Table of contents:
1. Data Visibility to Unlock Your Content Strategy | 1 |
2. Tools for Content Tagging | 2 |
3. Identifying Engaging Content | 2 |
4. Performance Prediction Models | 3 |
Conclusion | 3 |
At CBS Media Ventures, David Katz is tasked with driving monetization from Paramount’s extensive archive of content, identifying “what works where”.
His learnings were welcomed on the GDS platform as the executive attendees grapple with this question in a climate where video content reigns supreme. Globally, people are watching around 84 minutes of video(s) every single day.
What’s more, 52% of people are more likely to share videos over other types of content like social posts, product pages or blog posts.
But in such a saturated space, cutting through the noise comes with immense challenges.
“We’re buried under a mountain of data,” said Katz, before outlining four game-changing strategies that Paramount have used to successfully identify value within their mountain and generate engagement.
1. Find Out What’s in Your Data Mountain
Embarking on an effective content strategy (especially one including AI and ML) requires a crucial initial step: introspection. It entails exploring the data at your disposal and understanding its whereabouts. Without this essential knowledge and visibility, leaders will struggle to make discernible progress.
“I have a small content team, so I had to make sure they weren’t wasting limited time and resource searching for value in the wrong places,” said Katz.
Gaining visibility into your data will unlock your ability to set clear objectives, define your strategy, and track performance.
On this pursuit, implementing a robust meta-data tagging strategy has proven invaluable for the CBS Media Ventures team.
2. Do More with Less: Tools for Tagging
To maximise his small team’s time, Katz outlined how he had explored AI and ML to streamline particularly labour-intensive processes.
He used the American thriller series “Yellowjackets” as a case study since – at two seasons and 19 episodes – it consisted of a small number of high value assets.
Katz deployed Amazon Rekognition to identify what was happening in each scene – “It identified a person, a dog, water, grass…”. Then a media asset manager categorised and tagged each asset.
“It allows me to search for things that have been identified, so I don’t have to have a user go through and watch the video and manually tag it, I can have the machine do it.”
3. Find the Good Stuff: Tools for Timestamps
Tagging is instrumental to unlock access and searchability within your data. However, it doesn’t necessarily differentiate the good from the bad.
Content leaders must apply business specific insights to their data. In Katz’s case – insights revolved around stories.
“Content has a story behind it. There are emotions and sentiments there,” highlighted Katz. “So, I worked with a partner that allowed me to understand what was inside that content.”
Here, Katz highlighted a case study involving Dabl – a lifestyle network. It covers a broad scope of topics (cooking, interior design, travel, real estate and more), meaning there is a vast array of content and data points in its archive.
His content team use a language tool to identify timestamps for positive and negative sentiments. It maps out a timeline of each episode and displays “peaks” in emotion. This allows users to identify particularly engaging elements to potentially clip up and distribute.
4. Discover What Works Where
Of course, the challenge for content leaders does not end with the organization and visibility of their data. Even unearthing the most engaging data doesn’t immediately translate it into monetizable value or engagement.
Note, the top challenges executives shared in the Semrush State of Content Marketing 2023 report revolved around attracting quality leads with their content (41%), generating enough traffic (39%), creating content that resonates with their audience(s) (31%), and proving ROI (30%).
Katz experimented with performance prediction technology to help him understand which platforms would be most suited to his content, in turn generating more clicks, likes, shares, comments and more.
The case study he shared was around “Inside Edition” – a news magazine programme that has been on the air for 35 years but has also generated a huge new audience on YouTube.
Here, the machine takes things a step further than offering readable data; it makes suggestions for how to action it too.
The model sorts through the archives to uncover what stories have been covered and upon which platforms they would be well received: Instagram, Facebook, YouTube, TikTok etc. What’s more, it highlights stories and content that will not perform well at all.
“This points my team in the right direction so they can go right to the pieces that the machines think are going to be useful and get started on it.”
Conclusion: Don’t Experiment if You’re Not Prepared to Fail
These off-the-shelf tools and models have highlighted ways in which content leaders can enjoy routes to improved organization, understanding and decision-making around their data and content reserves.
Through experimentation and piloting, Katz has unearthed a vast array of manoeuvres to streamline processes, build strategies with confidence, and protect the working hours of his team(s).
But against a changeable economic backdrop, content leaders often find themselves under pressure to prove their value to the wider business. This does not always lend itself to creating a culture or experimentation that encourages – or even accepts – failure.
However, Katz urged the leaders in attendance at the Digital Innovation summit to acknowledge the integral role failure plays in innovation.
“Go play with these systems. Know that you’re going to get some things wrong but give yourself some grace and learn, because it’s a super exciting time.”