SEO / GEO & Content: Organic Engines

The framework

Intent first,
then architecture

01 · Map intent & markets

I start from what the SERP already rewards. Keyword data mined from the top 10–30 ranking pages per term tells me what a page has to cover before anyone writes a sentence.

In practice that's fintech keyword research run through NeuronWriter, sorted into intent tiers per market. I build one topic map, then split regional variants for NA, EU and SEA instead of translating a single page three times.

The brief describes what already ranks

Keyword research for a fintech company: how the intent map gets built

02 · Architect the site

I treat crawlability as infrastructure: if engines can't read the site cleanly, no amount of content fixes it. So the architecture gets fixed before the editorial calendar gets filled.

That means sitemaps, schema, canonicals, hreflang and Core Web Vitals worked with engineering across a multi-region site. It also means a crawler-access audit, since roughly 6% of sites block AI crawlers in robots.txt without realizing it. I rebuilt an international site on Webflow for exactly this: speed, control and SEO under one governed CMS.

If engines can't read the site, nothing else matters

Global web & tracking, the multi-region rebuild CMS Rebuild: solo-built Sanity platform

03 · Produce on a scoring system

I give writers a number to hit. Every article starts at a content score of 0, and the workflow moves it up.

A data-built outline takes the score to 40–50 before a sentence of prose exists. The full ~1,500-word draft lands in the mid-70s or 80s, with ChatGPT in the workflow for structure and metadata while a human makes the final call. Localization runs on the same briefs, so three regions ship from one calendar.

Writers hit numbers

Data + AI articles that rank: the scoring workflow, step by step ChatGPT for real SEO gains, where AI sits in the loop

04 · Optimize for both engines

I close the loop twice, because there are two engines now. GA4, Search Console and cohort attribution cover classic search; for AI search I watch whether the content gets cited at all.

The levers differ. In AI answers, mentions, structural clarity and freshness beat backlinks, and those engines trigger far more on specific 7+ word questions than on head terms. So I restructure pages to answer those questions directly and refresh them so they stay worth citing.

Rank in Google, get cited by AI, same content

GEO: what actually matters now, the AI-search playbook Running notes on GEO: field notes, updated as it shifts

Organic systems I have built across three regions:

⌕

Research & intent

✓ SERP-mined briefs

✓ Multi-market mapping

✓ Fintech keyword research

⚙

Technical SEO & IA

✓ Global-site architecture

✓ Core Web Vitals

✓ Webflow / CMS as a marketing tool

✎

Content systems

✓ Scored editorial pipeline

✓ Repeatable ~1,500-word drafts

✓ Cadence & localization

✦

GEO / AI search

✓ Crawler-access audits

✓ Answer-shaped content

✓ Mentions · structure · freshness

What I research and score with:
crawl, keyword and editorial tools

Research &
scoring

Ahrefs

SEMrush

NeuronWriter

Google Search Console

Technical
& site

Screaming Frog ⚡Core Web Vitals

Webflow

Hotjar

Editorial
system

▦Editorial calendars ✎Content briefs ⌗Research frameworks ⇄Localization workflow

Analytics
& AI

GA4 ◔Cohort attribution

ChatGPT in the workflow

Proof

A scoring system that turned ranking into a process

Every article starts at a content score of 0. A data-built outline takes it to 40–50 before a sentence of prose exists, using keyword data mined from the top 10–30 ranking pages. The full ~1,500-word draft lands in the mid-70s or 80s. Writers hit numbers, and the same discipline rebuilt an international site on Webflow for speed, control and SEO.

Read the full case ›

Documented on the blog Data + AI content workflow

0 → 80s content score from blank page to publish, repeatable by the whole team