Product launch video: 7 steps from 293 X launches

I analysed 293 product launch videos posted on X. Together, they had 380,875,388 views.

The 178 launches that reached at least 247,000 views had a median near 1 million. The other 115 were real software launches with a median of 168 views.

The two groups used different lengths, formats, captions, and posting times. Inside the high-reach group, none of those video choices predicted more engagement per viewer. Distribution still decided how far a launch travelled.

This product launch video playbook gives you a practical script, recording plan, export format, caption, posting window, and measurement check based on that sample.

Use Distribb to turn your launch into search pages, internal links, and outreach that keep working after launch day.

What the 293 product launch videos showed

I pulled the view count, likes, length, resolution, and posting time from every post. I also transcribed all 293 videos so I could measure what each script said and when it said it.

The launches people saw ran for a median of 83 seconds, often used a voiceover, showed a working demo, and were usually exported in 16:9 landscape. The low-reach launches ran for a median of 32 seconds, and 48% were under 30 seconds.

Timing was different too. Among the launches people saw, 82% went live from Tuesday through Thursday and 84% were posted between 14:00 and 18:00 UTC. Friday accounted for 28% of the low-reach group and 6% of the high-reach group.

Those differences show what established launches tend to look like. They do not prove that a 90-second landscape video or a Tuesday post caused the reach.

The same warning applies to the script. Inside the high-reach group, video length, spoken calls to action, price mentions, feature count, and problem-led openings did not predict more likes per viewer.

I also opened the visible quote reposts for ten launches. Five had at least one repost carrying X's Paid partnership label. Across 51 visible quote reposts, 15 were labelled as paid.

Step 1: write the script in five blocks

I coded all 125 narrated launches in the high-reach group.

Seventy-four percent named a problem the viewer already had. Sixty-seven percent named that problem in the first ten seconds. Seventy-four percent compared the current way of doing the job with the product's way, and 74% showed concrete proof. Among the launches that showed proof, 43% used a number.

The most common script shapes were problem then product at 30%, vision then product at 22%, and story then product at 17%. Straight feature lists made up 4%.

Write the first draft in five blocks:

1. Say the product name and what it is.
2. Name the problem your viewer already has.
3. Show the product doing that job and explain how the job gets done today.
4. Give proof, preferably a checked number.
5. End with one action, one practical vision, or a clean stop.

Use this shape because viewers recognise it as a product launch. The study did not show that any one block caused higher engagement.

Step 2: cut the script to about 219 words

The 125 narrated, high-reach launches used a median of 219 words over roughly 96 seconds. That works out to 143 words per minute, which is slower than normal conversation.

The median spoken-word budget looked like this:

• About 7 words by 3 seconds
• About 24 words by 10 seconds
• About 77 words by 30 seconds
• About 146 words by 60 seconds
• About 219 words by the end

Seven words is one short sentence. By ten seconds, the viewer has heard about two sentences. Use that space to name the product or the problem in plain words.

Read the whole script out loud with a timer. Cut any line that forces you to rush.

I expected problem-first openings to beat product-first openings before the analysis. They did not. Once the low-reach launches were included, the difference disappeared.

Step 3: record a voiceover over a working demo

Fifty-one percent of the high-reach launches used voiceover without putting a person on camera. Twenty-four percent showed a founder or another person, and 25% used several people.

Demo walkthroughs appeared in 39% of the high-reach launches and 19% of the low-reach group.

The simplest recording plan is a real screen capture with a separate voiceover. Record the product finishing one job. A sequence of menus and hover states does not show the result.

Keep the voiceover separate from the screen capture. You can replace one weak line without recording the whole demo again, and you can cut the screen recording to match the words.

Motion graphics were a separate format in the sample. I checked three frames from 105 launches. Motion graphic videos made up 21% of that group, had a median length of 44 seconds, and used large animated words in 82% of cases. Sixty-four percent had no voiceover.

Step 4: export in 16:9 landscape at about 90 seconds

Eighty-nine percent of the high-reach launches used 16:9 landscape. One of the 178 was vertical.

The median video was 83 seconds across the whole high-reach group and 96 seconds among narrated videos. Length still did not predict more engagement after the view count was taken into account.

Use 1920 by 1080 or a larger 16:9 export. Aim for 80 to 100 seconds when the script needs that time. Avoid forcing a useful demo into 15 seconds because 48% of the low-reach group was under 30 seconds, compared with 10% of the high-reach group.

Step 5: write a 44-word caption

Both groups used a median caption length of 44 words. Most captions were between 35 and 50 words.

Thirty-four percent of the high-reach launches started with the word Introducing, compared with 9% of the low-reach group. Emoji appeared in 12% of the high-reach posts and 50% of the low-reach posts.

Use a plain caption:

Introducing Distribb. It is an AI SEO platform that turns keyword research into pages, internal links, and outreach. It is for founders and SEO teams that want one place to plan and publish the work.

Put the product link in the post. The numbers describe the sample, so treat this caption as a clean convention rather than a reach guarantee.

Step 6: publish Tuesday through Thursday during US business hours

Among the high-reach launches, 82% were posted on Tuesday, Wednesday, or Thursday. Saturday and Sunday together accounted for 1%.

The busiest window was 14:00 to 18:00 UTC. The busiest single hour was 16:00 UTC, which is 9 a.m. in San Francisco, noon in New York, and 6 p.m. in Madrid during the offset used in the source study.

A company with a launch team may post during business hours for reasons that have nothing to do with the X algorithm. Moving a launch away from Friday costs nothing, so this is the easiest pattern to copy.

The distribution work happens here too. Prepare the people, customers, partners, and paid creators who will share the launch. Five of the ten launches checked closely used at least one paid quote repost.

Step 7: compare engagement with launches of a similar size

Raw likes make large launches look stronger than they are.

The smallest fifth of the 178 high-reach launches had a median of 344,417 views and 3.02 likes per 1,000 viewers. The middle fifth had a median of 1,023,927 views and 1.72 likes per 1,000 viewers. The largest fifth had a median of 4,925,424 views and 0.72 likes per 1,000 viewers.

A launch at 4.9 million views got about one quarter of the likes per viewer of a launch at 344,000 views. A wider audience contains more people who were never likely to buy.

Record views, likes per 1,000 viewers, clicks, signups, and sales. Compare the result with launches that reached a similar number of people. Use the business result to judge the launch, because a view does not tell you whether the viewer became a customer.

Frequently asked questions

How long should a product launch video be on X?

Aim for roughly 80 to 100 seconds when the product needs a real demo. The high-reach launches ran for a median of 83 seconds, and narrated launches ran for a median of 96 seconds. Length did not predict higher engagement inside that group, so keep the video only as long as the useful script requires.

What should I say in the first ten seconds of a product launch video?

Use about 24 words to name the product or the problem it solves. The narrated launches used a median of 7 words by three seconds and 24 by ten seconds. Product-first and problem-first openings both appeared in successful launches, and neither won in the full comparison.

Should I show my face in a product launch video?

You do not need to appear on camera. Fifty-one percent of the high-reach launches used voiceover with no person on screen, while 24% showed one person. A clear product demo with a separate voiceover is a common and practical format.

What video format should I use for an X product launch?

Use 16:9 landscape at 1920 by 1080 or higher. Eighty-nine percent of the high-reach launches were landscape, and one of 178 was vertical. The sample describes what established X launches used and does not prove that landscape caused the reach.

When should I publish a product launch video on X?

Tuesday through Thursday between 14:00 and 18:00 UTC matches the most common window in the study. Eighty-two percent of the high-reach launches used those days, and 84% used that four-hour window. A launch team's work schedule may explain part of the pattern, so treat the timing as a low-cost convention rather than an algorithm trick.

How do I measure whether a product launch video worked?

Compare likes per 1,000 viewers and business results with launches that reached a similar audience size. Engagement per viewer fell from 3.02 likes per 1,000 in the smallest high-reach group to 0.72 in the largest. Track clicks, signups, and sales because view count alone cannot show whether the launch brought customers.

Which part of your next launch needs the most work: the script, the demo, or the distribution list?

Try Distribb free and turn the launch into search pages, internal links, and outreach that keep bringing people back.