A question post is a post whose main job is to collect replies rather than deliver information. Most of them underperform for a boring reason: the question is badly designed. "What's your favourite productivity tool?" asks a stranger to stop scrolling, search their memory, compose a sentence, and publish it under their real name — in exchange for nothing.
Question design is a craft with four levers, and once you can name them you can fix almost any dead prompt in about thirty seconds.
Why "what's your favourite X?" gets three replies
Four things go wrong, usually at once.
The answering effort is higher than it looks. Open-ended recall is genuine cognitive work. "What's your favourite tool?" means scanning everything you've ever used and picking a winner you can defend. Recognition — choosing between two named options — costs a fraction of that.
Nothing is at stake. Answers that reveal nothing, cost nothing, and risk nothing are also worth nothing to the person typing them. People reply when the reply says something about them.
The frame is too wide. A question anyone could answer feels addressed to nobody. "What's your biggest marketing challenge?" is for everyone; "you're a solo consultant with eleven clients and no assistant — what breaks first?" is for someone.
The first reply is hard. An empty comment section is a blank page. The first person to answer is doing unpaid labour setting the tone, and most people would rather not.
The four levers
Take a flat question and pull each lever in turn:
- Lower the effort. Convert recall to recognition, or cap the answer length. "One word is fine" removes the fear of writing something not good enough.
- Raise the stakes. Ask for a position, a confession, or a number people are slightly nervous to publish. Mild social risk is the engine.
- Narrow the frame. Name the situation, the role, or the season. Specificity is what makes a stranger feel summoned.
- Make the first reply easy. Post your own answer in the comments within a minute of publishing. It sets the length, the tone, and the permission level for everyone after you.
Worked example. Flat: "How do you plan your content?" Fixed: "Batch on Sunday or write day-of? I'm a day-of person and it's costing me — tell me batching is worth it." That version is a binary pick (low effort), carries a stated position (stakes), names a specific tension (frame), and comes pre-answered (easy first reply).
Fifteen formats and where each one lands
Platform culture matters more than format cleverness. A confession prompt that thrives on Threads reads as unprofessional on LinkedIn; a ranking exercise that LinkedIn loves dies on TikTok.
| Format | What the ask looks like | Fits best on |
|---|---|---|
| Binary pick | "Batch on Sunday or write day-of? Pick one." | X, Threads, Bluesky, LinkedIn |
| Fill-in-the-blank | "I finally grew when I stopped ______." | Instagram, Threads, Facebook |
| Rank three | "Rank for a new account: consistency, quality, hooks." | LinkedIn, X |
| Defensible take | "Daily posting is overrated under 10k. Tell me I'm wrong." | LinkedIn, X, Threads |
| Situational judgement | "Client wants a rush edit at 6pm Friday. What do you do?" | LinkedIn, Facebook groups |
| Confession | "Worst typo you've ever published — go." | Threads, Instagram, TikTok |
| Missing piece | "Three ways I do this. I'm sure there's a fourth." | LinkedIn, Bluesky, Mastodon |
| Constrained favourite | "One tool for the next twelve months. Which?" | X, Threads |
| Day-one advice | "One sentence to someone freelancing on day one." | LinkedIn, Instagram |
| Show your screen | "Screenshot your calendar for this week." | Instagram, X, TikTok |
| Number only | "How many drafts before you publish? Just the number." | Threads, X, Bluesky |
| Two-word answer | "Your workflow in two words." | Instagram, Threads |
| Fix mine | "Here's my hook. Rewrite it — I'll pick a favourite." | LinkedIn, X |
| Time anchor | "Where was your account a year ago today?" | LinkedIn, Instagram, Facebook |
| Native prompt | Poll, question sticker, or Q&A box | Instagram Stories, LinkedIn, X, Facebook, YouTube Community |
A few notes on the mapping. The native prompt row is its own discipline — polls, quizzes and Q&A formats remove the writing step entirely, which is why they collect answers from people who would never leave a comment. On Instagram, the question and poll stickers in Stories are the lowest-friction question surface most accounts have, and the answers make good raw material for a feed post the next day. Treat LinkedIn polls as a distinct format with their own rhythm rather than a substitute for text questions. On YouTube, the Community tab is where questions belong; comments under a video are answering the video, not you.
TikTok is the odd one out. Questions there work when they're spoken in the video and answered in comments by people expecting a reply on camera — the platform's comment culture rewards creators who mine replies for the next video. A caption question with no video hook rarely lands.
And two honest gaps: Pinterest and Google Business Profile are not conversation surfaces. Pinterest traffic is search-driven, and questions posted on a business profile read as customer service rather than community. Don't force a weekly question slot onto either.
The bright line between a prompt and bait
This is where a lot of question-post advice goes wrong, so be precise about it. The distinction is not "asking for engagement" — every question asks for engagement. The distinction is whether an answer produces a conversation or just a countable action.
Engagement bait asks for a mechanical response: tag three friends, drop an emoji, type "YES" if you agree, comment GUIDE for the link. These generate volume without content. Platforms have spent years getting better at spotting the pattern, and as of October 2025 the safe assumption is that mechanical-response mechanics get discounted rather than rewarded — and that an audience trained on them will only ever perform the mechanic.
A genuine prompt asks for something only that person could write. The test I use: would a thoughtful reply to this be interesting to a stranger reading the comments? If the answers are all one word and interchangeable, it's bait wearing a question mark.
The other half of the line is what happens afterwards. Asking a real question and ignoring the replies is worse than not asking, because it teaches people their answers vanish. Every question post is a commitment to show up and turn those replies into actual conversations — question-back replies, not thumbs-up emoji.
Give questions a standing slot
The reason most accounts post questions badly is that they post them accidentally — a prompt tacked onto the end of a post that was really about something else. Question posts work better as their own scheduled format with a fixed day, the same way you'd treat a newsletter.
A weekly slot does three useful things: it trains your audience to expect a place where their opinion is wanted, it forces you to write the question in advance instead of improvising, and it gives you a clean comparison set. Fifteen questions posted on fifteen Wednesdays tell you far more than fifteen scattered across random posts.
Practically, block one recurring slot in your content calendar and fill it a month at a time. In SocialKit that means composing the question once and customising the wording per platform in the same post — the LinkedIn version gets a professional framing, the Threads version gets the blunt one — then letting it auto-publish on the day. If a prompt is genuinely evergreen, the recurring evergreen post workflow will re-run your strongest ones without a rewrite.
One boundary worth stating: SocialKit schedules the question, publishes it, and reports what happened — it has no unified inbox and no comment-moderation queue. Answering happens on the platform, by you, in the hour after it goes live. Scheduling makes the habit sustainable; the replying is still manual, and that's the part that decides whether the format works.
Read comment counts, not likes
Likes tell you a question was agreeable. Comments tell you it was answerable. When you review your posts in analytics, sort by comments and ignore everything else for this exercise.
Within about six weeks of a weekly slot, a pattern shows up: one or two formats consistently outperform the rest for your specific audience. Some audiences answer binary picks and nothing else. Some only show up for confessions. B2B audiences often ignore fun prompts entirely and answer situational-judgement questions at length, because those let them demonstrate competence.
Log four columns per question: format, platform, comment count, and whether any reply turned into a real thread. That fourth column matters more than the third. Nine comments that produced three genuine back-and-forths beat forty one-word answers, even though the engagement rate looks worse.
Start here
A two-week test that will tell you more than another month of guessing:
- Pick one slot. One platform, one day, same time each week.
- Write four questions now, one per week, using four different formats from the table. Don't improvise them later.
- Pull all four levers on each one. Low effort, real stakes, narrow frame, and your own answer ready to post as the first comment.
- Rewrite the ask line properly. The question itself is a call to action, and the same CTA formulas that make a save request land make a question answerable.
- Block twenty minutes after each one publishes to reply — question-back, not thanks.
- Log format, comments, and threads for each of the four.
- After four weeks, keep the two formats that produced threads and drop the rest. Then vary the topic, not the format.
The accounts with genuinely alive comment sections are rarely asking more questions than everyone else. They're asking narrower ones, answering first, and repeating the two formats that work instead of rotating through fifteen forever.