ArteLonga runs its entire network with Hostinger at the door — DNS, hosting, and now Horizons. We came as users and we stay as customers. Every small business we bring online is one more domain on Hostinger: horizontal scaling for Hostinger. This is the Data Analyst application, told from that seat.
Case study — Retro Umarizal: menu → live Horizons domain
A cold lead with a single restaurant menu became a paying Hostinger customer in ~24h — from a shareable draft to a domain live on Hostinger Horizons.
Jun 2 — arrive
Retro Umarizal arrives with one asset: a restaurant menu. A new lead enters the network.
Jun 2 — instant MVP
A self-hosted staging MVP goes live on a shareable subdomain — no purchase, no DNS for the client to touch. The owner sees a working site within the hour.
Jun 2 — validated
Client approves on the staging URL → buys retroumarizal.com.br. Investment follows validation, not before.
Jun 3 — converted to Horizons
Rebuilt on Hostinger Horizons; the customer compared both and chose the Horizons build. Live on the apex with a Let's Encrypt cert; the self-hosted MVP retired.
Every partner is one more funnel — and every funnel that lands is more horizontal scaling for Hostinger. Retro Umarizal is data point #1 of the Horizons funnel: real, dated HORIZONS_PROMPT → APP_GENERATED → PUBLISH events, ready to measure.
From first contact to a site live on its own domain, the path a customer travels reads as a funnel — a simple, shared way to see where Marketing Impact, activation and retention actually happen, with a number on every step and a department to answer for each (Marketing owns the open; Product the signup; Finance the payment; Infra the go-live; and so on). I modeled it that way, with data-integrity rails (observed-vs-pending, Wilson CIs); the full method — and the stage-by-team breakdown — is in the study.
ArriveMarketing · acquisition
OpenMarketing · engagement40.3% measured
ValidateProduct · activation
BuyRevenue · monetization
Go-liveInfra · Customer Success
RetainProduct · Revenue
What the numbers are — and aren't. The demo runs on two small samples: the Retro case study above, and my own inbox receiving Hostinger's emails. The second is a sample of one — no general proxy for anything — so the figures speak to engagement (did a message arrive and get opened), not churn or conversion. The point is the method and the honesty about its limits, not the counts.
Customer feedback on Horizons — what I'd report as the analyst
First-hand signal from building retroumarizal.com.br on Horizons — the honest product feedback this role turns into roadmap.
👍 Worked well
Quick intake, feedback incorporated effectively — credit-cheap to a working prototype. Approved within 24h, final edits on free credits. The completion sound is genuinely satisfying. Far easier than deploying and tuning DNS by hand.
🔧 Friction
Initial chat hides the start of the conversation at 100% zoom (recording ↗); some early instructions skipped (e.g. "scrape this site for content"); content needed structured HTML — a URL or pasted copy didn't take (3 exchanges, credit-heavy for content); unclear how to edit after publishing; English by default.
📊 Analyst ask
Surface build telemetry from the start — credits per stage, exchanges-to-prototype, edit-after-publish rate: the unit economics the Horizons dashboard needs.
Stats & insights
The numbers behind the network — and the one engagement metric this sample can actually measure, drawn with its error bar. Method and raw measurements live in the studies: DNS findings · API reference.
1,473commits
286merged PRs
9web services
6partners
Email open rate — the one number I can measure cleanly, with its error bar (hover the bar for the 95% CI)
This is one customer's worth of data — 77 Hostinger emails to a single inbox, effectively n=1. That's the honest limit. Even so, n=1 already surfaces signal: a ~40% open rate (with a wide, honest interval), the Horizons product feedback above, and a 24h menu-to-domain conversion. The method scales; the sample doesn't yet.
Technical portfolio — its own thing
The case study above is told from the customer's seat. The engineering behind it — the platform that produces this data — is a separate artifact: a System Portfolio with two timelines (data and systems), each running to today. It lives at yuri.artelonga.com.br — and runs interactively right here:
The role's five sample projects — what I've already touched
Hostinger lists five real projects for this seat. I've been on the customer side of most of them this month:
Horizons dashboard — growth, engagement, unit economics
I built a Horizons site myself (Retro Umarizal) → the first dated funnel data point; the analyst ask is exactly its unit economics (credits/stage, exchanges-to-prototype).
Measuring Marketing Impact — open → visit → purchase
The open-rate metric above (40.3%, Wilson CI) is this question measured; the MVP-vs-Horizons choice the client made is an A/B-style comparison.
Built and run ArteLonga, our network; real-time quant pricing on live markets.
Tableau / BI tools
Shipped dashboards in QuickSight at Propz; Tableau & Looker are interchangeable for this — the chart above is hand-built with honest error bars.
Partner across teams + Data Platform
Revenue, product, finance & customer success each own a stage of the journey; and I build the event pipeline end-to-end (events → bake → viz), so I work naturally with data & analytics engineers.
Product analytics (Amplitude)
Event funnels are my model here (HORIZONS_PROMPT → APP_GENERATED → PUBLISH); Amplitude / GA-style tools map straight onto it.
Before ArteLonga: Senior Data Scientist, Quant, and years of research. Full history in the résumé, publications in the bibliography. Already at the door as a customer — applying to build from the inside.