--- type: case slug: bipa title: Bipa tagline: "From marketing site to acquisition platform." company: Bipa role: "Design Lead (current job)" period: "2025-present" category: "Product & Growth" order: 1 featured: true status: active updated: "2026-07-11" lang: en summary: > Bitcoin-native fintech, 300 thousand users. I turned a marketing site into an acquisition platform: 93 pull requests in 4 months, web signup with 29.9% organic approval against 12.8% for the paid flow, and 40 JTBD interviews that became the team's decision base. metrics: - label: "Web signup" value: "29.9% organic approval (2.3x paid)" - label: "Delivery" value: "93 pull requests merged in 4 months" - label: "GEO" value: "live before all 10 competitors" - label: "Research" value: "40 JTBD interviews, 5 clusters" tags: [fintech, bitcoin, growth, jtbd, seo] --- # Bipa The site had pages. It had no funnel. It had no reliable data. There wasn't a single experiment running. It was a storefront with no cash register. Bipa is a Bitcoin-native fintech in Brazil: 300 thousand users, R$4 billion processed, business and personal accounts with Bitcoin, USDT and a Mastercard card with cashback in BTC. This is where I work today, my current job, not a closed case. I'm Design Lead, part of the product trio with a PM and a tech lead, in the strategic, tactical and operational design rituals, and I coordinate the team's brand designer. Looking at the business as a whole, I saw the biggest opportunity to scale was in growth, performance and marketing: I built the setup and I'm now handing it off. That's how I operate, scale a front, hand back the direction, move to the next one. ## Impact first In 4 months, 93 pull requests merged. A complete web signup converting 2.3x more than the old flow. GEO live before any of the 10 competitors. FullStory jumping from 30% to 100% of sessions recorded. 40 research interviews turned into permanent context for team decisions. A site that only existed now acquires. ## The decision that mattered most There was an untested assumption: that the best move was to capture the lead and send them to the app. We tested it. What did we find? An organic user completing signup in the browser: 29.9% approval. A paid user redirected to the app: 12.8%. Wrong assumptions can kill an opportunity. I built the complete 9-step flow from scratch, from tax ID to approval, with an A/B test across 9 iterations and a deterministic split by channel. Final decision: organic goes 100% to web signup, paid stays in the app. ## What was invisible 70% of sessions weren't recorded. The QR code pointed to the wrong URL. The site was flagged as phishing on Google Safe Browsing. 25 leads a month vanished in a silent API timeout. 162 articles went invisible whenever the CMS went down. Where did all of this show up? In no dashboard. Only when someone looked closely. I fixed tracking at the root: Ortto with no "source-not-set", 40 events cataloged, GA4 from 4 to 14 key events, UTMs from 2 to 5 parameters. Without measuring right, you can't decide right. ## SEO and GEO from scratch From a client-side blog invisible to Googlebot to 51 indexable articles in server-side rendering. Core Web Vitals from 56% "Poor" to optimized. A gap of 2,094 keywords and 2.28 million monthly searches attacked. And GEO before anyone else: llms.txt, speakable schema, citable paragraph. None of the 10 competitors had done it. ## Research as decision infrastructure There was high ambiguity about who Bipa's user is and why they choose the platform. Without research, product was deciding by market assumption. I ran 40 Jobs-to-be-Done interviews myself. A deliberate choice: the context and empathy of a real conversation can't be delegated. Synthesis done with AI, so what would take weeks took days without losing depth. On top of that I layered behavioral segmentation over dozens of transactional and socioeconomic variables, reaching 5 clusters and an ICP defined by real data. The jobs structure became the permanent context of an internal agent. Today the team uses it to validate hypotheses in real time, understand trade-offs before committing resources and set quarter bets. Every team RFC uses the jobs as a mandatory structural reference. ## Result 93 pull requests in 4 months. Web signup 2.3x more efficient on organic. SEO and GEO from scratch, with GEO ahead of the market. Tracking from 30% to 100% coverage. 40 interviews that stopped being an archive and became decisions. The work equals a senior squad of 5 to 6 people operating in parallel. It was one person, operating in every layer because the problem called for it. I talked about design, usability and Bitcoin adoption on Bipa Cast #74, available on YouTube.