Per our own RADAR data, the median of 583 tracked startup launches reached 23,429 views, and only 4.1% cleared a million.
FORKOFF RADAR · Research · Updated 2026-09-24 · n=967 launches tracked · 25 launch videos audited
The State of Startup Launch Videos 2026
FORKOFF RADAR tracks 967 startup launches on X. The typical one reached 23,429 views. Only 4.1% cleared a million. The launches that broke into the top 10 percent opened differently, launched on different days, and shipped a video of about a minute and a half.
583 launches with measured reach, 25 launch videos probed for length and format, 30 hand-audited case studies. Board read 2026-09-24.
Planning a launch? See how we build product launch videos around these numbers.
- 01The state of launch videos, in one answer
- 02How far 967 tracked launches reach
- 03Who launches: AI and SaaS, hardware, other
- 04What the top 10 percent do differently
- 05Launch video length and format
- 06Build a launch on the top-decile pattern
- 07Two layers: tracked field and audited case studies
- 08The audited case studies by vertical
- 09How far launch videos actually reach
- 10Organic vs amplified reach
- 11The wider market, sourced
- 12Method and dataset (CSV download)
- 13Adjacent RADAR reading
- 14Frequently asked questions
The state of launch videos, in one answer
Most startup launches on X stay small. Across 583 launches FORKOFF RADAR tracked with measured reach, the median reached 23,429 views, 22.6% cleared 100,000 and 4.1% cleared a million. The top 10 percent (at or above 351,876 views) opened with "Introducing", "Meet" or a launch announcement 36% of the time, against 16% for everyone else, and 71% of them launched Tuesday to Thursday. The launch videos RADAR audited run a median of 93s, and 24 of 25 are landscape. On 30 audited case studies, 67% carried a distribution-amplified reach signature and 33% read organic.
This report reads how launch reach was built, never whether anyone was honest. A high views-to-likes ratio is a read on mechanics, and paying for or coordinating distribution is a legal, ordinary launch tactic.
How far 967 tracked startup launches reach
RADAR tracks 967 launch posts on X, 851 of them from 2026. 583 have a measured peak view count so far. The middle half landed between 8,798 and 82,524 views. 24 launches cleared a million. A launch that reaches six figures is already in the top quarter.
| Measure | Views |
|---|---|
| 25th percentile | 8,798 |
| Median | 23,429 |
| 75th percentile | 82,524 |
| Top-decile threshold (90th) | 351,876 |
| Top-decile median | 832,187 |
| Share over 100K | 22.6% |
| Share over 1M | 4.1% |
AI and SaaS crowd the field, hardware is rare
AI and SaaS launches are 58% of the tracked field, and only 8.9% of them reach the top 10 percent. More launches in one category means less attention for each. Hardware launches are rare, 9 with measured reach, and 2 of them made the top decile. That is too few to call a pattern, so it is shown here and not claimed.
| Segment / category | Launches | Median views | Reached the top decile |
|---|---|---|---|
| AI and SaaS | 338 | 21,566 | 30 (8.9%) |
| Other | 236 | 25,950 | 27 (11.4%) |
| Hardware | 9 | 27,135 | 2 (22.2%) |
| AI | 318 | 22,172 | 28 (8.8%) |
| Other | 122 | 26,541 | 17 (13.9%) |
| Crypto | 54 | 44,029 | 5 (9.3%) |
| B2B | 30 | 8,915 | 1 (3.3%) |
| SaaS | 20 | 18,154 | 2 (10.0%) |
| Media | 18 | 14,653 | 1 (5.6%) |
| Consumer | 12 | 40,108 | 3 (25.0%) |
| Hardware | 9 | 27,135 | 2 (22.2%) |
Segment rows in bold group the RADAR categories: AI and SaaS = AI + SaaS; Other = every category except AI, SaaS and Hardware. n=583.
Launching in one of these segments? The playbook differs. What FORKOFF runs for AI startups covers standing out in the crowded 58%. Hardware and robotics launch videos covers the rarer launches, where the product has to be shown working on camera.
What the top 10 percent of launches do differently
59 launches reached 351,876 views or more. Two things separate them from the other 524. The first is the opening line. A post that opens with "Introducing" or "Meet" reached the top decile 19.7% of the time. An announcement opener reached it 20.0% of the time, and a plain product statement 7.9%. The second is the day: Tuesday launches broke out most often. Post length, a link in the post and a morning slot made no difference.
Top-decile rate by opening line
Top-decile rate by launch day (Pacific)
| Trait | Top decile (n=59) | Everyone else (n=524) | Read |
|---|---|---|---|
| Opens with "Introducing", "Meet" or an announcement | 36% | 16% | Separates |
| Launched Tuesday to Thursday (Pacific) | 71% | 58% | Separates |
| AI or SaaS product | 51% | 59% | Slightly lower in the top decile |
| Posted 6 to 11 am Pacific | 53% | 54% | No difference |
| Link in the launch post | 32% | 31% | No difference |
| Median launch post length | 199 chars | 200 chars | No difference |
These are correlations, not causes. Large labs and funded companies tend to open with "Introducing" and launch mid-week, so part of the gap is who launches that way, not only the wording.
Launch video length and format
RADAR measured the video file of 25 launches that broke out, with a median of 2,613,305 views. They run a median of 93s, with the middle half between 60s and 106s. 24 of 25 are landscape and every one has an audio track. The short ones earned the most engagement per view. The 4 videos under 45 seconds reached 0.5% of views, against 0.2% for 90 to 150 seconds.
| Length | Videos | Median views | Engagement per view | Median views per like |
|---|---|---|---|---|
| under 45s | 4 | 1,308,855 | 0.5% | 252 |
| 45 to 90s | 7 | 3,770,458 | 0.2% | 825 |
| 90 to 150s | 14 | 2,929,989 | 0.2% | 736 |
n=25 launch videos, probed for duration, width, height and audio. Engagement per view = (likes + reposts + replies + quotes) / views. Small buckets: read as direction.
Posting slot shows the same trade-off. The 8 videos posted on the hour drew more views (median 4,150,428), but at 1,441 views per like. The 17 posted off the hour ran 651. More reach per like is the pattern the RADAR method reads as distribution doing more of the work.
Build your launch on the top-decile pattern
- Open the post the way the top decile does: name the product in the first words ("Introducing", "Meet", or a launch announcement).
- Launch Tuesday to Thursday. 71% of breakout launches did.
- Ship a landscape video with sound, around 60s to 106s, and cut a shorter version for replies and quotes.
- Judge the result by views per like as well as views, so you know whether the audience earned the reach.
FORKOFF's product launch video service pairs the launch video with a 14-day distribution warm-up, planned against this data. There are dedicated versions for AI startups and hardware launches. RADAR is how we measure the launch afterwards.
Only 4.1% of 583 tracked launches cleared a million views. Bring your launch date and product, and we will map it against the 59 launches that reached the top decile.
Two layers: 967 tracked launches and 30 audited case studies
The field above answers what a normal launch reaches. It cannot say whether a big number was earned, because a view count alone does not show that. The sections below answer that on 30 hand-audited case studies. Each is a public post with public view, like, reply, quote and repost counts, deduplicated to one canonical launch post per product, and 18 carry a full forensic trace. Both layers ship as CSVs under CC BY 4.0, so a reader, an analyst or an answer engine can re-derive every figure.
Every launch is public
All 30 launches are public posts with public view, like, reply, and quote counts, pulled via our own X data pipeline. Nothing here is a private or unnamed data point.
The dataset ships
The full aggregate dataset downloads as a CSV under CC BY 4.0. The numbers on this page and the numbers in the file come from the same single source, so they cannot disagree.
The method is disclosed
A full forensic trace is verified on 18 of the 30 launches; the rest are read from the public views-to-likes ratio at lower confidence, and labeled as such.
The 30 audited case studies across 6 verticals
The tracked launches span 6 product verticals. AI leads on combined reach with 7 launches and 42.8 million views. The vertical is a neutral product-domain tag, never a verdict, and per-vertical cuts on a 30-launch set are directional rather than definitive.
| Vertical | Launches | Combined views | Median views per like | Read organic |
|---|---|---|---|---|
| AI | 7 | 42,809,039 | 568 | 3 of 7 |
| Marketing and Sales | 9 | 27,551,311 | 821 | 1 of 9 |
| Consumer and Health | 4 | 15,141,230 | 892 | 1 of 4 |
| Dev tools | 5 | 13,141,341 | 754 | 2 of 5 |
| Fintech | 3 | 7,972,752 | 724 | 1 of 3 |
| Crypto/Web3 | 2 | 3,802,959 | 1,420 | 2 of 2 |
| All verticals | 30 | 110,418,632 | 788 | 10 of 30 |
FORKOFF RADAR launch dataset, n=30, Oct 2025 to Jun 2026. Verticals are neutral product-domain tags. Counts recomputed live from the tracked set.
How far the audited breakout launches reached
The case studies were picked because they broke out, so they sit far above the field median of 23,429: the median audited launch reached about 3,098,892 views, and the distribution is still heavily skewed. Reach is grouped into public view-count bands below. These are aggregate counts of how many launches landed in each reach tier, a factual read on the ceiling a launch video hits, not a verdict on any of them.
2
10M and over views
7% of tracked launches
4
5M to 10M views
13% of tracked launches
19
1M to 5M views
63% of tracked launches
5
under 1M views
17% of tracked launches
What an organic launch looks like
The three biggest launches RADAR reads as organic, reach and engagement that grew together. Each links to its full reading.
Two in three launches ran ahead of their engagement
20 of the 30 tracked launches (67%) carried a distribution-amplified reach signature; 10 (33%) read organic. The organic cohort ran a median of 354 views per like, the amplified cohort 1,441, about 4.1 times higher. That gap is the whole finding: amplified reach runs further ahead of the likes than organic reach ever does.
The full band calibration, the distribution chart, and the reproduce-it-yourself steps live on the views-to-likes benchmark, the sibling study this split is measured on.
The market these launches sit inside, sourced
The first-party numbers above are X-native and measured on this tracked set. The figures below are named third-party benchmarks that frame the market a launch video competes in. They are directional context, each with its source and year, never pooled into this report's first-party numbers.
91%
of businesses use video as a marketing tool in 2026, back to an all-time high.
Wyzowl State of Video Marketing 2026, 202693%
of video marketers say video is an important part of their overall marketing strategy.
Wyzowl State of Video Marketing 2026, 202682%
of marketers say video marketing gives them a good return on investment.
Wyzowl State of Video Marketing 2026, 202683%
of video marketers say video has directly helped them increase sales.
Wyzowl State of Video Marketing 2026, 202682%
of video marketers say video has helped them increase website traffic.
Wyzowl State of Video Marketing 2026, 202685%
of people say they have been convinced to buy a product or service by watching a video.
Wyzowl State of Video Marketing 2026, 202696%
of people have watched an explainer video to learn more about a product or service.
Wyzowl State of Video Marketing 2026, 202684%
of consumers say they want to see more video content from brands in 2026.
Wyzowl State of Video Marketing 2026, 202689%
of consumers say a video's quality directly impacts how much they trust the brand behind it.
Wyzowl State of Video Marketing 2026, 202663%
of people say a short video is how they would most like to learn about a new product or service, well ahead of text.
Wyzowl State of Video Marketing 2026, 202686%
of ad buyers now use or plan to use generative AI to build video creative.
IAB 2025 Digital Video Ad Spend and Strategy Report, 20250.12%
is X's per-follower engagement floor, the lowest of the major platforms, so a launch video's raw view count cannot be read at face value.
Socialinsider Social Media Benchmarks 2026, 2026External figures are reported as published by their named sources, read live against the source page on the last-updated date. Put together they frame the stakes: video is now used by almost every business and converts, generative AI is standard in production, and X carries the lowest engagement floor of the major platforms, which is exactly why a launch video's raw view count needs the views-to-likes read to be understood.
How the report was measured
The tracked field is every launch on the public RADAR board, all 967 read on 2026-09-24. Reach is the peak view count RADAR recorded on each launch post, measured so far on 583. Opening lines are classified by pattern from the post text; a hand check of 24 random posts found the named patterns right 17 of 18 times. Launch times are converted to Pacific. The 25 launch videos were probed for duration, orientation and audio, and the video layer leaves out 5 audited posts that are a funding post, a grant or carry no video. The 30 case studies are public product-launch posts across Oct 2025 to Jun 2026, totaling 110.4 million views, with a full forensic trace on 18.
Tracked field
967 launches
583 with measured reach, read 2026-09-24
Launch videos
25 probed
median 93s, 0 probe failures
Case studies
30 launches
Oct 2025 to Jun 2026, 18 of 30 verified trace
The views-to-likes ratio is the load-bearing signal because it is a pure function of two public numbers, reproducible by anyone from the public post. The organic ceiling near 500 views per like and the calibration bands come from the published RADAR methodology. A verified trace adds the amplification-wave shape and the posting-time fingerprint as corroboration, but the ratio anchors every read.
Caveats, stated plainly. The top-decile traits are correlations, not causes. Hardware has only 9 launches with reach and the video length buckets hold 4, 7, 14 videos, so those cuts are directional. A view count includes replays and non-human views; the views-to-likes read on the case studies is the check on that. The case-study sample is 30launches, so per-vertical cuts are directional, not definitive. Metrics are point-in-time and trace to a public pull; bot purges and late engagement shift the numbers, which is why the set is re-checked and dated. Reads without a full trace are never rendered as high-confidence bought verdicts. And amplified distribution is a legal, common, often-sound way to launch at scale, so a high ratio is a read on how reach was built, not a claim about anyone's honesty.
The RADAR aggregates CSV carries the reach bands and percentiles, the category, opening-line and weekday breakdowns, and the launch-video length and format figures. The case-study CSV carries the overview, the per-vertical breakdown, the ratio percentiles and the anonymized per-launch ratio distribution. Launch-level rows are anonymized by design: the files publish distributions, not a per-named verdict.
Adjacent RADAR reading
- Launch Radar hub , the tracked launches ranked by reach, each with its public metrics and views-to-likes ratio.
- Best product launch videos, ranked , the tracked launch videos ordered by views, each with the same reach read applied.
- Views-to-Likes Benchmark 2026 , the sibling study with the full band calibration this report's organic vs amplified split is measured on.
- RADAR methodology , how RADAR reads whether a launch's reach was earned or amplified, from three public signals, with the full calibration.
- Product Launch Video , the managed launch service this report comes out of, priced on audited reach rather than a production fee.
State of launch videos, common questions
Frequently asked questions
What is the state of launch videos in 2026?
FORKOFF RADAR tracks 967 startup launches on X, 851 of them from 2026. Of the 583 with measured reach, the median launch reached 23,429 views, 22.6% cleared 100,000 and only 4.1% cleared a million. The launches that reach the top 10 percent open their post with "Introducing", "Meet" or a launch announcement far more often, and launch Tuesday to Thursday more often. On the 30 audited case studies, 67 percent carried a distribution-amplified reach signature and 33 percent read organic.
How many views does a startup launch video get on X?
Across 583 tracked startup launches with measured reach, the median was 23,429 views and the middle half sat between 8,798 and 82,524. Reaching the top 10 percent took 351,876 views. 22.6% of launches cleared 100,000 views and 4.1% (24 launches) cleared a million. The 30 audited launch videos in the case-study layer are a different population, selected because they broke out, with a median of 2,613,305 views.
How long should a product launch video be?
The 25 launch videos RADAR audited run a median of 93s, with the middle half between 60s and 106s. 24 of 25 are landscape and all carry an audio track. The 4 videos under 45 seconds earned the highest engagement rate, 0.5% of views against 0.2% for 90 to 150 seconds, though with fewer views. That cut holds only 25 videos, so read it as a direction, not a rule.
What do the top 10 percent of startup launches do differently?
The top decile (59 launches at or above 351,876 views) opens with "Introducing", "Meet" or a launch announcement 36% of the time, against 16% for everyone else. 71% launched Tuesday to Thursday, against 58%. Post length, a link in the post, and a 6 to 11 am posting slot made no difference. These are correlations: large companies tend to post that way too.
What day should a startup launch on X?
In RADAR's tracked set, 14.9% of Tuesday launches reached the top 10 percent, the best of any day, against 6.5% on Friday. Tuesday is also the most crowded day (141 launches). The typical result follows a different order: Friday has the highest median reach (28,570 views against 23,470 on Tuesday). So the day shifts the odds of a breakout more than it shifts the median.
Why two datasets, a tracked field and 30 case studies?
They answer different questions. The tracked field (967 launches, public board data) shows what a normal launch reaches and what the top decile has in common. The 30 case studies are hand-audited launches with full public engagement counts and a forensic trace on 18 of them, which is what the organic versus amplified reading needs. Both are published with their method so every figure can be re-derived.
Which vertical launches biggest?
Across the 583 tracked launches with measured reach, AI and SaaS make up 58% of the field but reach the top 10 percent at only 8.9%, because the category is crowded. Hardware is rare (9 launches) and 2 of them reached the top decile, too few to call a pattern. In the 30 audited case studies, AI leads the set on combined reach, with 7 launches totaling 42.8 million views. The set spans 6 verticals in all, from AI and developer tools to fintech, crypto, marketing, and consumer and health. Per-vertical cuts on a 30-launch set are directional rather than definitive, which the report states plainly.
What is a good views-to-likes ratio for a launch video?
On X, organic reach tops out near 500 views per like, the point where reach and engagement grew together. Across these launches the median was 788 views per like and the 90th percentile was 3,126. The organic launches ran a median of 354, the amplified launches 1,441. Lower is closer to organic. The full band calibration lives on the views-to-likes benchmark.
Is a launch video with millions of views fake?
Not necessarily. A high view count paired with the likes, replies, and quote-posts to match is genuine reach that connected with a real audience. A high view count with the engagement far behind it is the signature of distribution-amplified reach. Paying for or coordinating distribution is legal and common, so a wide gap describes how the reach was built, not whether anyone was dishonest.
How was this report measured?
Two layers. The tracked field is every launch on the public RADAR board (967 launches, all read on 2026-09-24), with reach taken as the peak view count RADAR recorded on the launch post; 583 have measured reach so far. Opening lines are classified by pattern from the post text, and launch times are converted to Pacific. The launch-video figures come from the 25 audited launch videos, each probed for duration, orientation and audio. The 30 case studies add public views, likes, replies, quotes and reposts, with a full forensic trace on 18 of them. Every figure is recomputed from the published dataset.
Can I download the launch video dataset?
Yes, as two CSVs under CC BY 4.0. The RADAR aggregates file carries the reach bands, percentiles, category, opening-line and weekday breakdowns, and the launch-video length and format figures. The case-study file carries the overview, the per-vertical breakdown, the ratio percentiles and the anonymized per-launch ratio distribution. Launch-level rows are anonymized by design, so the files publish distributions, not a per-named verdict. Attribution is FORKOFF (2026), State of Startup Launch Videos 2026.
Authorship
Simba
Launch Intelligence Analyst, FORKOFF
Reviewed by: Kshitij JK
Last reviewed:
Published:
Methodology
FORKOFF RADAR tracked field: 967 startup launches on the public RADAR board read 2026-09-24, 583 with measured peak reach; 25 launch videos probed for length, orientation and audio. Case studies: 30 public product-launch posts on X, Oct 2025 to Jun 2026, 110.4M combined views. Reach and views-to-likes ratio computed per launch; organic ceiling near 500 views per like from the published RADAR methodology. Full forensic trace on 18 of 30 launches; the rest read from the public ratio at lower confidence. Numbers recomputed live as the set grows.
Sources cited
- RADAR methodology , FORKOFF
- Launch Radar hub , FORKOFF
- Views-to-Likes Benchmark 2026 , FORKOFF
- IAB 2025 Digital Video Ad Spend and Strategy Report , IAB
The measurement side of how FORKOFF moves
Want a launch video that earns its reach?
This report grows out of running managed launches at the scale where the difference between earned and amplified reach stops being a guess and becomes a measured signal. Talk to a strategist about a launch built to earn its ratio.
Receipts, deep dives, and playbooks.
Launch Distribution for Tech, SaaS and AI Companies: How It Works in 2026
Launch distribution decides who sees a SaaS or AI launch in its first 72 hours. The five stages, the 14-day warm-up and what billboards miss.
Read
What Makes a Launch Video Go Viral? 1,075 Startup Launch Videos, Measured
What makes a launch video go viral, from 1,075 startup launch videos we labelled: the opening hook, 60 to 120 seconds and a working demo beat polish.
Read
The Biggest Launch Videos on X This Week, Measured: What Earned the Views
The biggest launch videos on X from Oct 17 to 24, 2026, measured on our launch RADAR: who earned the views, who amplified them, and what to brief an agency on.
Read
