You spend an hour in a topic-ideation meeting and end up settling on the same “feels right” ideas. After writing and publishing, organic traffic is in the single digits — because readers never search with the word you had in mind. The problem isn’t the writing; it’s that the starting point was already off. What you should really reference is the set of elements in a writer’s brief: figure out who you’re writing for before you start.
Treat the user persona as the starting point of topic selection, walk the four-step chain of “task — pain point — search term — title”, and leave a reviewable record at every step. Don’t use age-and-gender tags as your persona; look at “what is this person doing, where are they stuck, and what would they type into the search box”. Once this flow runs smoothly, both topic hit rate and title CTR climb at the same time.
Where the persona comes from: four cross-verifiable sources
A description like “women aged 25–35, in first- and second-tier cities” is useless for topic selection, because it can’t derive a single search term. A useful persona lands on a concrete task scenario: is she choosing baby food for a two-year-old, or picking an expense-reporting system for her company?
Cross-referencing four sources is far more reliable than running one survey:
- Customer service and pre-sales records: high-frequency questions are a ready-made topic pool, and in the users’ own words.
- Search Console queries report: filter for queries with impressions but extremely low CTR — those are “questions with demand you haven’t answered well”.
- Comments and community chats: users’ phrasing is closest to what they type into the search box.
- Competitor content comments and negative reviews: where rivals failed to explain clearly is your entry point.
Organize the results into three-column cards: task, pain point, possible search terms. From then on, pull topics directly from the cards instead of brainstorming from zero every time.
Translate needs into keywords, not internal jargon
The same need can be expressed in five or six ways in the search box, while the team often uses only one made-up name internally. The place this translation step most often fails is using internal terminology as the search term.
The approach is to build a three-column reference table: the left column records the user’s exact words, the middle column holds the real search terms, and the right column states this article’s positioning. Pull search terms from autocomplete, related searches, and Q&A sections — don’t invent them.
Figure: Persona-driven content topic selection: from need to title — key points (compiled by Operations GO)
| User’s exact words | Corresponding search-term form | Article positioning |
|---|---|---|
| Don’t know which to pick | how to choose / compare / difference | Decision guide |
| Want to spend less | save money / cheap / dupe / options | Value-for-money list |
| Made a mistake, can’t fix it | error / failed / how to solve | Troubleshooting |
| Just starting, don’t know where to begin | beginner / starter / steps / tutorial | Getting-started guide |
| Worried about being ripped off | pitfalls / cautions / is it reliable | Risk checklist |
The same need can map to several intents, so confirm the primary intent before writing — don’t let one article try to be both a tutorial and a comparison at once. For the judgment criteria, cross-reference the four-type search intent classification and align intent with landing-page format first.
Use three-dimension scoring to prioritize topics
Once you’ve collected needs, there are usually dozens, and writing them all isn’t realistic. Score quickly on three dimensions on a 1–5 scale, weight them, and sort:
- Search scale (weight 4): look at the total search volume of the word family and its long-tail count; a single word with no family has limited value.
- Commercial value (weight 3): can this article naturally lead to a conversion page? If it can’t, it’s pure top-of-funnel.
- Can you go deep (weight 3): does the team have first-hand data, case studies, or operational screenshots? Without them, it’s hard to differentiate.
Topics high on all three go into the must-write queue; ones high on search scale but hard to go deep can first become a roundup or list page; ones high on commercial value but low on search volume go into supporting pages along the conversion path, not judged on traffic alone.
After sorting, don’t write everything at once. Pull 2–3 from the top of the pool each week and keep the rest rolling. Align the pacing with the dual-axis content calendar planned around search seasonality and business rhythm, so you’re not starting to write right when peak season hits.
From topic to title: the first 15 characters decide the outcome
A title has to do two things at once: let the search engine know which query you’re answering, and make users feel there’s a clear benefit to clicking. Structurally, you can split it into three parts.
- Carry segment: put the core search term in the first 15 characters; don’t let it get pushed back by prefixes.
- Benefit segment: signal the content form with words like “list”, “steps”, “compare”, or “pitfalls” to lower the expected reading cost.
- Qualifier segment: pick one of a number, a year, or a target audience — for example “5 steps”, “2026 edition”, or “for small teams”.
Prepare 3 candidate titles for the same topic and look again after a day — you’ll usually find the first version stuffed too full of modifiers. The meta description and title should divide the labor, not restate each other; for the specifics, see the title and meta description CTR details.
Run three quick checks before publishing
Before finalizing, spend ten minutes on three checks — they filter out most self-congratulatory titles. If a check fails, change it; don’t force it out just because “it’s already written”.
| Check | Passing standard | If it fails |
|---|---|---|
| People really search this | Autocomplete fills in the word | Switch to a phrasing that appears in the dropdown |
| Not clickbait | Body delivers on the title’s promise | Shrink the promise, or flesh out the content |
| Not truncated on mobile | First 30 characters express the full meaning | Move the qualifier to the second half |
| No internal cannibalization | No published article covers the same intent | Merge the two, or rewrite one article’s intent |
The last one is often overlooked: two articles on the site fighting over one keyword usually end with neither ranking. Searching whether the site already has same-intent content before publishing is far less work than merging afterward.
Turn persona cards into a daily topic bank
Building persona cards isn’t for display. Pin them to the first column of your topic board; every time someone says “this topic feels good”, first find the corresponding task and pain point in the cards. If you can’t find it, the need isn’t validated yet — hold off. One SaaS company used a high-frequency customer-service question card and 9 of 12 articles produced in three months got organic traffic in their first month, because the topics came straight from users’ own words and the search terms held up naturally.
- The first column of every card always holds the user’s exact words, never internal terminology.
- New needs get checked against the cards first; they enter the pool only when they map to a task.
- Review card hit rate monthly and retire pain points with persistent zero clicks.
| Topic | Search scale | Commercial value | Can go deep | Verdict |
|---|---|---|---|---|
| Expense-system selection | 5 | 4 | 5 | Must write |
| Expense-policy explanation | 3 | 5 | 3 | Supporting page |
| Expense pitfalls checklist | 4 | 3 | 4 | Must write |
When it comes to execution, the moves are concentrated: today export the last three months of high-frequency customer-service questions and organize them into “task — pain point — search term” three-column cards; filter Search Console for queries with impressions and CTR under 1% as a supplementary demand source; fill needs into the three-column reference table, confirming each search term’s real phrasing via autocomplete; score on the three dimensions of search scale, commercial value, and depth, and pick the 3 to write next week; write 3 candidate titles for each and run them through the four pre-publish checks before finalizing. Topic selection done by gut feeling misses seven out of ten times; put your users’ own words on the table and the hit rate takes care of itself.
