply data-driven testing (A/B, retention metrics) at scale; lean on intuition and psychology for novel features