Content is insights! Expressed in myriad forms and styles, it is information that makes readers feel, think and change their lives. Latest advances in AI such as Natural Language Generation (NLG) have enabled companies to automate routing content generation/automation tasks. We explained all related terms and technology:
What is content marketing?
Content marketing is one of the most effective strategies for inbound and digital marketing. Content raises awareness of your brand, builds trust with your audience and turns them into customers who may become product evangelists. Content marketing is related to both your company’s own website and blog as well as social media platforms such as Twitter, Facebook, Pinterest which have audiences that are exploring new content on a daily basis.
What is the content experience?
Content marketing process is creating and distributing the content and improving the content experience. Improving content experience can be more important than improving content quality.
For example; I like swimming. If you take me to swimming on a snowy day, probably it would not be the best swimming experience I have ever had (I know there are Nordic people swimming in ice-cold waters, I am not one of them).
Therefore, the channel, time, frequency or rather content distribution are also as important as the content. There are many software tools that can help marketers reach the right customer via the right channel with the right message at the right time such as Nudge (helps identify which accounts to contact) and Crystal Knows (helps personalize your message taking into account the recipient’s personality).
Here we focus on creating the content but feel free to examine our AI in marketing section to explore other parts of the content experience.
What are AI uses cases in content marketing?
AI can assist organizations in content ideation by facilitating content creation, curation and distribution, and analysis of the content’s performance.
Identifying topics to write about
Success of a piece of content relies on the interest in the topic and how existing content satisfies that interest. For example, both of these values can be estimated using free tools like Google Search Console for the web. It is a relatively easy and automatable task to come up with adjacent topics to write about, analyze their potential and pick the topics with the most potential.
Coming up with adjacent topic may seem difficult, however even relatively simple word embedding models trained on Google search console data can identify similar words based on how frequently they appear together. Word embedding is one of Natural Language Processing (NLP) techniques
With the help of NLP and natural language generation (NLG) technologies, organizations can fasten the process of content creation. Though content creation bots are not equipped to create complete emotionally moving blog articles yet, it can create content such as text, image, and sound.
Proofreading content that is written by a human is another application area of AI in content marketing. For example, Grammarly is a sophisticated artificial intelligence product built to analyze sentences written in English and corrects user when it identifies grammar mistakes.
AI algorithms can be used to track individuals’ behavioral patterns and their preferences regarding content consumption. These insights can be used to personalize content.
Real-Time SEO Recommendations
There are tools that claim to increase an article’s chances of getting ranked higher on search engines. We are quite sceptical about the current products on the market that tend to use simplistic approaches such as keyword counting but this is an area where a sophisticated solution could make an impact.
Content that remains unchanged quickly loses value. Auto-updating can take a significant burden from content producers.
How can content generation be automated?
Content creation is still mostly manual. Getting started from scratch, creating innovative, effective and original content is a hard job. There are many aspects that should be considered during content creation such as SEO (Search Engine Optimization), style, uniqueness and content objectives.
Different types of content such as visual, audio and text can be automatically created by AI algorithms. Machine learning and deep learning techniques search, analyze and learn from articles that are related to a given keyword to create unique and optimized content. However, most machine generated content will be perceived as derivative, may not fit the style of your organization or serve your content goals. Therefore, humans in most cases either need to quality check or build upon the content auto generated by machines.
The subbranch of AI that deals with text generation is called Natural Language Generation (NLG). We list several companies offering NLG services below. Since we are also in the business of preparing text research, you could ask if we leverage these companies’ services. The short answer is No. When we tested these solutions, we couldn’t find a solution that could prepare B2B-ready, concise, data-driven, well-researched articles. They couldn’t even serve as a starting point, deleting their text and restarting on our own was faster.
However, in B2C as well as internal documents, the requirements are different. We have seen testimonials from B2C companies using such tools. Additionally, creating reports from data is straightforward and companies like Quill and Wordsmith offer promising tools for that. However, our articles are based on more than simple spreadsheets. We rely on interviews, expert insights and industry analysis so we couldn’t use those tools.
Articoolo is an automated tool for article generation. For the full version and other features like SEO optimization, subscription is necessary.
Automated Insight’s Wordsmith turns documents into text. When you connect Wordsmith to Excel, Word or Tableau document, it produces press-ready concise verbal summaries of the data from the underlying data.
Wordsmith is already creating content for large customers:
- 4K company earnings reports/quarter for the Associated Press
- 50K personalized narratives/week for GreatCall
- 100K workout recaps/week for bodybuilding.com
Wordsmith operates as a SaaS platform where you can get 500 outputs/month for $24k/year. With increased volume, subscription prices fall, enabling large platforms to produce content for just cents.
Narrative Science’s Quill is a similar content generation service. Quill is sold as a managed service with prices ~$10K/month for most use cases.
In 2017, Google succeeded in creating meaningful images from sentences. You can watch the video below, or read the article to learn how it works.
Deep Dream Generator combines two given images to transfer the style of one image onto another. Such filters are now common place in social media applications. Here are some examples created by Deep Dream Generator:
You can see the original images at Deep Dream Generator. (You need to sign up first)
Reading content is a relatively easy task and computers have been reading for years. Google’s duplex system even introduced human-like pauses and filler sounds and words in its speech.
Some artists also claim to be using AI algorithms to create songs however these are more novelties rather than mainstream entertainment. In the song below, the music was claimed to be composed with artificial intelligence while the lyrics were written by Taryn.
What is content automation?
Content automation is more than just automated content generation. It means automating every stage of the content life cycle.
Content is made up of components like images, text, graphs. These components or the data underlying them can be used in different documents and formats. When there is a change, it is time-consuming to change every piece of content in every document by copy/paste. There are marketing technologies that help brand managers, marketing managers or product marketing managers to keep their content up-to-date.
Quark, for example, automatically updates every document, content, and component. It also automatically adjusts the format for different devices to ensure that each content is delivered to right customer in a right way to improve the customer experience.
Überflip aggregates the contents and uses artificial intelligence and intent data to predict, recommend, and automate personalized content experiences.
Why rely on content automation?
Automating content has advantages for content management compared to traditional content marketing. It can
- Increase efficiency of content creation and management
- Provide workflow automation and reduce manual mistakes
- Improve customer experience
- Provide regulatory compliance
What are the best practices of content automation?
- Caution is advised while investing in this space: Even though some vendors are making big claims about the capabilities of their tools, we have not yet come across a content automation tool compelling enough for us to use. It is still not possible for a bot to write popular, engaging content without any human guidance.
- Even the top content automation tool of today can be augmented by human effort. Even with state of the art technology, full or significant automation of content is not possible. Companies should focus on finding tools that augment their teams. These could be proofreading tools or auto content creators that create articles that are later improved by humans. Proofreading tools are already quite helpful since they provide automated plagiarism checks, engagement level gauging, and tonality assessments.
Happy to help if you have questions related to NLG or content automation: