Store build · SEO and GEO · Content and email
A new store, €0 on ads: half of the revenue comes from ChatGPT and Perplexity
How we built Pinlandia.bg as a brand rather than just a store: a custom build, two businesses on one site, and six months of organic growth in which AI assistants brought in half of all sales.
01 · The brief
A brand, not just a store
Pinlandia.bg was never meant to be just another online badge shop. The client was building a brand, and the website was a long-term asset that would grow with it. So we ranked the requirements not only by what launch day needed, but by what the site would be asked to do a year or two later.
- 1
A product that grows in stages
Launch with one size and one type of button badge, then add new diameters, rectangles, hearts and other shapes. The catalogue had to absorb all of it without a rebuild.
- 2
Personalisation inside the store
Buyers upload their own artwork or generate an image with AI, and see the finished badge before they order.
- 3
Two businesses on one site
A store for consumers, plus a separate offer for companies: merchandise, events, holidays, employee gifts.
- 4
Organic growth only for the first 12 months
Ads come later, and only where they can speed up something that already works.
- 5
Scale further down the road
A marketplace for local artists, and a multilingual version for the Balkans and Eastern Europe after 18 to 24 months.
02 · The decisions
What that meant for the build
A custom build instead of an off-the-shelf platform
A configurator with mockups, file uploads, business enquiries and a future marketplace don't fit into a template store without compromises.
We designed for the owner, not just the buyer
Most of the admin work is automated, so the time goes into production.
An architecture where the two halves don't cannibalise each other
The site is split by search intent, in Google and in AI assistants.
A structure ready for new languages and markets
No rewrite later, when the brand goes beyond Bulgaria.
03 · SEO architecture
One site, two intents
We split the site by what the searcher actually wants. Each half goes after its own queries in Google and in AI assistants, without taking rankings from the other.
Behind the scenes: what the owner sees
- Status emails for every order
- Speedy, Econt and BOX NOW integrations
- Alerts for new orders and warranties
The owner spends time on production, not paperwork.
04 · The first six months
From launch to steady monthly revenue
The store went live at the end of March 2026. The plan was simple: no paid ads, organic growth only, and the patience to see how the market responds.
- Day 0Launch, end of March 2026
- Day 40First business order: 150 badges
- Day 60 to 70First order from a complete stranger
- Months 5 and 6€300 revenue each month, all organic
Months 1 to 3: foundations and testing
We tried content formats, new site features, and A/B tests on placement, colour and messaging. The goal wasn't a quick sale; it was to learn what makes people stop, browse and order.
We kept upgrading the site after launch
A new designer where buyers work with AI or their own files; pages with real examples, so people can see what is and isn't possible; plus more changes to navigation and checkout.
Months 5 and 6: €300 revenue every month
Two months in a row, entirely from organic channels. It's not a big number, but there is a €0 ad budget behind it: just optimisation, content, email and word of mouth from the first customers. And it keeps growing.
05 · Where revenue comes from
Half of all sales come from AI assistants
- 50%ChatGPT and Perplexity
- 50%Organic search, social media and referrals
Traffic from ChatGPT and Perplexity arrives with UTM tags. We cross-checked them in GA4 and asked some customers how they found the store.
06 · The tests
What testing taught us
Checkout: three steps were too many, one was too intimidating
People dropped off along the way.
The long form put people off before they even started, and left us with no contact for a reminder.
Step one captures the contact, step two completes the order.
The homepage button
We tested three messages and several shades. Shown here in English; the test ran in Bulgarian.
The simplest one won.
07 · The channels
Which channels delivered
Beyond AI assistants, two of the four social channels bring in orders. Visual content works best.
The main source of consumer orders
The surprise: it brought two business orders, not consumer ones
A showcase of examples
A showcase of examples
We can't attribute orders to Instagram or LinkedIn yet, so we don't claim any.
Email: four automated flows
- 1Welcome
- 2Abandoned cart
- 3Order status
- 4Promo code 7 days after the order
08 · What's next
The site is ready for the next stages
These are plans, not finished work. What matters is that none of them needs a new site.
- New badge shapes and sizes
- A marketplace for local artists selling their own designs
- New languages for the Balkans and Eastern Europe
- Paid ads after the first 12 months, only where they speed up what works
Stack
- React + Vite
- Supabase
- Gemini API
- Speedy, Econt and BOX NOW APIs
- GA4
Launching a new online store or brand? Start with a free check of how Google and ChatGPT see you.
Request a free checkVerified via:
- GA4
- UTM tags from ChatGPT and Perplexity
- Customer conversations
- Real invoices