Why Dan Martell's best Reels start with a question

His latest 200 Reels were pulled and transcribed, and an AI model labelled the 186 real talking scripts on eight things: hook, structure, opening, evidence, emotion, advice, topic and call to action. Then the labels were compared against his plays ^[1]. The model read the words only, never the view counts ^[2]. In his best-performing fifth of Reels, with a top Reel of 2.8M plays, the most common labels are a curiosity hook (a question or an open loop), a comparison structure, a claim as the first line, reasoning as the proof, and no call to action ^[1].

Here the common labels and the medians mostly agree. Curiosity is his best hook type per Reel, and comparison is his second-best structure, behind short explanations on a small group of 9 ^[1]. So with Martell you can copy what he does most and what works best at the same time.

Five findings sit behind that, each with the figure and what it means for your next Reel ^[1]:

Each finding is one part of the same Reel. Someone's question opens it, one rule or a comparison answers it, an example proves it, the topic lands on the viewer's money or routine, and it ends on the lesson with no ask. Across the other creators decoded the same way, numbers, experiments and comparisons beat "here's my story" too, unless the story carries real proof ^[3].

Two cautions. Several groups are small, like 8 mistake hooks, 7 money Reels and 9 explanations ^[1]. Every figure is a median from his own account, and nobody tested the gaps against chance ^[2], so treat the shape as his best bet and check it on your own plays.

So build your next Reel on a question a client asked you. For a freelancer, that's "How do I tell my clients I'm putting my rates up?", answered with one rule and an example from past client work, names left out.

His top two Reels open with a question from someone in front of him: "I have 500 bucks, what do I do with it?" and "When do you know it's time to leave your 9 to 5?" ^[1]

Letting a real person ask the question makes the answer feel earned instead of preached ^[1], because the viewer hears their own worry in someone else's voice and then watches it get answered.

Film the question being asked, by a friend, a client or a comment on screen, then answer it in under 30 seconds ^[1]. You already hear these questions every week:

• A real estate agent hears "Should I sell before I buy?"
• A career coach hears "When do you know it's time to quit?", sometimes twice in one week.
• A freelancer hears "How do I tell a client I'm putting my rate up?"
• A family lawyer hears "Do we need a will if we're not married?" from almost every couple who buys a home together.

Write down the exact words, typos and all, because the client's phrasing is the hook. A comment screenshot on screen counts as someone asking, as long as you blur the name. The content ideas guide has more ways to collect questions like these.

When it doesn't fit: the gap between questions and claims is modest, 200K against 174K across 45 Reels, and the bigger gap between curiosity and mistake hooks rests on only 8 mistake hooks ^[1]. Never make a question up and pin it on a client, either. If nobody asks you much yet, use the question you had yourself when you started.

When the question arrived as a comment, Postmosis's post writer can start from it: its idea source Reply to a comment on your post opens the draft by quoting the comment, then answers it, written for what you do. Once your accounts are connected, schedule the post for the day you planned.

Here's his whole shape as one Reel you can film this week, built on one question a client asked you:

1. Write down a question a client asked you this month, in their words, and open on it. His curiosity hooks reach 228K median plays, against his 173.9K median ^[1].
2. Ask their goal before you answer, the second job in his top Reel ^[1]. It turns a general tip into advice for one person.
3. Answer with one rule, or as a comparison: what most people do, what you do instead, and why. His comparisons reach 253K, and his stories 95K ^[1].
4. Prove it with a client's situation, names taken out. His example Reels reach 202.4K, against 165.3K for Reels backed by his own experience ^[1].
5. End on the lesson, as 158 of his 186 Reels do, once your bio has one next step for the people who want more ^[1].

Here are the five steps filled in by a freelance copywriter, with the question on screen and the answer to camera:

• The question is the client's own words, shown on screen, so the viewer hears their worry in someone else's voice.
• "What's the goal?" is his second job, and it turns the advice from general to personal.
• The rule is a comparison, a paragraph of apology against two lines, which is his strongest large structure.
• The proof is a client example with the two who left kept in. A loss told straight is what makes the seven believable.
• It ends on something to do tonight, and nothing is asked for.

One question fills your week. Keep a note on your phone and add every question you hear on calls, in DMs and in comments, word for word, so you rarely run out. On Sunday, pick the one you heard most and film three takes in one sitting: the question and your rule for Monday, the comparison for Wednesday, and the client example for Friday. Next Sunday, check which of the three brought messages as well as plays, and pick the next question from the same kind of client.

When it doesn't fit: some questions can't be answered for everyone, like "Can my landlord keep my deposit?" put to a lawyer. Answer with what to gather or ask, like the lease and dated photos, and say plainly that their own case needs its own look.

Postmosis edits the clip you filmed on your phone in the browser: it cuts the pauses, retakes and filler words, adds captions in 6 styles and exports 9:16, 4:5, 1:1 and 16:9 in one run. See the video editor

His top Reel has every part of that shape. At 2.8M plays, it starts when a guy tells him he has 500 bucks and asks what to do with it. Dan asks his goal, then gives one rule: pick one thing, because people who win stay with one thing for a long time ^[1]. He names it, FOCUS, for follow one course until successful. He reframes losing the money as learning, and closes with a surfing picture: falling off the board isn't failing, it's learning how not to stand ^[1].

The model labelled it a curiosity hook with a problem-and-fix structure and principle-level advice, with no call to action ^[1]. Here are its eight jobs taken apart line by line, with his Reel beside a marketing consultant's version written for this page:

The second job is the one most people skip. Asking the goal before you answer turns a general tip into advice for one person, and the viewer who shares that goal leans in. Film the question first, read by a friend or shown as a comment on screen, then cut to your answer. Read straight through, the consultant's answer runs well under half a minute.

These four templates come from his patterns ^[1], rewritten so the brackets ask about your clients. Fill each with your own topic, proof and wording, then time it with the script length calculator.

[A client asks:] "I have [a small amount of time, money or help]. What should I do with it?" What's the goal? [Their answer]. Then [one rule]. The people who get there all [what they have in common]. I call it [a short name for the rule]. And if it doesn't work, [reframe the setback]. It's like [a simple picture from everyday life].

[The thing everyone in your field recommends]? Skip it. I know. [Why it misleads people]. Instead, [the better option]. Here's why: [one reason].

[Tool A], [what it costs], because you need [what it does for your clients' work]. [Tool B], [what it costs], only if [situation]. Skip [Tool C] until [the point where it starts to pay for itself].

I don't care about [the thing everyone in your clients' position worries about]. I care about [what you watch instead]. Because [reason]. Here's what I do: [habit].

Each keeps his order: the question or the surprise first, one rule in the middle and an ending on the lesson. Put your client example after the rule. The video script template has a blank page laid out the same way.

His four scripts don't fit every trade equally well. Here's where to start for each kind of work, a question to open on, and what to keep out of the Reel:

Whichever row is yours, the first line is still a question a client really asked you.

Here's "Skip this, do this" filled in by a real estate agent who's tired of quiet Sunday open houses, and read aloud it runs under half a minute.

• "Skip them. I know." is his move from the Rich Dad Poor Dad Reel, and it admits the viewer has heard the opposite ^[1].
• The reason is one sentence a seller can repeat to their partner over dinner.
• The proof is a count from the agent's own listings, small and checkable, with no prices in it.

When it doesn't fit: three listings is a small count, and a busy month can make any method look good. Say how many it rests on, as this script does. If sellers in your area expect open houses, test the swap on one listing and tell the seller why first.

A freelancer could use his tool-decision shape the same week:

After the question comes the answer, and on his account the shape of the answer matters most. His 15 Reels built around a story from his own life have the lowest median of any structure. None of them reach his top fifth, and 7 sit in the bottom fifth ^[1]. Comparisons, this versus that, do more than 2.5 times better, and short explanations of one idea are strong too, at 331K across 9 Reels ^[1].

It's the same lesson as Hormozi and Shelby Sapp: on these accounts, the take beats the backstory ^[1]. The Hormozi breakdown and the Shelby Sapp breakdown show the same pattern in their own numbers.

Turn your next story into a comparison: what most people do, what you do instead, and why ^[1].

• "Most freelancers" against "I" is the comparison, so the viewer checks which side they're on.
• The lost client stays in the Reel as the reason, which is where a story earns its place for him.

When it doesn't fit: these are medians inside one account, and explanation is a small group of 9 Reels ^[1]. On Kallaway's far smaller account, Reels backed by his own experience reach about 1.9 times his reasoning-only Reels ^[4]. If your viewers don't know you yet, a one-line story after the question can be your proof.

Under the answer sits the proof, and the kind of proof matters too. Reels that make the point with an example reach 202.4K, while Reels backed by his own experience reach 165.3K, under his median ^[1].

For a professional, an example is usually a client's situation with the names taken out. Here's one agent's point made both ways:

The example hands the viewer a decision they can picture making themselves, and since nobody in it is named, the couple would recognise the advice before they recognised themselves.

When it doesn't fit: an example has to be yours to tell. If you're a lawyer, or a consultant under a confidentiality agreement, change the case until nobody involved would know it, or use your own mistake. Numbers as proof cover only 4 of his Reels, so his data says little about them ^[1].

What the question is about counts as well. Business and mindset make up 97 of his 186 talking Reels, and both sit at a 135K median, well under his account median. They fill 26 of the 37 Reels in his bottom fifth and only 13 of the 38 in his top fifth ^[1].

Personal money, daily habits and health travel much further for him: 10 of his top 38 are life topics, against 1 of his bottom 37. His biggest Reel is a guy asking what to do with 500 bucks ^[1].

Take a business lesson and aim it at someone's personal money or routine ^[1]. A consultant's "how to run a planning day" becomes "the one meeting I'd cut from your week." A coach's "how I structure a programme" becomes "what I'd do with the first hour of your morning." The lesson stays yours, and it lands on the viewer's own day.

When it doesn't fit: money has 7 Reels and health 4, so the exact gap is a direction, though business sitting at the bottom across 57 Reels is solid ^[1]. Keep the routine tied to what you sell. A lawyer's Reel about morning habits reaches people who'll never need a lawyer, so aim it at a routine your clients have, like the paperwork they put off.

One topic is growing fast for him. One in five of his recent Reels is about AI: which tools to pay for, ChatGPT versus Claude, and red and green flags when using AI. His 38 AI and tech Reels reach a 212K median, about 20% over his account median, and 11 of them are in his top fifth ^[1].

For a business creator, AI is the topic that keeps him on brand and still reaches people who don't care about scaling a company ^[1]. It's likely the same for you, since most of your clients are working out which AI tools to trust in their own jobs.

Make one Reel a week on a specific AI tool decision your audience faces: which one, what it costs, what to skip ^[1]. Keep the tool inside their work: a real estate agent can compare two tools for writing listing descriptions, and a consultant can say which note-taker belongs on a client call.

When it doesn't fit: AI content is riding a wave, and the gap may shrink as the topic gets crowded ^[1]. Only review a tool you've used on real client work, and say what you used it for, because a tool Reel goes out of date fast.

The Reel ends on the lesson. 158 of his 186 Reels have no call to action ^[1]. The 28 that ask viewers to comment, follow or buy do slightly worse, at 143K against 178K, and only 3 of them reach his top fifth while 7 sit in the bottom fifth ^[1]. Like Hormozi, he lets the content earn the follow and sells elsewhere ^[1].

Pick each ending by what you want from that Reel, reach or leads ^[3]. Here are three for the copywriter's rate Reel:

End most Reels on the lesson, the way he does, and keep one clear next step in your bio, like a booking page. Put the plain ask on a few, tie it to what the Reel taught, and compare their plays and messages with the rest.

When it doesn't fit: Reels with a call to action may also be the more promotional ones, so the ask itself may not be what holds them back ^[1]. If you need enquiries this month, a comment ask tied to the lesson is a fair trade on some Reels.

Martell's shape starts with a real question, so start your AI draft there too. Collect five questions clients asked you this month, then paste this into ChatGPT or Claude:

I'm a [what you do] and my clients are [who they are]. Here are five questions clients asked me this month, in their own words: [paste them].
1. Pick the two that the most people like my clients would also ask. Say why in one line each.
2. For the best one, write a script under 30 seconds in this order: the question as they asked it, me asking their goal, their answer, one rule, a line on why it works, a short name for the rule, a reframe if it goes wrong, a simple picture from everyday life.
3. Write a second version as a comparison: what most people do, what I do instead, and why.
4. No call to action. Don't invent results, numbers or client stories. Where proof belongs, write [YOUR PROOF].

Keep the client's question word for word in the final script. It's the line that makes the rest feel earned. Then swap [YOUR PROOF] for a client example you're allowed to tell.

In Postmosis, Write shows you hooks on named formulas first, with one marked as recommended, then three versions of the post or short-video script for the work you do. See the post writer

Martell's numbers come from an account where the median Reel gets about 174K plays ^[1]. Yours may get a few hundred, from people who hire someone like you once every few years. Here's where his pattern breaks on an account like that, and what to do instead:

• The question needs a real person, since his best Reels have someone in front of him asking ^[1]. If you film alone, read the question off a client's message on screen, blur the name, then answer to camera.
• Skipping the ask works because he sells elsewhere ^[1]. Before you drop your call to action, make sure your bio and its link point at one clear next step.
• The AI lane is getting crowded, and the gap may shrink ^[1]. Stay specific to your clients' work, since "which AI tool should a property agent use" has far fewer people answering it than "best AI tools".
• Plays aren't your score. On a small account, one message from the right client is worth more than a jump in plays, so count enquiries beside plays when you judge a Reel.
• A question Reel that flops still tells you something, so note the question, the plays and the messages, and try a sharper question from the same client next time.

Instagram shows each eligible Reel to a small audience first, regardless of whether they follow you, and only the top performers go wider ^[5]. That first group has never met you, so a question they recognise does more for you than a big name could.

His hook types back up the question opening. His curiosity hooks lead at 31% over his median, and hooks built on a mistake ("you're doing this wrong") trail at 47% under it ^[1].

The opening sentence tells the same story. Question openings beat plain claims, 200K against 174K across 45 Reels ^[1].

The mistake hook is worth a second look if you use it a lot. "You're doing your marketing wrong" is an easy first line for any professional to write, and it's his weakest type. A question in your client's own words does the same job without telling them off. Here's one freelancer's Reel opened three ways:

On his numbers those three sit at +31%, +5% and -47% against his median ^[1]. The curiosity one is also a question a client could have asked you, which is the shape his top two Reels share.

Read his top 12 for how they open, since that's where the question shows up. Here they are by plays, with the first words of each and the labels the model gave it ^[1]:

Five of the twelve open with a question, and three open by rejecting something people expect, like "I don't care about my kids grades" ^[1]. Most are about money, habits, careers or AI tools, and very few are about running a company.

Likes per 1,000 plays is a rough read on agreement. The "how would you become successful at my age" question and the bedtime alarm sit near 63 and 49, while the sales Reel sits at 15 ^[1].

For your next three Reels, borrow one shape each. Here they are for a career coach:

Dan Martell's site describes him as an author, entrepreneur and coach, and lists the companies and projects he runs: Martell Ventures, a network of companies offering software, AI development and recruiting services to entrepreneurs, plus the Kings Club, free events for 13 to 22 year olds, and the Martell Legacy Foundation ^[6].

His books page lists two titles: Buy Back Your Time, about scaling a business without burning out, and Software as a Science, a guide for B2B SaaS founders ^[7]. One of his top Reels opens by telling viewers to skip Rich Dad Poor Dad by Robert Kiyosaki ^[1].

On Instagram he posts business advice, and his median Reel gets about 174K plays ^[1]. A coach answering questions on camera is your job on a bigger stage. You hear the same kind of questions every week, on calls, in DMs and in comments, which is why his shape carries over, as long as the question is a real one from your clients and the proof comes from your work.

His other three label groups, each against his account median of 173.9K plays ^[1]. Rows marked small sample have fewer than five Reels, so read them as noise.

The steady reads for your planning: comparisons and explanations above stories, examples above his own experience, and money, habits and AI above business and mindset ^[1].

To use the tables, find the row nearest what you already post. A consultant's posts about running a firm sit in his business row, 22% under his median, and a coach's morning-routine posts sit closest to habits, 26% over ^[1]. Either way the cue is the same: keep the lesson yours and aim it at the viewer's own day.

His numbers are medians inside one account, and nobody tested whether chance could explain the gaps. They're patterns worth trying on your own content, and they don't prove what causes reach ^[2]. The figures worth having are your own. Take your last 30 Reels with their plays and label them with this prompt ^[3]:

For each video below (transcript and plays), label:
hook type (curiosity / result / direct / contradiction / recognition / story / mistake)
opening (question / instruction / claim / story)
structure (explanation / comparison / Q&A / steps / opinion / problem and fix / story)
evidence (example / reasoning / personal / numbers / none)
topic, and call to action (none / comment / follow / link).

Then for each label, give the count and the median plays per value.
Only report a difference when each group has at least 5 videos.
End with three rules for my account, each backed by a number from this data.

[paste one row per video: plays, then the transcript]

1. Use at least 30 videos, compare medians and ignore any group under 5 videos ^[3]. One big Reel can drag an average far enough up to fool you.
2. Write down 3 rules for your account, each backed by a number, and use one in each of your next 3 scripts ^[3].
3. Run the question shape for 20 posts before you judge it ^[4], with everything else the same as your usual Reels.
4. Post some as trial reels, which go to people who don't follow you first, and look for each one's numbers in the reels viewer about 24 hours after you share it ^[8].
5. Make each test Reel new, since a trial reel may get limited reach if Instagram finds you've already shared the same content ^[9].

The reel insights guide shows where each number lives in the app.

The data was read on October 3, 2026, from 200 Reels: 186 with speech were decoded, 6 without speech were skipped and 8 music or vlog Reels without a script were set aside. Plays and likes keep changing after a read like this, so treat every figure as a snapshot of that week ^[1].

An AI model labelled each script on eight dimensions from the words alone, and median plays were compared across the labels ^[2]. A median is the middle Reel of each group. One huge Reel can't drag a group up, and groups with only a handful of Reels carry a small-sample flag in the tables above ^[1].

It's an independent analysis of public Reels, not affiliated with Dan Martell, and it quotes only short opening lines ^[2].