Paul Soubigou

Paul SoubigouJun 24, 2026

Web-to-app is one of those things everyone in the mobile app industry talks about, but nobody tells you how brutal it actually is. We struggled with it for almost a year. But this week it finally clicked! Quick context: web-to-app means getting users to subscribe on the web before they ever download the app. The whole point is to dodge Apple's 30% cut on every in-app purchase. The catch is you have to convince a complete stranger to pay before they've even installed the app. No product in their hands, no trust. It's the hardest sell in mobile. The usual way to do it is paying a subscription to a SaaS to build the funnels (around 1k€/ month). Insane, right? Because you don't even know if it'll work, and sometimes it never does. I skipped all that and built the funnel myself: → A clean custom domain. → A Google tag with events firing on every single page. → A Meta pixel mapping the events correctly. → A funnel dashboard to see exactly where people drop off. Stripe subscriptions synced with the in-app ones, so web and app talk to each other. No SaaS editor boxing me in on design or animation, full control over every screen, and the whole setup is duplicable for any other app in a few hours. Now I just prompt the changes I want. And here's the part I still can't fully believe: our CPAs on the web flow are now in line with our in-app campaigns. Same cost to acquire a paying user, except now we keep the 30% Apple used to take. This is exactly the kind of thing we work on at Rocapine, where we lead UA for multiple apps, both our own and through publishing deals with indie devs and studios. If you've built something great and want to scale it, let's talk ;)

Paul SoubigouJul 8, 2026

If you're building an app first and figuring out distribution later, you're already behind. Most people start with the app and hope it spreads. I start with what's already spreading, and build the app into it. So that's what I did. I found a TikTok format that clearly worked, then built a Gen Z app around it in 2 hours one weekend. I tested the onboarding and the pricing early, until they converted. Then I automated the content creation, so posting now takes me 5 minutes a day. Total time on the actual app: under 10 hours. Everything else went into distribution. In one month it crossed $1,000 and climbed to No.88 in Lifestyle on the French App Store. 100% organic, 0 spent on ads. The format was the bet, not the app. I never had to convince anyone to pay. I just built exactly what a viral format was already making them want. Distribution first. App second. That's the new order. Now I'm going after the US, and that's where it gets tricky. Posting from France means TikTok mostly serves your content to a French audience, so you never really reach American users. That's the wall most people hit. I'm using TokPortal to get around it and post as if I were based in the US, so the algorithm actually pushes the content to the market I want. And I'm not the only one working this way. It's exactly what we do at rocapine : test fast, lead with distribution, and put our weight behind the apps that actually travel. If you've built something great and want to scale it, let's talk.

Paul SoubigouJul 1, 2026

Back in January, we had a strange problem at rocapine: we were shipping apps faster than we could make the ads to test them. Sounds like a good problem, right? It isn't. Some weeks we'd ship close to 10 new apps, and we couldn't produce ads fast enough to properly test even a fraction of them. Testing means a constant stream of creatives: ad images, ad videos, icons, App Store screenshots, all of it in multiple languages. The bottleneck was never building the apps. It was producing all of that fast enough. So I built KISS, our Creative Intelligence Software System. (Jean-Gabriel Boinot-Tramoni came up with the name, we reverse-engineered the acronym after :p) Here's how it works: it takes the best-performing ads from the Facebook ad library and adapts them to our apps. The copywriting, the colors, the style, our app's own UI and visuals, all reworked into something native to us. It runs all the latest AI models in one place, so anyone on the team can launch a full batch of test creatives in 5 minutes, no creative eye required. And these aren't throwaway ads: some of our top spenders now come straight out of KISS. Work that used to take hours now takes minutes. The result: creative stopped being the bottleneck. This is the kind of tooling that lets us test fast and move into scaling faster than any other studio out there. If you've got an app and want to scale it, let's talk.