Video marketing statistics: the famous ones, traced back to their sources

The method is the same for every claim. Take the sentence as it circulates. Follow its citations backwards until they stop, which is usually two or three hops. Read what the earliest locatable publication actually measured, on what sample, in which year. Compare that with the sentence on the slide.

HOLDS means a real, named source stands behind the claim and still describes something true, even if it deserves a caveat. OUTDATED means the origin is real but it measured a different year, a different platform or a different thing than the sentence now implies. UNVERIFIABLE means no primary source can be found at all, or the origin itself disowns the number.

Only two of the thirteen earn a HOLDS. That is not because video marketing is a fraud. It is because a number that survives ten years of being copied without its context usually left the context behind in year one.

Claim one: 85% of videos are watched without sound. The chain runs to one article: Digiday, May 17, 2016. Digiday asked publishers about their own Facebook data. LittleThings said 85% of its roughly 150 million monthly Facebook views played silent. Mic said 85% too. PopSugar put its share between 50% and 80%, and the agency MEC North America said branded videos averaged 85% to 90% silent (Digiday, 2016).

What it measured: a handful of publishers' videos, autoplaying muted in the 2016 Facebook feed, where a view counted at three seconds. It never measured all video, all platforms or all years. Verdict: OUTDATED. Quoted as a law of video in 2026, it is a decade-old observation about one feed that autoplayed on mute by design.

Claim two, its mirror: TikTok is watched with the sound on. The chain here is short and real. In June 2021 TikTok published research it had commissioned from Kantar: 88% of TikTok users said sound is essential to the experience, and 73% said they would stop and look at ads with audio (Kantar for TikTok, 2021). Verdict: HOLDS, with the caveat said out loud: it is a survey about one platform, paid for by that platform, and it measured what users said rather than what players logged.

The honest read for anyone posting a weekly talking video: burn in captions, because feeds still mute by default, and keep the voice worth hearing, because the platform where short video lives rewards sound.

Claim three: viewers retain 95% of a message from video and 10% from text. This one traces to a single origin, and the origin has publicly described it. Insivia, the agency whose name is on most citations, states the figures came from a small survey it ran in the late 2000s, of roughly 200 B2B buyers, comparing self-reported recall. In its own words: it was directional, and it was not a peer-reviewed neuroscience study. The exact percentages, again in its own words, were a pattern in that survey, not universal laws of cognition (Insivia, origin's own account).

Verdict: UNVERIFIABLE. Not because the origin is hidden, but because the origin itself says the precision is borrowed. A twenty-year-old survey of 200 people, reported as self-assessed recall, cannot carry the sentence 'viewers retain 95% of a message', and its author does not claim it can.

Claim four: the average attention span is 8 seconds, shorter than a goldfish's. The chain: a Microsoft Canada consumer report from 2015 carried the number, the world's press repeated it, and in 2017 the BBC's More or Less programme traced it. The 8-second figure appears in the Microsoft report only as a citation of a site called Statistic Brain, which could not produce a source when asked. Microsoft's own experiments in that report never produced the number. The goldfish half fails separately: researchers use goldfish in memory studies precisely because they learn and remember over months (BBC, 2017). Verdict: UNVERIFIABLE, both halves.

If you are a coach or an advisor who presents to clients, this pair is the one to quietly retire from the deck. The case for short, clear video survives without a fake neuroscience floor under it.

Claim five: video on a landing page increases conversions by 80%. The chain runs through Unbounce and hundreds of listicles to EyeView Digital, a video advertising firm, and then stops. The underlying report is not publicly available, no year or sample is attached, and the retellings disagree with each other on the number itself, 80% in some, 86% in others. Verdict: UNVERIFIABLE. A real company sits at the bottom of the chain, but a study nobody can read, with a headline figure that drifts between citations, is not a statistic. It is a rumour with a percent sign.

Claim six: real estate listings with video receive 403% more inquiries. This is the most instructive chain on the page because it goes nowhere. Current listicles attribute it to the National Association of Realtors, to PropertyNest, or to DigitalSherpa, and the attributions contradict each other. No NAR publication containing the figure can be found. No methodology, no year, no market has ever been attached to it. Verdict: UNVERIFIABLE. For a realtor filming a weekly talking video, the honest version needs no decimal places: the agent a buyer already recognises gets the call, and that is built by showing up on camera, not by quoting a number nobody can source.

Claim seven: adding video to email increases click-through rates by 200% to 300%. This one has a real origin, and the origin is smaller than the legend. Forrester published a trend report in 2009 called As Seen In The Inbox. It reported interviews with marketing executives suggesting that compressed video without sound, animated GIFs and clickable screenshots of videos could lift click-through by two to three times (Forrester, 2009). Verdict: OUTDATED. It was practitioner opinion, collected in 2009, largely about thumbnails and GIFs that linked out. Time turned an interview finding into a measured-sounding law about embedding video in email.

Claim eight: social video generates 1,200% more shares than text and images combined. The chain runs through G2 and a 2016 Forbes column to an attribution to Brightcove, and stops there. No Brightcove report containing the figure is publicly locatable, and no year, platform or sample has ever surfaced (attribution via Forbes, 2016). Verdict: UNVERIFIABLE.

Claim nine: video is 82% of all internet traffic. The origin is real and precise: Cisco's Visual Networking Index, 2018 edition, forecast that video would reach 82% of IP traffic by 2022, up from 75% in 2017 (Cisco VNI, 2018). Two problems. It was a forecast, not a measurement, and it was about 2022. The VNI series then ended; Cisco's Annual Internet Report, published in 2020 and covering 2018 to 2023, replaced it (Cisco, 2020). Nobody is publishing the current-year figure this claim pretends to be. Verdict: OUTDATED, quoted in the present tense about a forecast horizon that closed four years ago.

Claim ten: YouTube is the world's second-largest search engine. The chain runs to comScore analyses of the US search market from 2008, which counted queries typed into YouTube's own search box alongside search engines and found the volume beat Yahoo's (comScore, 2008). That measured searches on one website, not search-engine market share. On actual web search, StatCounter data reported by Statista puts Google at 91.32% in July 2026, Bing at 4.46% and Yahoo at 1.24% (StatCounter via Statista, 2026). Verdict: OUTDATED. It described 2008 site-search counts, and it answers a question nobody is asking when they put it on a slide.

Claim eleven: TikTok users watch 95 minutes a day. Real number, real source: Sensor Tower reported roughly 95 minutes a day as a global average in the second quarter of 2022 (Sensor Tower, 2022). It still circulates as current. eMarketer's 2025 forecast put US adult users at about 52 minutes a day, the first projected decline in the platform's history (eMarketer, 2025). Verdict: OUTDATED. The 2022 global figure is not the 2026 US figure, and the gap is not small.

Claim twelve: one minute of video is worth 1.8 million words. The origin is Forrester analyst James McQuivey's 2008 report How Video Will Take Over the World, and the number was never a measurement. It is arithmetic on an adage: if a picture is worth 1,000 words, and video runs 30 frames a second, then 60 seconds is 1.8 million words (Forrester, 2008). Verdict: UNVERIFIABLE, in the strict sense that there is nothing to verify. Its own author built it as a rhetorical flourish, and it has been cited as research ever since.

Claim thirteen: 87% of marketers say video gives them a good ROI. This one traces cleanly. It is from Wyzowl's State of Video Marketing survey, the 2021 edition (Wyzowl, 2021). The survey is real, named, and has run annually for twelve years, which makes it the closest thing this industry has to a longitudinal record.

The caveats are the sample and the year. The 2026 edition surveyed 266 respondents, a mix of marketing professionals and consumers, polled in late 2025 (Wyzowl, 2026). In that edition, 91% of businesses said they use video as a marketing tool, and 82% of marketers reported good ROI, down from 93% the year before (Wyzowl, 2026). So the famous 87% is not wrong, it is simply 2021. Anyone quoting it today is quoting a five-year-old edition of a survey that publishes a fresh number every year.

Verdict: HOLDS, said properly: a couple of hundred self-selected respondents, surveyed by a video company, still say video pays. Quote the current edition, name the sample, and it is an honest sentence. For a recruiter or advisor deciding whether a weekly talking-head video is worth the habit, the trendline matters more than any single percentage: twelve consecutive years of the same survey saying roughly the same thing.

A usable statistic has a source, a year and a sample you can name in the same breath. These survived the tracing, quoted with all three.

The chains above rest on these publications. Primary sources are listed where one exists; where only an attribution exists, that is stated, because that is the finding.

This page is on a video tool's website, so here is the honest version of the pitch, held to the same standard as everything above.

Cutroom turns one spoken phone take, up to 3 minutes, into a finished vertical video. It transcribes every word, writes the hook, places b-roll on the exact words that describe something, burns in captions, and returns a 9:16 MP4 in about a minute. You direct on the transcript, not on a timeline. That is the whole product, and it exists for people who publish a talking video every week and lose the evening to editing it.

The price, stated the way we wish more statistics were: a batch is 100 credits, export adds 20 credits per finished minute, so a finished 30 second video is about 110 credits. The first 3 finished videos are free in your first 7 days, with 600 credits and no card, and they export with a Cutroom mark across the middle. Lite is $19.99 a month for 1,250 credits, about 11 of those videos, takes the mark off, and cancels in one click under Settings. Basic is $39.99 for 2,500 credits and adds avatars and generated music. Premium is $79.99 for 5,000 credits.

What it will not do: sources cap at 3 minutes, exports are 9:16 MP4 only, and no statistic on this page promises you results. Neither do we.