Nobody trusts a dealer’s brochure. They trust other owners.
A buyer about to spend $6,700 on a Chinese mini-excavator doesn’t trust the dealer’s website — they find the owners’ group. Rippa had one: eight months, 383 owners, 57,494 messages. Nobody had turned it into a website. We did.
- Services
- Custom SaaS development, Research and analytics, Data and integrations, Digital marketing, E-commerce
- Year
- 2026
Results
57,494
community messages analysed, from 383 owners
60
knowledge articles across 8 categories, from real owners
3
languages, with auto transliteration for Uzbek-Cyrillic
100%
routes pre-rendered to static HTML in all three languages
What was missing
Buyers are cautious about a Chinese mini-excavator in a market full of clones and unreliable information. Before they call, they need to know which model fits their work, what breaks, what it costs to run, whether they can register it, how to haul it. A brochure site answers none of that, and no dealer’s marketing copy is trusted on it. The answers existed — in the #RippaVody owners’ Telegram — but not on a website.
Before
- A cautious buyer facing clones, half-truths and no way to self-qualify
- No trustworthy source on what ownership actually costs
- Model choice, registration, haulage and servicing all unanswered pre-purchase
- Eight months of owner knowledge locked in a Telegram export
The insight
Any dealer can list horsepower. Almost no dealer can show eight months of real ownership experience — what breaks, what to check, what it costs — in the owners’ own words, with names and dates attached. That’s the defensible asset. Not the catalogue. The community.
What we built
A trilingual React storefront with five model pages (R06–R22) plus a story-driven flagship page for the R06, full spec sheets, USD ‘from’ pricing with in-stock and pre-order badges, a 12-type attachments catalogue with compatibility, parts availability and one-tap Telegram, WhatsApp and phone contact. The differentiator is a knowledge base mined from the full Telegram export: a Python pipeline parses 58 HTML files into a structured message stream, rebuilds reply chains into scored threads, promotes the top 200 into article candidates, and matches owner photos to each topic. Selected threads are rewritten for clarity with every quote keeping its real author, date and message id, and the categories map directly to the questions a buyer asks — model choice, registration, haulage, running costs, earning, spotting a clone. Russian and Uzbek-Latin are authored by hand; Uzbek-Cyrillic is generated by a longest-match transliterator. After the Vite build, Puppeteer freezes every route in all three languages to static HTML with per-language JSON-LD and sitemaps.
After
- Five model pages with full specs, USD pricing and stock status
- A searchable knowledge base of 60 owner-sourced articles with credited authors
- 4,158 owner photos and 1,085 field videos as quantified social proof
- A trilingual, statically pre-rendered site live in production
The R06 model page, with the compactness badge and full description. 
The attachments catalogue, filtered by type, with compatibility and stock badges. 
The owner knowledge base, filtered by topic and by machine model.
Stack
Next case
Three audiences. Nobody introduced them.Building materials
Read the caseContact
Want one of these?
Tell us what is not working. We will tell you what we would do about it.
- Response time
- We answer within one working day.
- How we work
- A remote-first team across EU and US time zones. Everything runs online — your meetings happen where your team already is.