Live SEO Experiment

AI Trend Prompt Hub SEO Case Study

This page documents the public SEO experiment behind AI Trend Prompt Hub: a focused prompt library that tests how AI image prompt pages get discovered, indexed, clicked, and cited. The current lesson is practical: broad prompt pages are weaker than specific clusters with real query evidence.

Current Baseline

The site is not blocked technically: sitemap URLs return 200, robots allows crawling, and crawlers are reaching the site. The bottleneck is trust and selection. Google has indexed a minority of submitted URLs, and most visible traffic comes from a small outfit-editing cluster.

Published prompt pages

50

Small programmatic library, not a scraped mass site.

GSC clicks

77

Measured from 2026-06-03 to 2026-06-29.

GSC impressions

3,849

Measured from 2026-06-05 to 2026-06-30.

Sitemap URLs

52 / 52

All sitemap URLs returned 200 on 2026-07-03.

What Changed

From generic pages to validated clusters

Early pages covered many AI prompt trends. Search Console showed that Google was responding more clearly to exact outfit-editing needs: same face, change only clothes, dress change, clothes swap, and photobooth prompts. The current strategy is to expand the validated clusters first instead of publishing many thin trend pages.

  1. 1Find early query signals in Search Console and server logs.
  2. 2Separate validated clusters from generic prompt topics.
  3. 3Add a focused long-tail page only when a query pattern repeats.
  4. 4Strengthen internal links from indexed pages into the new target page.
  5. 5Add examples, FAQs, and strict prompt rules that answer the exact search intent.
  6. 6Resubmit sitemap and inspect whether Google moves the page from unknown to indexed.

Pages Being Used As The Test Cluster

AI Clothes Changer Prompt

Main traffic page for outfit swap and keep-face-same edits.

Same Face Outfit Change Prompt

New long-tail page built from GSC outfit-change query patterns.

AI Dress Change Prompt

Supports dress, shirt, and clothes-change searches.

Korean Photobooth Prompt

CTR opportunity page for photobooth and four-cut strip searches.

What This Demonstrates

  • Search Console driven keyword selection instead of intuition-only page growth.
  • Programmatic content with manual quality gates, examples, FAQs, and related-page routing.
  • Index tracking across sitemap, URL Inspection, server logs, and public crawl signals.
  • Traffic-first SEO thinking: get real impressions and clicks, then decide which traffic is worth keeping.

Next Experiments

  • Track whether the same-face outfit page moves from crawl discovery to indexed status.
  • Add external references from a project README, portfolio notes, and legitimate social posts.
  • Improve title and meta copy on low-CTR pages after new GSC data arrives.
  • Stop expanding pages that remain unknown unless a real query cluster appears.