Why I Removed Every K-Pop Guide From This Site

When I rebuilt the guide system on this site, I had 119 published "BEST 10" lists across every region and theme. On paper, it looked like a fully stocked catalog. Then one morning a reader asked me a simple question: are the places on the K-pop list actually about K-pop? I opened the Seoul K-pop guide and read it. The answer was no.

What the K-pop guide actually contained

Ten places were listed under "Seoul K-Pop Spots BEST 10". The actual contents: Kansong Art Museum (traditional Korean painting), the Korean Museum of Straw and Life (agricultural folk objects), the National Museum of Korean Contemporary History, the Sool Gallery (traditional liquor), and several small theater halls. None of them had any meaningful connection to K-pop.

I had built a filter that searched place names for words like 'k-pop', 'idol', 'sm', 'hybe', and 'jyp'. The intent was right. The implementation was wrong: the keyword check was a tiebreaker, not a requirement. When zero places in our database matched those keywords, the generator silently fell back to whatever was popular in the region. The page got created, the slug got published, and the title still said "K-Pop Spots BEST 10".

The audit that made the decision easy

Before deleting anything, I ran a relevance audit across all 119 guides. The result was harder to ignore than I expected: 26 guides had zero places that matched their stated theme, 60 had under 30% match, and only 20 cleared an 80% threshold. The K-pop case was not an isolated bug. It was a systemic mismatch between what the titles claimed and what the pages delivered.

I have hosted over a hundred foreign friends in Korea over the years. If any of them had clicked one of these K-pop guides expecting K-pop spots, they would have ended up at a traditional liquor gallery. That is the moment a travel site loses someone's trust. It is hard to win back.

Why I chose to delete rather than "fix later"

There was a tempting middle path: leave the guides up and improve them gradually. I decided against it for two reasons. First, Google's misleading-content policy is unambiguous — a page titled "K-Pop Spots BEST 10" that surfaces unrelated places is misleading, full stop, regardless of intent. Second, every day those pages stayed indexed was a day they could be clicked by a real traveler. Holding bad content live to avoid a "thin catalog" feeling was trading reader trust for vanity metrics.

I deleted all 119 guides — 238 rows including English and Korean — in a single migration that afternoon. The Guides menu went empty. Then I rewrote the selection logic so it now requires either a strict KTO category match (festival, wellness, camping, etc.) or a real keyword hit in the place name. If a region-theme pair cannot find at least five qualifying places, the generator now refuses to create the guide.

What the rebuild produced

Round one of the strict generator produced 23 guides instead of 119. K-pop, shopping, photo-spots, date-spots, and other themes that the public data does not actually support were removed from the catalog entirely. Every remaining guide was verified at 100% theme match before it went live. The catalog shrank by 80%. The accuracy went from "mostly wrong" to "all correct."

I would rather run a small honest catalog than a large misleading one. If you visit the site today, the Guides menu has fewer pages than it did last week. Every page that is there says what it does, and does what it says. That is the version of this site I want to keep building.