Project 05SEO / GEO
Search & GEO rebuild
Three hundred pages rebuilt for a search landscape where the answer is increasingly written by a machine.
- 300+pages rebuilt
- 20country markets
- GEOcertified by Semrush
01
At a glance
The whole project in three steps.
- The problem
300+ pages across 20 markets with no shared structure, competing with each other just as search moved to AI answers.
- What I did
Audited, consolidated and rewrote the estate for ranking and for citation by answer engines, then handed the standard to 20 offices.
- The result
300+ pages on one architecture, with the self-competition removed. Semrush-certified in GEO along the way.
Skills it shows
- SEO and GEO
- Content architecture
- Structured data
- Site auditing
- Stakeholder enablement
Tools
- Semrush
- WordPress
- Elementor
- Google Analytics 4
- Looker Studio
02
The challenge
Why the project had to happen.
A twenty-country consultancy accumulates pages the way a building accumulates cabling: every office adding what it needed, nobody removing anything. Three hundred-plus pages, no shared structure, competing against each other for the same terms.
Read the full context
The timing mattered more than the mess. Search was shifting from ten blue links to generated answers: AI Overviews, ChatGPT, Perplexity. A page written to rank is not automatically a page an answer engine will cite. Rebuilding for the old target would have meant doing this twice.
03
How I did it
5 steps. Open any one for the detail.
Audit before touching anything
Crawled and inventoried every page: what ranked, what competed, what should not be indexed.
How
Full crawl and inventory: what existed, what ranked, what cannibalised what, and what had no reason to be indexed at all. A meaningful share of the estate was competing with itself.
Fix the architecture, not the pages
Consolidated first: deciding what becomes one page beats optimising four.
How
Consolidation first. Deciding what should be one page instead of four is worth more than optimising four pages, and no amount of on-page work rescues a structure that splits its own authority.
Write for extraction, not just for ranking
Wrote self-contained, attributed passages with structured data so answer engines can cite them.
How
Answer engines quote passages. That changes the unit of writing: self-contained claims, clear attribution, structured data that states plainly what a page is and who stands behind it. This is the GEO half, and it is the half most rebuilds skip.
Instrument it
Tracked position and technical health in Semrush and RankMath, traffic in Analytics.
How
Semrush and RankMath for position and technical health, Analytics for what the traffic did after it landed. Rankings that do not convert are a vanity result.
Hand it over
Left the standard with twenty offices so page 301 follows the same rules.
How
Twenty offices publish; I do not. The rebuild only holds if the people adding page 301 know the rules, so the output included the standard, not just the pages.
04
See it
The page estate
Before — competing with itself
Each tile is a page. Consolidation came first: deciding what should be one page instead of four is worth more than optimising four pages that split their own authority.
05
What it achieved
Over 300 pages rebuilt on a single architecture, with the duplication and self-competition removed.
Content structured for citation by AI answer engines as well as for classical ranking. The work is aimed at where search is going, not where it was.
Certified in On-Page SEO and AI Search (GEO) by Semrush during the project.
06
What I took from it
Deciding what should be one page instead of four is worth more than optimising four pages.
Answer engines quote passages, so the unit of writing changes: self-contained claims, clear attribution, structured data. Most rebuilds skip this half.
A rebuild only holds if the people adding page 301 know the rules, so the output included the standard as well as the pages.