How we improved the SEO quality of every article Clustea generates
When I launched Clustea, the articles it generated were decent but generic.
After user feedback and a full quality audit, I rebuilt the entire article
generation pipeline. Here's what changed.
Before: Clustea picked a keyword, sent it to Claude, got an article back.
Now, before writing a single word, Clustea:
- Analyzes the top 10 Google results for that keyword
- Identifies People Also Ask questions
- Detects SERP features (featured snippets, video boxes, knowledge graphs)
- Determines the search intent (informational, comparison, how-to, commercial)
Then the article is written to match exactly what Google is showing for that keyword.
Concrete improvements:
→ Featured snippet paragraphs — a 40-60 word answer immediately after the H1,
formatted for Google to extract
→ PAA integration — FAQ questions pulled from real People Also Ask results
→ Dynamic word count — 2,500+ for ultimate guides, 1,200 for definitions.
Not a flat 1,500 words for everything
→ Internal linking — automatic links to related articles on the client's site
→ 16 AI writing tells blocked — no more "delve into", "leverage",
"in today's fast-paced world"
The goal is simple: every article Clustea publishes should be more complete,
more specific, and more useful than what's currently ranking on page 1.
Still early. Still zero paying customers. First real test incoming
with clients on June 3rd.
3 free articles → clustea.com
Nice upgrade. The part I would test with first users is not just “does the article look better than page 1?”, but “can the customer point to the part they could not have written themselves?”
For AI content tools, generic output is only half the problem. The other half is that buyers need a visible quality delta they can trust fast. I would show the before/after audit next to each generated article: missing PAA questions, intent mismatch, weak snippet answer, thin internal links. That turns the tool from a writer into a quality-control system, which is much easier to believe in.
This is a sharp reframe — shifting from "we wrote a better article"
to "here's what was missing and how we fixed it" makes the value
tangible immediately.
The before/after audit idea (missing PAA questions, intent mismatch,
weak snippet answer, low-quality internal links) is something I can
actually build. It would make Clustea feel more like a quality system
than a writing tool.
Adding this to the roadmap. Thanks for the concrete direction.