The problem is rarely the writing
When a page underperforms, the instinct is to blame the draft: not deep enough, wrong tone, too short. Occasionally that's true. Far more often the page was finished before it was started, because nobody decided which reader it served or what specific question it answered.
Publish forty pages without those decisions and you don't get forty chances to rank. You get forty pages splitting whatever authority the site has, several of them chasing the same query, and a graph that looks flat despite genuine effort.
Publishing less, on purpose, beats publishing more, on instinct.
The order that works
1. Decide the reader, in one sentence
Not a persona document. One sentence: who is typing this, and what do they need to do next? If you can't write it, the page has no center, and everything downstream (structure, depth, examples) becomes guesswork.
2. Read the SERP before planning the page
What already ranks tells you what Google has concluded the query wants. If the top ten are comparison tables and you write a personal essay, you're not competing. You're submitting the wrong format. I open every result, note the page type, and count how many are doing the same thing. That count is the brief.
3. Commit to a page type
Most content that "just won't rank" isn't low quality; it's the wrong shape. A definition page can't win a query where searchers want a tool. A listicle can't win where they want a step-by-step. Decide the shape explicitly, before the outline.
4. Then write
By this point the outline mostly writes itself, because the hard decisions are already made. This is also why briefs built this way take pressure off writers: they're not being asked to guess intent while also producing prose.
What changes now that AI answers sit above the links
Being citable now matters alongside ranking. AI summaries lift sentences that make clear, self-contained claims: a specific number, a named process, a direct answer in one line. That rewards exactly the writing that was already good: concrete, structured, committed. And it punishes hedged padding, which was always worth cutting anyway.
Practically, I now write at least one section per page that answers the core question in a single quotable paragraph, near the top. Not for the algorithm's sake, but because it's what a hurried reader wanted too.
How to tell if your plan has this problem
Take any ten published pages and ask two questions of each:
- What single query is this page for?
- Which other page on this site also targets that query?
If you can't answer the first for more than half of them, the plan was never a plan. If you keep answering the second with a real page, you have cannibalization, and consolidating is probably the fastest win available to you. Either way the fix starts in a spreadsheet, not in a draft.
What I'd do first
Before commissioning anything new: audit what exists, map each page to one query, find the overlaps, and pick the ten pages with impressions but poor positions. Those already have Google's attention. Fixing them is faster and cheaper than earning attention from scratch, and it tells you whether your content problem was ever a volume problem.
Aashish Rai is an SEO and content specialist in Kathmandu, Nepal, working on content strategy and on-page SEO. More about me · What I do