A client content platform with the organic engine built on top

Context and scope

A Brazilian digital-accounting company, online since 2018, had almost no non-branded organic traffic. The site could not sustain the content formats a competitive category needs, and the editor could not publish a rich page without engineering work.

The engagement covered both halves of the problem in one place. I built the platform on PayloadCMS and Next.js 15 with a hand-built page builder, then ran the SEO consulting, the content strategy, and the content production on top of it. Solo, as a solopreneur, with an intern or freelancer helping on content.

This is the record that shows the combination directly: the platform is the reason the content could be produced at volume, and the content is the reason the platform moved the numbers.

The problem

• Non-branded organic traffic sat in the tens of visits a month, and almost all existing traffic was branded.
• Editors could not compose the page formats the category needed without a developer each time.
• The content model and the SEO structure were not the same thing, so every page type was a new decision.
• A keyword set above forty thousand terms was clutter rather than a plan.
• Content production was a bottleneck in its own right, on top of publishing.
• Multi-language needs (pt, en, es) were coming, and the site was not built for them.

Constraints

• A solo engagement, so the platform, the SEO, and the content all had to share one schedule and one owner.
• A full go-live date in January 2026, with real content already in the plan.
• The client stack lived on AWS, so the platform had to fit it rather than replace it.
• Performance at launch mattered, because the category is competitive on page experience.
• Client details stay private, so the record is written from the engagement scope and the measured outcomes.

What I built

• A content platform on PayloadCMS 3, Next.js 15, and React 19, with 18 collections and around 25 modular blocks in a hand-built page builder.
• A large variant-rich component library, so one block could render many layouts without new code.
• A block-resilience layer so a broken or empty block degrades cleanly instead of taking the page down.
• Role-based access, live preview, and localization across pt, en, and es.
• Cloud and media work on AWS S3 and CloudFront, with Mux for video, and a later move to Amplify.
• SEO treated as an engineering problem: information architecture, structured data, internal linking, and programmatic page structure.
• A keyword program that reduced more than forty thousand terms to 40 core and 400 complementary targets.
• A content engine producing around 30 dense articles a month, in the 2,000 to 5,000 word range.

Decisions and trade-offs

Outcome

Non-branded organic moved from 50 to 200 visits a month, to 10,000 a month at six months, to more than 20,000 a month at nine months. The platform shipped at a PageSpeed score between 82 and 92, and that performance is part of why the content ranked. Later marketing scripts added after launch moved the score down; loading them properly is the fastest available win on the site.