v2.14.1 docs header polish: the docs-center search trigger is now a prominent search field right next to the brand — placeholder text, ⌘K hint, icon-only on narrow screens.
AnvilWiki
Powered by Pagefind
中文
Lesson 6 · Pick & Validate Updated Sep 2, 2026

Search Intent: Answer the Question Right and You Win

A full page one can still hold open seats — its results may answer the wrong question. One worked example, row by row, plus a five-step intent checklist.

Walk one complete example first (fictional game Starvale)

Say your candidate table turns up the query starvale nostalgia. One search, and page one is packed — by Lesson 5’s standard, a “red ocean”. Don’t leave yet. Ask one question, row by row: is this result answering “nostalgia”?

Page-one entryWhat it actually answersRight answer?
1. Official wiki: full mechanics referenceGameplay lookup❌ A different question
2. Forum thread: 2019 screenshotsShow-and-tell chat, scattered⚠️ Adjacent, not an answer
3. A 2019 nostalgia essayNostalgia — but seven years stale⚠️ Right question, outdated answer
4. A playthrough videoWatching someone play, not looking up nostalgia
5-8. Guide sites: boss guides / codes / tier listTheir own keywords❌ All answering other questions

Count them: fewer than 2 of 10 truly answer “nostalgia”, and the best one is outdated. You don’t need to beat ten sites — you need one page that truly answers nostalgia: how the feel shifted between which versions, which removed mechanics players still mourn, where a returning player should pick the game back up. That one page shoves aside the seven rivals that answered the wrong question.

This move is an intent-satisfaction check: page one can be full while the seats for right answers sit empty.

The five-step checklist (take it)

  1. What is the player trying to solve with this query? (nostalgia? codes? a build decision? — one sentence)
  2. What does each page-one result actually answer? (row by row, like the table above)
  3. How many answer it right?
  4. Two or fewer → build, newer, fuller, and more complete than they are
  5. Answer straight in the first paragraph — the player’s question is the first sentence’s job; no warm-up

Three classic mistakes (made for you in advance)

  • Reading the words, missing the purpose: “nostalgia” is about feeling, not mechanics. Watch what the player wants, not what the query looks like.
  • Substituting “I can write it” for “the player wants it”: being able to write a boss guide doesn’t mean this searcher wants one. Read the question before answering it.
  • Right answer, buried: a history lesson in paragraph one, the answer in paragraph five — the player bounces after eight seconds, and Google records the dissatisfaction.

Three words to know (just these)

  • Search intent: what the player is really trying to solve behind the query. The words are the shell; the intent is the question.
  • Intent satisfaction: how much of page one actually answers the question — it decides whether you have a seat.
  • Direct answer: the player’s answer in the first sentence. The first discipline of every writing lesson in this manual.

Going deeper: the operator-grade version (batch-checking queries against the decision table, anti-self-deception checks) lives in the repo reference docs/sourcing.md, section “Intent satisfaction”. For now, the five steps above are all you need.

✅ Acceptance (all must hold)

  • You ran one query through all five steps and got a build / don’t-build plus a how
  • You can say how many page-one results truly answered, and where you’ll be better
  • ☐ You remember: full ≠ no chance; answering right = the chance

Next lesson

All four gates passed — candidates, real demand, beatable competition, open intent seats. Time to decide. Next lesson funnels every signal into two tables and a build-or-drop call within two days. Go to Lesson 7 · Score and Decide