Type a question into Google, and underneath the first few results you'll see a small box of related questions that expand when clicked. Most people see it as a shortcut. I see it as the cheapest keyword research on the internet — free, always current, and sorted by what people are actually asking right now.
Here's what makes it powerful for content ideas: every question in that box is a phrase real users typed, not something a tool invented. And the questions change as your topic evolves, which means you get a live feed of demand instead of a snapshot from last year's database.
I've built entire content calendars from this box alone. No paid tool. No spreadsheet gymnastics. Here's how to do the same — and where the approach quietly breaks if you're not careful.
Key Takeaways
- The question box on Google is a free, constantly updated source of long-tail keywords you can use for articles, FAQ sections, and even video scripts.
- Expanding the same question across several topics reveals which queries repeat — those are your priority targets.
- Two questions can look almost identical and still deserve two separate pages if the intent behind them differs.
- Personalization and location skew what you see, so cross-check questions in an incognito window before committing.
- A well-answered question can be lifted into an AI-generated answer, so write the answer, not just the heading.
- Don't build a page for every question. Map the ones that share an intent onto one strong URL.
What are People Also Ask keywords, and why do they beat most paid tools?
When you search for something on Google, the results page often shows a cluster of questions related to your query. Click one and it expands with a short answer and a fresh set of questions underneath. That stack, sitting right on the results page, is the source most content teams underuse.
Why it matters more than a keyword database: those questions reflect live search behaviour, ranked by Google's own relevance signals. A question wouldn't appear there if people weren't asking it or if Google didn't think an answer existed. That combination is rare.
The three properties that matter
Not every keyword source has all three. This one does:
- It's free. No subscription, no credit card, no daily query limit.
- It self-updates. Questions shift as news and seasonality change — you see the shift, not a lagging snapshot.
- It's already in question form. You don't need to rewrite a keyword into a heading. The heading is right there.
Compare that to a paid keyword tool that shows you volume and difficulty for a phrase nobody phrases that way in conversation. Useful, yes. But the box on the results page tells you what the query sounds like when a human types it into a search bar at 11pm.
How many questions do you actually get?
Search anything moderately broad and you'll see four questions initially. Click them, and each opens a new set. Chain that expansion a few levels deep and you can easily surface twenty to forty questions from a single seed topic.
I do this by hand for every article I outline. It takes about fifteen minutes and produces a mind-map worth of headings. If you want to speed it up, tools like AlsoAsked or AnswerThePublic automate the expansion and let you export a tree, but the manual pass teaches you something an export never will: which questions keep reappearing across topics. Those repeats are the ones worth writing first.
How to turn that question box into a repeatable content method
Randomly grabbing questions produces random content. The value comes from a process you run the same way every time, so comparisons stay honest.
Step 1: pick a seed and expand deliberately
Start with the core phrase of your topic. Search it, open the question box, and click each question to load the nested set. Keep going two or three levels deep. Write everything down in a single document — not curated yet, just captured.
Step 2: deduplicate on intent, not on wording
This is the step almost everyone skips, and it's the one that decides whether you get traffic or cannibalization. Two questions like "how long does it take" and "how much time does it need" are the same intent — merge them. But "how long does it take" and "how long should it take" can be genuinely different: one is descriptive, the other normative.
Ask yourself: would the same page satisfy both? If yes, it's one entry. If no, they're two angles and possibly two headings.
Step 3: group by intent and map to one URL per group
Now cluster the deduplicated list. Each cluster gets one destination — usually an H2 section in an article, occasionally a standalone page if the cluster is big enough and the intent is distinct.
| Cluster size | Best destination | Why |
|---|---|---|
| 1–2 questions | H3 subsection in an existing article | Too thin for its own URL; adds depth to a page you already rank for |
| 3–5 questions | FAQ block or H2 section | Enough substance to carry a section, not enough to justify a page |
| 6+ related questions | Standalone article | The intent is clearly distinct and the demand justifies a dedicated URL |
Notice the pattern isn't "one question, one page." That approach multiplies thin content and splits your authority. The table above is the rule I use — and the rule I broke early on, to my cost.
Step 4: write the answer first, the heading second
A mistake I made for months: I wrote the heading, then padded around it. Wrong order. Draft the direct answer in one or two sentences, then shape the heading to match what you actually said. This forces clarity and it mirrors how a search engine wants to use your content.
Where do free tools fit, and which ones are worth the click?
The manual method is the foundation. Tools just scale it. Here's how the common options compare:
| Approach | Cost | Best for | Limitation |
|---|---|---|---|
| Manual expansion on the results page | Free | Learning intent; small-to-mid topics | Slow; no export |
| AlsoAsked | Free tier, paid for depth | Visualising the question tree | Free queries capped |
| AnswerThePublic | Free tier, paid for exports | Broad topic ideation | Noisy; mixes question types |
| Rank-tracking suites with a question module | Paid | Teams already paying for SEO software | Overkill if you just need ideas |
If you're starting out, the free tiers are enough. I ran my first year of content planning with nothing but the browser and a text file. The tools became useful only once the manual method had taught me what to filter for.
The problems nobody mentions about this approach
It's not a clean data source. Anyone who tells you otherwise hasn't done it at scale.
- Personalization skews what you see. Your search history and location change the questions. Always confirm in an incognito window without a logged-in Google account.
- Questions drift. What's there today may be gone next month. Capture your list when you research; don't assume it'll still be there when you write.
- No volume data. You can't tell how many people ask a question. Treat repeat appearances across topics as your proxy for demand.
- Cannibalization risk. Build a page per question and you'll compete with yourself. Group by intent — the step two rule above.
The drift issue in particular bit me. I outlined an article in one month and wrote it six weeks later, only to find two of my chosen headings no longer matched anything people were asking. The fix is simple: always do a quick re-check the day you start drafting.
From a search question to an answer that gets cited
The same questions that appear on a results page increasingly feed the AI-generated summaries that sit above organic results. That's an opportunity, and it changes how you should write.
An AI summary pulls a direct, self-contained answer. So when you build a section around a question, lead with the answer in the first sentence, in plain language, without a long wind-up. Don't bury it under three paragraphs of context. The reader gets what they came for; the citation follows.
Two practical habits from my own drafting:
- Answer the question in the first sentence, then expand. Never open with throat-clearing.
- Keep the heading phrased the way people ask it. A natural question makes a natural match.
This is where the free-source advantage compounds. Because your headings come from real phrasing, they align with how the answer engine recognises the intent — no rewriting required.
The question almost everyone skips
Some of the best ideas hide in the questions that look redundant. "What is X" and "why is X important" feel like the same topic — until you notice the first wants a definition and the second wants a justification. Different audiences, different angles, different pages.
Next time you open that box of related questions, don't just collect. Sort. The ones that look repetitive are often the ones nobody has answered well, because everyone assumed they were already covered. That's the gap worth walking into.
