Marko Bera
ENSR
Let’s talk

← All work · 06 / 06

Finding clients for website work

Lead Finder + dental sites

A script finds businesses in Belgrade that don't have a website. For a clinic I pick, I build the site before I even call them, so I show up with a finished site.

Role
Builder and seller
Status
Tool done · 12 pitch sites built
Period
June – September 2026
Stack
Python, Google Places API, HTML/CSS/JS
2,394
unique businesses found
751
of them without a website
827
duplicates removed across districts
12
dental clinic sites built for pitches

Video and screens

Lead Finder

A command-line tool on the Google Places API. 20 business categories × 4 Belgrade districts = 80 searches, 3,221 raw results, and 2,394 unique businesses once duplicates across overlapping districts were removed. The output is a CSV with phone, rating, reviews and whether the business has a website. The 751 without one are the warm list. A second scraper focuses on dental clinics only.

Pitch sites

For each chosen clinic I build a complete single-page site from its real data: services, prices, opening hours, team. Every site gets its own palette and structure, and none of them look like a template.

The rules came from getting it wrong. No stock photo stands in for a real person: if the owner or a dentist is shown, there is an empty frame labelled “insert photo”, because a random face where the real owner should be kills trust immediately. Line icons, no emoji. Before a site is sent, every image URL is checked for HTTP 200 and has a fallback.

What I decided

  • Which categories and districts to target.
  • Pitch with a finished site, not a proposal.
  • Never fake the faces of real people.

What the AI did

  • Wrote the Places API tool and the de-duplication.
  • Built the 12 sites from briefs.

Where I corrected it

  • Used stock faces for real team members. I banned it.
  • Defaulted to the same blue-teal “AI look”. Now every site gets its own design direction.

Next case study

heybera.com