How AI Impacts Video View Time and Engagement [New Data]


Online video marketing has been a big topic of discussion for marketers much of this decade – and rightfully so. Even before companies like Wistia came on the scene late last decade there were very specialized blogs dedicated to the topic of online video (remember ReelSEO? Loved that blog). Even my good friend, Rocky Walls, was leading the online video thought leadership at 12 Stars Media in Indianapolis back then. Much of what I know about video I learned from him (h/t Rocky).

Fast forward to today. The statistics around consumer and brand use of online video are striking in many instances. Its use has skyrocketed and is predicted to have even further growth moving into next decade.

We in the inPowered labs decided to open up the big data machine and look at all of the online video content we amplified in 2018 to see if we could glean anything new or unique in the numbers pertaining to online video amplification and engagement. We looked at nearly one half of a million video views. Our artificial intelligence (AI) specifically optimizes for engagement – whether video or written. Based on research done by Chartbeat Analytics is how we define engagement (more on this in the methodology section below). As a result, we’ll explore, on average, how long users watched video from our platform.

However, before we get to the data-science stuff. Let’s look at some remarkable video marketing stats for 2019. Luckily, the good folks over at Biteable curated 55 of them. We grabbed 20 to highlight.

  • 81% of businesses use video as a marketing tool — up from 63% over the last year. (Hubspot)
  • 6 out of 10 people would rather watch online videos than television. (Google)
  • Mobile video consumption rises by 100% every year. (Insivia)
  • By 2022, online videos will make up more than 82% of all consumer internet traffic — 15 times higher than it was in 2017. (Cisco)
  • 78% of people watch online videos every week, and 55% view online videos every day. (HubSpot)
  • A Facebook executive predicted that their platform will be all video and no text by 2021. (Quartz)
  • YouTube is the second most popular website after Google. (Alexa)
  • Users view more than 1 billion hours of video each day on YouTube. (YouTube)
  • 59% of executives say they would rather watch a video than read text. (Wordstream)
  • 75% of all video plays are on mobile devices. (eMarketer)
  • Viewers retain 95% of a message when they watch it in a video, compared to 10% when reading it in text. (Insivia)
  • 72% of customers would rather learn about a product or service by way of video. (HubSpot)
  • People are 1.5 times more likely to watch video on their mobile phones. (Facebook)
  • 92% of users watching video on mobile will share it with others. (Wordstream)
  • By 2020 there will be close to 1 million minutes of video crossing the internet per second. (Cisco)
  • Nearly 50% of internet users look for videos related to a product or service before visiting a store. (Hubspot)
  • 85% of consumers want to see more video content from brands. (HubSpot)
  • 65% of executives have gone to the marketer’s site and 39% have called them on the phone after watching a marketing video. (Forbes)
  • 97% of marketers say video has helped users gain a better understanding of their products and services. (Hubspot)
  • 52% of marketers say video is the type of content with the best ROI. (HubSpot)

Here’s some other interesting numbers worth noting:

  • Video marketers get 66% more qualified leads per year. (Optinmonster)
  • Video marketers achieve a 54% increase in brand awareness. (Optinmonster)

The above statistics speak for themselves. Video is a powerful tool in the marketer’s content quiver, indeed. This is unquestionable. However, what happens when video is optimized using AI for engagement metrics? Let’s find out.


First and foremost, we have to define what an online engagement is. There’re several ways of doing it, but in the inPowered labs we like the 15 second benchmark from Chartbeat Analytics. All of our customers only pay per engagement at this benchmark. We don’t charge for clicks. They were able to show that at 15 seconds or longer 70% of the people consume 80% of the content or more.

And what do you know. . ? The numbers below bear this out. Of the users that viewed at least 15 seconds of video out of nearly one half a million, 83.6% of the video was consumed. This is native paid media folks, not the organic measurements. It’s the use of AI doing micro-testing on thousands of variations and permutations of native ad units across over 40 social media and native networks that makes performance like this possible.

Avg. Completion Rate Avg. Engaged Time Post :15 Avg. Video Length
54% 51 61


Next, we decided to break down the engagement into quartiles. As shown below the levels of actual engagement are very high for video. In total, post-15 second engagements nearly reached 65,000 hours last year. Given the stats mentioned above, that can have real advantages for a brand trying to build awareness, affinity and consideration.

                                                                                                 Percentage of Video Viewed
Total Views 25% 50% 75% 100%
442,941 123,224 101,830 89,225 65,510
Average # of Views in Each Quartile 28% 23% 20% 15%


Another thing we know, and Facebook backs this up in its new TruPlays product, that the vast majority of viewers don’t actually watch the video to its end due to the credits and post-meaty content at the end of any video. Having over 65,000 people watch videos to the end is nothing short of amazing.

The takeaway here is simple. Paying per click for a video is a fool’s errand. We’ve found zero correlation between clicks and actual content engagement. In fact, many times we’ve found a statistical negative correlation between clicks and actual content engagement. This is from nefarious bot traffic that eats away at budget and is precisely the reason P&G pulled millions of dollars out of paid media. However, in order to drive performance like this it requires AI. An actual human being would take weeks to micro-test thousands of native ad unit permutations.

Learn Cutting Edge Content Marketing & Native Advertising Insights