Content Demand Research: Finding What to Actually Write About
Stop Guessing, Start Mining Real Demand
Most content calendars get filled by intuition: someone decides a topic sounds interesting and writes about it. A more reliable approach treats topic selection as a research routine with real inputs, your own first-party data, external demand signals, and live conversation, run on a fixed cadence rather than an occasional brainstorm.
This lesson lays out a structured, repeatable routine that surfaces genuine demand before you write a single word, so content time goes toward topics people are actually searching for and arguing about, not just topics that feel timely.
Quick Summary
- Your own analytics (search console data, on-site search, past post performance) is the single highest-value input, since nobody else has access to it
- Reddit and niche forums reveal contested, high-engagement topics faster than most keyword tools, especially when you read the top comment rather than just the post
- Google Trends' default 12-month view hides seasonality, always expand to the 5-year view before calling something a genuine trend
- "People Also Ask" style tools turn one strong topic into a full content outline, since the nested questions map directly to subtopics
- A search query and a piece of public content serve different psychological modes, a query is asked privately in low-status mode, a public post needs to add tension or a stance, not just restate the question
Start With Your Own Data First
Before looking anywhere external, mine what you already have. This is the one category of signal genuinely unavailable to competitors.
- Search Console (or equivalent): filter queries containing "how," "why," "vs," and "best." Sort by impressions and look specifically for high impressions paired with low click-through rate, that combination is real demand your existing content isn't satisfying.
- On-site search and past content performance: which existing pages or posts got unusually high saves, shares, or repeat visits relative to their reach? A high save-to-view ratio signals genuine utility, not just passing interest, and that topic is worth a deeper follow-up.
- Any AI-search-specific reporting your analytics platform offers: several platforms now report AI Overview or AI-answer impression data separately from classic organic clicks; watching that trend specifically surfaces which topics are gaining visibility in AI-driven search before it shows up anywhere else.
Do not skip this step because external signals feel more exciting. Your own first-party data is the one input nobody else researching the same space has access to, which makes it the highest-differentiation source even though it takes the most discipline to check consistently.
External Demand: Forums and Communities
Reddit and niche community forums surface real, unfiltered pain points faster than most keyword research tools, provided you search them correctly.
A focused routine: pick 2-3 relevant communities, sort by "Top, This Week" rather than "Hot" (Hot favors recency over genuine engagement), then apply two specific reading techniques. First, read the top comment on a thread, not just the original post, the actual pain point is frequently buried in the highest-upvoted reply rather than stated plainly in the post itself. Second, look specifically for threads with a high comment-to-upvote ratio, that pattern signals a contested topic people are actively arguing about, which is exactly the kind of topic that supports a strong point-of-view piece rather than a neutral explainer.
Note any question that recurs three or more times across a month of browsing, that repetition is a strong signal the topic deserves a comprehensive piece rather than a quick answer. Set a hard time limit on this research (10-15 minutes per session); community browsing is deliberately designed to be a time sink, and an unbounded session easily turns a quick research routine into an hour of unstructured scrolling.
Trend or Fad: The 5-Year View Test
Google Trends' default view shows the past 12 months, which hides whether a rising topic is a genuine trend, a one-off spike, or ordinary seasonality. Before committing meaningful content investment to an apparently "hot" topic, expand to the 5-year view and check the shape of the curve.
A vertical spike that collapses shortly after is a fad. A broad, smooth curve sustained over 6 or more months is a genuine trend. The same spike recurring at the same time each year is seasonality, not a trend at all, worth planning content around annually but not worth chasing as if it were new. This one check prevents a large share of wasted effort chasing topics that were never going to sustain search demand.
Turning "People Also Ask" Data Into an Outline
Question-hierarchy tools (the kind that expand a search topic into its nested follow-up questions) are one of the fastest ways to go from "here's a promising topic" to "here's a complete content outline." Enter your strongest candidate topic and treat the resulting child questions as the actual subheadings for a comprehensive piece, since they represent the real, varied ways people ask about that same underlying topic.
This turns topic research directly into structure, rather than requiring a separate outlining step, and it tends to produce more comprehensive coverage than outlining from memory or assumption, since it's grounded in actual aggregated search behavior rather than a single person's mental model of the topic.
The Query-to-Post Translation
A search query gets typed privately, in what amounts to a low-status information-seeking mode ("how do I do X"). Public content, by contrast, gets read in a more public, status-aware mode (readers are implicitly asking "who here is sharp, who's wrong, what am I missing"). Directly restating a search query as a headline usually underperforms; the more effective move translates the query into its underlying tension.
| Query pattern | Underlying tension | Content angle |
|---|---|---|
| "How do I do X" | A real skill or process gap | A tutorial or framework piece, tends to perform well as a saveable, structured asset |
| "X vs Y" | An unresolved comparison | Take an actual side rather than staying neutral, generates more discussion than a balanced comparison |
| "Why did X happen" | An unarticulated frustration | A diagnosis piece that names the pain the reader hasn't yet put into words themselves |
| "Is X dead" | A status anxiety about a discipline or tool | A contrarian piece backed by real data or experience, weak without genuine evidence behind it |
One practical rule: build one piece of content around one core query pattern. Bundling several distinct query intents into a single piece dilutes the hook and makes the resulting content harder to promote around one clear angle.
Balancing a Content Calendar: Arc vs. Live Signal
A useful split for an ongoing content calendar: roughly 60% of scheduled slots follow a pre-planned thematic arc (so the overall body of content tells a coherent, building story rather than reading as scattered topics), while roughly 40% stay open for whatever this week's actual research surfaces as the strongest live demand signal.
When a live signal strongly contradicts the pre-planned arc, follow the signal. The arc exists to prevent random scattering, not to override genuine evidence about what an audience actually wants right now. If a particular piece unexpectedly performs very well, a deeper follow-up on that same topic generally outranks whatever the arc had scheduled for the following period.
A Repeatable Weekly Routine
A structured version of this research, run consistently (a reasonable target is around 40 minutes weekly), tends to outperform occasional, unstructured brainstorming by a wide margin, mainly because consistency compounds: recurring topics become visible over multiple weeks in a way a single research session can never reveal. Score each candidate topic against a few axes, demand (did more than one source surface it), proof (do you have first-hand data or experience to draw on, this should count more heavily than the other axes), tension (is there a genuine disagreement to engage with), and fit (does this topic serve your actual audience and goals). Then assign the top few candidates to the coming period and let lower-scoring ones sit in a dated backlog, discarding anything that goes unused for several weeks, since an unused signal that old has likely already expired.
The One-Line Takeaway
Mine your own first-party data first since nobody else has it, verify any "hot" topic against a 5-year trend view before committing to it, and translate search queries into their underlying tension rather than restating them directly, that combination turns content planning from guesswork into a repeatable research routine.
Related Concepts
- Topic Clusters and Pillar Pages, the output of this research routine feeds directly into building a topic cluster.
- Search Intent, the query-to-post translation table here is a content-strategy application of search intent classification.
- Content Strategy, this lesson's routine is the demand-research input layer that a broader content strategy is built on top of.