What Content Generation Means
Simply put, content generation or automated content creation is when a computer “writes” content without needing a human author. We will focus on text-based content for the sake of this explanation, but this content could also be represented visually or audibly. Of course, the machine doesn’t just telepathically generate paragraphs of human-readable text from nothing. Instead, it is fed some form of guidance (inputs), and it has been trained on the various options for how the end-result (output) could look and feel.
From a technical perspective, content generation is Natural Language Generation (NLG), an artificial intelligence task and concept closely related to Natural Language Processing (NLP) and Natural Language Understanding (NLU). All these tasks are a part of what is considered computational linguistics – the practice of a machine understanding, processing, and generating language in the same way that humans might do so naturally.
The models and AI capabilities driving content generation today are quite intelligent but are not yet at the point (nor may we ever really want them to be) where all or most of your content can be generated purely by a machine – at least not while still providing an effective and enjoyable customer experience. But there are several high-value use cases where leveraging content generation creates major competitive advantages. This article focuses on one of the biggest: personalized micro-content.
What Micro-Content Means
The term micro-content can have a few different meanings, but the primary definition used here is small, quickly consumable pieces of content that can have a high-level of impact on the overall customer experience. It could be as small as a Call to Action (CTA), or as big as a paragraph of text convincing a prospect of why they need to Add to Cart. Or even a brief email that endears a customer to your brand and builds instant trust and brand equity. These pieces of content can be a part of, or the entirety of, a micro-moment – a special moment that is memorable or effective in advancing a customer’s journey.
Why Generate Micro-Content
There are two big business needs that justify an investment into content generation, and they happen to fall on completely different ends of the spectrum. The first is when massive scale is necessary and the need for quantity edges out quality, In that case, it is not worth the cost to manually create content for every possible product or market or language, etc. The second is when quality, or rather relevance, is far more important than quantity and it’s virtually impossible to manually create content that is personalized and relevant enough to achieve that content’s goal. This is where micro-content comes in.
The key here is not that these are just small pieces of content, but rather that they pack a punch well-beyond their weight, so to speak. If you were to measure your investment as well as the return on that investment (ROI) as dollars per word, both values would be substantially higher for these crucial pieces. If you told your CMO that generating just two sentences of content could lead to, for example, an uptick of hotel bookings by 17%, they’d surely take notice. Therefore, an investment into training or deploying a content generation model to craft a piece of micro-content for one or more very specific uses is both desirable and justifiable.
Why Generate Instead of Simply Personalize
Many times, personalizing manually created content is the right way to go. It requires less investment and is more straight-forward technically. But sometimes, to really alter content enough to achieve a very high level of relevance, context, and personalization on a 1:1 basis across all your possible customers, simply modifying a one-size-fits-all piece of content does not fit the bill. You still want a piece of content that looks like it was written by hand to maintain authenticity, but you want it written uniquely for each individual interaction. You want to automate the experience of a small-town shop owner hand-writing a note to each of their long-time regulars. This is where the concepts of personalization and content generation overlap and it’s where you can realize the greatest value of real-time generated content.
A Real Use-Case of Generated Micro-Content
When a business traveler books a flight for an upcoming trip, the next step is to find an appropriate hotel. Many find that searching their favorite hotel aggregator just leaves them with an endless list of hotels with generic descriptions that tell them a lot of irrelevant details. Descriptions become noise and don’t add-value because they are not personalized to this specific traveler or contextual to this specific trip.
An example of a regular, manually-created hotel description that a traveler would see today:
This is a lot of information – and almost none of it applies to the traveler’s needs for this trip. This might be the perfect hotel for them, but they would never know it from this description. As an example, they never rent a car when staying close to an airport because they like to get in and out quickly, yet parking is mentioned three times. They will be there for work and care about how this hotel makes their trip easier and more effective, not about poolside bars or proximity to a hospital. There is also a large map that adds no value for this traveler – they likely have no idea where that pin is or how the location relates to the airport.
As mentioned at the start of this use case, the business traveler has just booked their flight. The airline, or travel booking app, knows:
- The passengers, dates, times, and airports of each of the flights on this itinerary
- How often they travel, how often they visit this city, how long they usually stay after they arrive, whether this is business or pleasure
- Fly Economy or business class, etc.
It may even know their preferences like
- Where they like to eat
- What they care about in a hotel
- How much they typically spend
- If they rent a car on most of their trips
It can also make intelligent assumptions based on that traveler’s history as to what they really need out of a hotel for this specific trip based on models of similar traveler behaviors. When the hotel description is generated in real-time, for this specific traveler, for this specific trip, all that information and more can and should be used to deliver a better, more effective, and more enjoyable customer experience. A manually-written description, or even a personalized templated description, will not accomplish the goal here.
An example of a real-time generated hotel description:
The above description was not written by a human, it was generated in real-time. It tells the traveler not WHAT, but WHY. Why is this the right hotel for them for this trip, why should they book it right now instead of endlessly searching other listings. It correctly assumes that fast reliable wi-fi is very important and it knows that the traveler prefers steak houses to take clients out to dinner at. But perhaps most interesting, it knows what time the traveler must be back at the airport and highlights how they don’t have to checkout until its time to get an Uber back. It also shows the cost and time for an Uber ride and the location relative to the only location that means anything to a traveler – the airport they are arriving at.
Explore Content Generation Further for Micro-Content
Content generation isn’t usually helpful or a worthwhile investment for most of a brand’s content. But for specific micro-content or micro-experience pieces, it can add an incredible amount of value that really moves the needle on impacting customer behavior and positively influencing a journey. In our example, it means an airline or travel aggregator can drive conversions and a higher customer lifetime value (CLV). Running an A/B test, with the manually written content served to the control group and the generated content used as variant B, is an easy way to demonstrate ROI as well.
Develop or refine your overall personalization strategy, mapped to fulfilling your business goals, to highlight what pieces of content or experience could most benefit from this approach. Start small and very targeted in your use of this technology, constantly measure its effectiveness, take an agile approach, and work with your implementation partner to continually tune your models so micro-experiences are always improving. Content generation makes tremendous experiential power available to your brand. The key is understanding exactly how and where to apply that power.