I deleted my favorite chapter, and the book got better. Here's how I knew

I loved my chapter four more than almost anything else I'd written — a tense confrontation scene I'd rewritten eleven times until every line sang. Cutting it entirely, on an editor's suggestion, felt like malpractice. The book was better within a week of finishing the cut.

The test that finally convinced me wasn't a craft principle, it was a practical one: I asked three beta readers to summarize the plot after chapter six, without telling them chapter four was gone in their copy versus present in the others' copies. Nobody who read the version without it mentioned missing information. Several who read the version with it mentioned that scene as their favorite, which had confirmed my own bias for a full year.

The uncomfortable lesson: a scene can be excellently written and still not load-bearing, and writers are the worst-positioned people to tell the difference, because craft quality and structural necessity feel identical from inside our own attachment to a scene we worked hard on.

My rule now, painful as it still is every time: if beta readers don't ask a single question that the scene would have answered, it goes, no matter how good the sentences are. The sentences can usually be repurposed elsewhere. The eleven rewrites were not wasted; they just weren't for that chapter.