The Future of Content Marketing: How Machine Learning and UGC Drive Scale
From content classification to curation and optimization, machine learning can make creating content so much more efficient for marketers.
In the United States, marketers spend $10 billion a year on content marketing, of which half — $5.2 billion — is dedicated to content creation alone. Interestingly, research also indicates marketers waste about 20 cents of every dollar creating ineffective content, which comes to a whopping $1 billion yearly loss.
While that number is jarring, it shouldn’t be a surprise. According to the Content Marketing Institute, only 27 percent of marketers say they see very concrete returns from their content marketing campaigns. Despite these dismal returns, 50 percent of marketers say they will be increasing their marketing spend this year. What is driving this dissonance?
The truth is, creating effective content at scale is difficult, costly, and resource-intensive. To make matters worse, in an increasingly personalized world — where everyone competes for attention online — developing a continuous stream of engaging content is no longer a differentiator, but a necessity to compete. Companies keep “throwing money at the problem,” but are they investing where it counts?
An effective but often overlooked resource to alleviate content scale issues is user-generated content (UGC).
UGC: A plentiful mine of content waiting to be capitalized
UGC is the broad spectrum of consumer-created content, shared across digital properties and social networks. Beyond Instagram posts, YouTube videos, and tweets, UGC includes reviews, blog comments, audio files, and all other digital breadcrumbs that users leave on the web.
As a civilization, we are creating content at an unprecedented rate, generating billions of pieces of content daily. While a majority of UGC is noisy, there are thousands of micro-influencers creating high-quality content about your brands and experiences that is waiting to be discovered and leveraged. In fact, numerous studies show that UGC is more trustworthy, engaging, and converts better than branded content. This shouldn’t come as a surprise, as letting your customer tell the story for you is much more authentic and believable. So the question is, why aren’t more companies embracing it?
Simply put, finding the “golden nuggets” in the mountains of web content has been historically difficult. However, the recent progress and democratization of machine learning — particularly of computer vision and natural language processing — is changing the marketing technology space by automating many of the arduous manual tasks of the past.
The machine-learning advantage
To better understand how machine learning is accelerating content marketing, it’s helpful to look at the past: how artificial intelligence “AI” has evolved since the field first emerged in the 1950s, advancing from the automation of narrowly defined tasks to the automation of human-centered tasks, which include content curation, management, and optimization.
The first wave of AI, ushered by the Symbolist school of thought, revolved around building rules-based expert systems that codified a narrow problem as a set of rules that a computer could interpret and automate tasks from. Wave 1 AI is great at reasoning but not at perceiving, learning, or abstracting — the ability to apply knowledge learned in one area to another.
Then came along the Connectists, whose AI inventions sought to mirror how human neural networks operate. These systems are great at perceiving the natural world, such as voice, images, and text. They use statistical models to make predictions from the inputs they perceive and learn from the feedback they receive. Thanks to the cloud computing revolution, the availability of big data sets and the steady innovation in the field, these systems — like Adobe Sensei, Adobe’s artificial intelligence and machine-learning framework — have evolved today to the point where they can perform certain tasks at or above human level, like object recognition or natural language processing.
Further, a third wave of AI is emerging, which combines the learnings of wave 1 and 2 systems, to build programs that can learn in an unsupervised fashion from the context in which they operate, and even abstract those learnings into other areas. Deep Mind’s Alpha Go and Alpha Zero are fascinating examples of the superhuman potential that wave 3 systems hold.
All of these capabilities mean machine learning can bring a level of efficiency to human tasks that is difficult for professionals to achieve on their own. That could be key in helping you overcome the time-intensive work required to create great content experiences for your customers — many marketers say the top challenges they face are producing engaging content, measuring content effectiveness, measuring ROI, producing content consistently, and producing a variety of content.
By combining the data collection, automation, and predictive capabilities of machine learning with your creativity as a marketer, your company can optimize and scale its campaigns to create more personalized and engaging content.
What does AI look like in the marketing space?
To assess the roles of AI in software, MIT Professor Patrick Winston’s framework is extremely helpful. It breaks AI down into four categories: tools, assistants, peers, and managers.
Tools such as autocomplete, autocorrect, or auto-translate are simple tasks a system performs upon request, but that require human oversight and approval. In marketing technology, AI tools include content classification models, which can automatically tag everything that is inside content, extracting entities, aesthetic properties, sentiment, and even emotions.
Assistants can perform tasks without supervision. A chatbot, for example, is trained to answer based on a previous database of knowledge, but no one is constantly monitoring the chatbot’s activity per se. In marketing technology, content recommendation engines, assisted curation, and experience optimization systems also fall under this category.
Peers can complete human tasks independently, kicking problems to human backups as needed. In the future, there will be automated content moderation, automated rights management, and digital peers that help marketers with day-to-day marketing activities.
Then there are managers, which function like AI-powered supervisors, that allocate work intelligently and provide real-time feedback, just like the service Cogito does during customer-support calls.
All of these technologies enable marketing organizations to operate more efficiently and to scale their customer interactions. But companies need the right technology partner to fully take advantage of these tools, and to build an effective customer engagement strategy around them.
Adobe Sensei: Achieving content marketing at scale with AI
One of the greatest benefits of machine learning is its ability to more deeply understand content.
Adobe Sensei already features many of the capabilities marketers need to optimize content management and delivery.
From a curation perspective, Adobe’s UGC platform Livefyre uses Sensei’s computer vision models to filter UGC based on what’s inside of the media, and not just what’s in the text description, which is often misused by spammers. When you combine keyword queries with smart filtering and automated rules, you get super accurate curation results, significantly reducing the time spent moderating noisy content. Additionally, the auto-tagging functionality significantly improves content discovery and management in your DAM.
A new feature that is currently in R&D within Livefyre focuses on custom classifiers. Custom classifiers allow you to train the image recognition models to recognize vital proprietary assets such as logos, products, or specific VIPs for your brand. The ability to automatically recognize these assets enables you to create shoppable UGC experiences at scale, as they automate the process of matching UGC with specific SKUs in a product catalog.
From an experience-creation perspective, in the latest version of Adobe Experience Manager we’ve also introduced experience fragments which, in conjunction with machine learning, let you create different experiences for different channels in one place, without the need to write code. This feature facilitates the delivery of personalized experiences, and should ultimately improve customer engagement and conversion.
From a content delivery and optimization perspective, the integration of Adobe Target, an AI-powered optimization platform, and Livefyre allows you to dynamically A/B test UGC experiences in your site that can be optimized on specific engagement KPIs. It also allows you to personalize these experiences for specific audience segments, so that each segment gets hyper-relevant content to their interests or behaviors. Pairing a complete optimization engine like Target with a constant flow of high-quality UGC unlocks authentic personalization at scale.
The future of content creation
Combining machine-learning marketing technology with UGC alleviates many of the scale challenges modern marketers face today. The ability to deliver personalized and optimized digital experiences with authentic content has never been easier. What used to take significant human and financial resources to accomplish can now be relegated in part to AI-powered systems that automate manual execution tasks. In an era where your company needs to exceed your customers’ ever-increasing expectations, machine learning allows you not only to scale your content, but to move at the speed your customers expect.