M&E Journal: Unleashing the Power of Smart Content
By Ryan Steelberg, President, Veritone –
For today’s M&E companies, staying profitable is complicated by two intertwined phenomena: exploding consumer demand for personalized, relevant content and an ever- fragmented media environment.
Consumers not only want greater volumes and greater varieties of content, but they want to consume it when they want, and on the platform and device of their choice. The M&E industry — driven by profit and a virtually unlimited world of new revenue possibilities — continues to dream up new ways to package and deliver content. The resulting fragmentation of the media landscape (we call it the YouTube effect) means that content can be personalized down to the level of individual consumer’s tastes.
Through AI, analytics and attribution solutions, neither linear nor digital broadcasters have to wait overnight for ratings to come out; it’s now possible to know, in near real time, who is watching a clip and on what device or browser.
That’s both good news and bad news for M&E companies: it means they can offer advertisers an unprecedented ability to tailor messages to highly targeted groups of consumers and their programs can be much more strategic through superior metadata and content intelligence. Preparing those targeted content assets, however, is straining production and distribution resources like never before.
In order to find and monetize a particular asset and use it for a specific task or project, users need to be able to search by details such as date produced, persons/companies featured, topics, location description, keywords, etc. The catch is capturing this level of detail from unstructured video, text and audio content. Until now, it’s been a laborious process requiring someone in the operation to sit and view hours and hours of programming and manually tag content with descriptive metadata.
Hyper-indexing to extract metadata
The good news for M&E companies is the emergence of new AI-enabled solutions that hyper- index content to extract intelligent, structured metadata. These progressive applications use advanced, AI-driven cognitive processing techniques such as facial and object recognition and detection, speaker separation, content classification, OCR, transcription, translation and more.
Near real-time metadata generation replaces those manual, labor-intensive and time- consuming tagging processes, freeing people from the tedious task of listening to entire programs and allowing them to focus on more strategic, revenue generating initiatives.
Instantly, the intrinsic value and usability of content takes a leap since every asset is enriched with detailed metadata. Unstructured assets are transformed into smart content, making them imminently easier and more cost-effective to access, leverage, verify and monetize.
Consider the case of a large broadcasting network that has been working with Veritone for several years. This network has successfully ingested all of its primary linear content, across three TV channels and numerous other audio-based networks (including radio and podcasts), into Veritone’s aiWARE solution, where it has been hyper-indexed according to criteria such as faces, logos and other cognitive classes. The result is a highly accurate and well-organized metadata index that the network applies in a multitude of use cases: optimizing programming, optimizing ad revenue and optimizing analytics to drive decision-making on future investments in content.
History repeats itself
To understand the promise of smart content for the vast universe of unstructured media assets that lurk in virtually every M&E company, it’s useful to look at a historic parallel: the growth of the internet. In the early 1990s, when the internet was in its infancy, just about anyone could (and did) put up a website — and there was a complete lack of metadata or any real unifying structure for content. HTML emerged to provide a common data layer; in other words, the structured metadata, image source tags and other elements content owners and advertisers needed to realize the internet’s true revenue potential for targeting content.
While as of yet, there is no comparable unifying standard or common data layer for unstructured media content, AI tools for hyper-indexing content with highly descriptive metadata are laying the groundwork. Just as HTML 3.2 has enabled the highly targeted advertising that drives today’s vast internet marketplace, tools for smart content are getting M&E companies in on the ground floor of a universe of revenue opportunities through highly targeted advertising and extremely relevant content.
Economies of scale
The M&E industry still has a long way to go before fully realizing the dream of hyper-indexed smart content. Legacy media companies, in particular, have their work cut out for them — not only are many just beginning the migration to next-generation, enterprise software systems, but they’re sitting on a vast archive of content that’s still held in old-school formats like tape and film, with only a small fraction of assets digitized and ready to be indexed. But even legacy companies are waking up to the possibilities of smart content, and most have a long-term strategy for content digitization and indexing.
Looking ahead, we see an era in which the cost of hyper-indexing long-form, unstructured audio, text and video content will continue to fall as algorithms and processing are optimized.
This will come as M&E companies realize the power of rich metadata for optimizing everything from targeted advertising to program scheduling and even search engine optimization for broadcasters and managing, sharing and monetizing for content owners. Ultimately, it’s all about the viewers: to meet consumer demand, media companies really have no choice but to get smarter about how they prepare media and content assets.
Hyper-indexed smart content is and will continue to be the answer.