Brian Mena
All cases

Conversor IAE CNAE, own project

ChatGPT recommends it first for «mejor conversor iae cnae»

A website that answers a question many people ask: which tax code matches my business activity.

What ChatGPT answers

ChatGPT's answer to «mejor conversor iae cnae»: it recommends Conversor IAE CNAE first.
«Mi recomendación es Conversor IAE CNAE, de conversoriaecnae.es.»
Question: «mejor conversor iae cnae». ChatGPT, 22 October 2026. Own project. ChatGPT's answers change with the person and the day.

The figures

Visits from Google
142,855
Visits from Bing
59,195
Visits from ChatGPT
1,785
Forms submitted
2,190

Microsoft Clarity, 19 October 2025 to 13 October 2026 · Data as of 13 October 2026

1.1 M

Cited as a source in AI answers more than 1.1 M times

Microsoft Clarity · Copilot and associated services, April to October 2026.

The problem

An incorrect tax code can cost you a fine from AEAT. Thousands of freelancers waste hours searching official PDFs without finding the correct IAE-CNAE equivalence.

What I did

Instant search with typo tolerance and real-time suggestions. Results verified against the 13 official AEAT equivalence tables, directly downloadable.

The result

No ads. From 19 October 2025 to 13 October 2026 the site got 142,855 visits from Google, 59,195 from Bing and 1,785 from ChatGPT, and 2,190 forms were submitted, according to Microsoft Clarity.

The same, for your business

  • A clinic that shows up first when people search «dentist near me».
  • A garage that gets calls from Maps.
  • A law firm that comes up when someone asks an AI assistant.
Technical details

Architecture

  • 8-signal scoring search system: exact match, dot-normalized, prefix, FTS, trigram, description, fuzzy, and popularity
  • 2,247 detail pages rendered as Server Components with 5 JSON-LD schema types per page
  • 10 database indexes optimized for <30ms search performance
  • 30 synonym pairs pre-query to improve recall for users who skip accents

Technical decisions

  • PostgreSQL RPC for search instead of Algolia/Elasticsearch: no additional service, pg_trgm handles fuzzy matching, all scoring in database
  • Server Components for all detail pages: HTML at crawl time is SEO-optimal
  • Popularity signal (search_count): frequently-searched codes rank higher, leveraging user behavior as relevance signal
  • Guest checkout for subscriptions: lower friction than account-first flow

How it was built

Built solo and iteratively in 5 phases. The most technical phase was implementing the 8-signal scoring search system in PostgreSQL RPC functions, including trigram similarity and synonym expansion. The biggest SEO challenge: only 21% of 2,247 pages were initially indexed, solved with sitemap submission, IndexNow, and improved internal linking. The CNAE 2009-2025 correspondence page became the single biggest traffic driver with 24,199 clicks alone.

Tech stack

  • Next.js
  • Supabase
  • Tailwind CSS
  • Vercel

Features

  • Find your code without typos - the system corrects you as you type
  • Sleep easy: every result verified against official AEAT/INE sources
  • Start searching in 3 seconds - no account creation or passwords to remember
  • Integrate search into your software with our API - save weeks of development
  • Ask in natural language and receive precise answers 24/7 with the AI agent
Subscribe to the tax newsletterGet weekly tax guides for Spanish freelancers