Back to portfolioCase study · Data system

Mapping Iran's startup ecosystem: 13,636 companies in one working panel

Belink lists Iran's startups and tech companies one profile at a time. I turned that directory into a cleaned, scored and searchable dataset — and learned what the ecosystem actually looks like when you can see all of it at once.

13636company profiles collected and cleaned
45%of dated companies were founded in 2014–2018
410rows share one email with another company

The working panel

What it looks like inside

Screens from the private panel. Every company-level field — name, address, founders, email, phone, website, score — is replaced by grey bars before capture.

Overview: totals, field coverage and the hot / warm / cold / skip split across all 13,636 companies.Redacted
Overview: totals, field coverage and the hot / warm / cold / skip split across all 13,636 companies.

A look at the data

Ten rows from the sheet

One company from each of the ten largest sectors, exactly as the panel holds them. Company emails and phones are masked here; no founder names or lead scores are shown.

belink_sample.xlsx10 / 13,636
A1fx=FILTER(belink, ISNUMBER(MATCH(sector, TOP10, 0)))
1CompanySectorFoundedWebsiteEmailPhoneSocial
2ALWERAdvertising, marketing & sales2024alwer.iri***@alwer.ir443-••••-90inigtg
3ویزیسانHealth2022vizisun.iri***@sarirai.com173-••••-22igtg
4سیمی نصر مادTrade2012snmaad.comi***@siminasrmaad.com021-••••-93inig
5فضای کار اشتراکی فندکReal estate2024fandak.spacei***@fandak.space937-••••-75in
6هزاره سومFinancial services2021hz3.iri***@hz3.ir021-••••-52—
5 more rows in the full case study

Sample of 10 from 13,636 rows. Contact fields masked on purpose. •••• = masked

Inside the full case study

What you get by email

  • The full dataset as Excel: 13,636 companies, emails and phones masked
  • The five-stage pipeline: collect, clean, classify, score, work
  • Screens of the working panel, with every company-level field redacted
  • Ten sample rows in a spreadsheet view, contacts masked
  • How complete the directory really is, field by field
  • The founding wave of 2014–2018, and what came after
  • The ten largest sectors, and why half the directory has none
  • Four lessons for anyone turning a public directory into a dataset
  • Method and caveats — only aggregate figures, no company or contact data

The full case study publishes only totals and shares. The company-level panel is private and is not shared.

Get the case study

Sent to your inbox

Enter your email and I'll send you a private link to the full case study and the Excel file of all 13,636 companies.

One email with one link. Nothing else unless you tick the second box. Privacy policy