How to Measure Video Content: Crafting Insights Using the Scientific Method (Part 4)

by Emily E. Steck

How to Measure Video Content: Crafting Insights Using the Scientific Method (Part 4)

You’ve created content; you’ve distributed content and you’ve analyzed the video metrics and data from said content. Now you have to craft insights (and actions) from your content’s data.

Part of being a content marketer or savvy brand is having the ability to remove the “noise” of the data landscape to isolate the most important insights that can yield value for your content marketing in the future.  Video marketing is no exception. In fact, there’s a science to using the data collected from YouTube Analytics and Facebook Video Insights to measure video content (and content in general). Here’s how you can craft insights from your video content.

Crafting Insights Using the Scientific Method

Everyone uses data. Government agencies, marketers, business executives, oh, and scientists! Scientists use data all the time. So to use data, and craft insights about your data, think like a scientist. Nay, think like a data scientist.

To adopt this mindset, data scientists and content marketers use a modified version of the scientific method. The scientific method is an ongoing process, and the modified one usually looks like this:

  • Ask Questions
  • Do Background Research
  • Construct a Working Hypothesis
  • Test Your Hypothesis
  • Analyze Data and Draw a Conclusion
  • Communicate Insights into Actions

The Scientific Method for Content Marketers

By Emily E. Steck

  • Ask Questions

    By Emily E. Steck

    **Any good data scientist starts by asking questions.** _Why is my data doing this? How can I prove that X is performing because of Y._ So on and so forth. The quickest way to ask important questions is look for anomalies in the data. Aka _Why did traffic spike here? Why did it dip here? What is exciting our audience?_ Data scientists look for data outliers, spikes or any other funny business because those are the most pertinent questions. Spikes in data could mean any number of things; maybe you shared it on social media day, maybe your video was picked up by a site like Buzzfeed, maybe it was featured on YouTube’s home page or maybe more of your audience uses YouTube or Facebook on the weekend versus the week.

  • Do Background Research

    By Emily E. Steck

    **To answer those “maybes” into definite-lys, you’ll need to do background research.** By digging deep into your various reports from YouTube Analytics and Facebook Video Insights, you’ll begin to be able to trace spikes in traffic, for instance, to its source. For example, by looking at the referral traffic from one of your videos, you may see that people are watching your video on_ Gawker._ Or by looking at another report, your videos receive the best traffic when you post on Mondays than Wednesdays.

  • Construct a Working Hypothesis

    By Emily E. Steck

    Most of the time, however, tracing the origins of said spikes or dips is more complicated than pinpointing one reason. There are many. That’s why content marketers, strategists and data scientists **develop working hypotheses or theories to explain the questions that need to be answered.** For example, if content marketers are trying to determine why the average bounce rate increased over the last month, they’ll develop a working theory based on data and industry knowledge. For example, the [dreaded 100% bounce rate](http://blog.quiet.ly/industry/the-dreaded-100-bounce-rate-and-what-its-really-saying-about-your-content/) strikes fear in any marketer. However, the bounce rate measured by Google Analytics does not always accurately account for session duration and time on page. (For a detailed explanation on that, head [here](http://blog.quiet.ly/industry/the-dreaded-100-bounce-rate-and-what-its-really-saying-about-your-content/)).

  • Example: Hypothesis

    By Emily E. Steck

    Here’s a **hypothesis:** “The bounce rate increased this month most likely because of a paid Facebook marketing campaign that sent users to our site, who then promptly left. In addition, the data for pages visited per session were historically low while our best-performing stories were driven by search. This leads us to believe that people could have found our content and then left the site. Additionally, bounce rates are sometimes inaccurate measuring session duration and time on page, which caused our bounce rate to increase.” Complicated, but comprehensive hypotheses like this one can help craft deeper and better insights about your data.

  • Test Your Hypothesis

    By Emily E. Steck

    But to see if any of this is correct, you’ll need to** test your hypothesis.** If it’s more complicated, like the one above, it’s not bad idea to test multiple parts of the hypothesis. To do this, you’ll need a flexible content strategy and rigid date periods. Consider testing month-to-month or week-to-week for better, more accurate comparisons. Use control groups to test against. For example, if you used a paid Facebook post the month before, do not use a paid Facebook post for the month you are testing; do everything else the same. Over time, you’ll see emerging patterns and will be able to develop complex testing for your hypotheses.

  • Analyze Data

    By Emily E. Steck

    After testing your hypothesis, you’ll then **analyze the data once again**. Perhaps the bounce rate increased that month for every reason you guessed, but the bounce rate mostly increased because of the paid Facebook campaign. Or it was mainly because of search and the paid Facebook campaign had little influence. Or perhaps the low pages visited per session hurt bounce rate the most because people were not staying on the site. And sometimes, your hypothesis could just be wrong; that’s why the scientific method is an ongoing method. Then you’ll have to create another hypothesis and start again.

  • Draw a Conclusion

    By Emily E. Steck

    But eventually, you’ll **draw a conclusion from your data and create insights** from said data. For scientists, that can mean publishing a paper. For content marketers, that means **using insights to develop content strategy, creation and distribution.** To go with our bounce rate example, a good content marketer could take an insight like the low pages visited per session and suggest ways to keep visitors on the site longer. This could manifest as using different types of content—from video to infographics to text—or shorter-form content or longer-form content or incentives to keep people on the site longer. That’s up to the ingenuity of the content marketer, isn’t it?

  • Communicate Insights into Actions

    By Emily E. Steck

    And finally, FINALLY, comes the time to **communicate those insights into actions**. This is a combo-collaborative effort between content marketers, strategists, editors or however else you build your [superhero content marketing team](http://blog.quiet.ly/industry/what-superheroes-can-teach-you-about-building-a-content-marketing-team/). What it comes down to, though, is making sure that your data transforms into insights and that those insights are put into action.

And that’s how you use the scientific method to observe and act on your data.

5 Useful Tips to Create Better Insights from Video Data

  1. Keywords are your best friends. How are people searching for your videos? Namely, look at the keywords they are using to find them. Some you will have optimized for, but others will be a pleasant surprise. Keyword optimization will help your videos be found, but you can also use keywords to create better ideation of content strategy and creation.
  2. Analyze your average view duration/audience retention graph to know where and when to place CTAs. Plenty of creators place annotated CTA cards at the end of videos, asking viewers to click on a landing page. However, plenty of people don’t make it to the end of a video. Look at the average view duration to understand where your viewers drop off at so you can place these CTAs in the video in more strategic spots. You can also look at the annotation and cards report to see where and when viewers click-through the most.
  3. Use demographics data to learn who’s watching your content. More Canadians than Americans? Women than men? Who watches what when? All of these better inform you of your audience and content strategy. If you’ve adjusted your content to better suit your audience, you’ll see differences in the average view duration and engagement reports.
  4. Optimize for time. When is your audience online? Because that’s the time you want to post your content. Look at views reports to understand when people are there. Most importantly, though, look to the length of your videos. Where is your audience dropping off? Take a look at the audience retention graphs/average view duration to see this.
  5. Understand different platforms. Facebook and YouTube are very different platforms. Treat them as such with your content. The average length of Facebook videos was 44 seconds and research indicates that 21-second videos were most frequently watched in full. YouTube, meanwhile, has an average online content video of 4.4 minutes. Additionally, consider where your audience is most likely to watch your content: Facebook or YouTube?

Of course, there are an infinite amount of ways to interpret and analyze data and to create insights and actions from that data. This is just a starting point. But hopefully by now, you understand the nuances of crafting insights from data.

Stay tuned for the final entry in this series. In the meantime, catch up with the rest of the series:

Understand how Quietly can help play a role in your content marketing efforts.

Speak to a Strategist Today

Get a free consultation for your content marketing strategy.

Speak to a Strategist Today