You’ve probably heard of Natural Language Processing (NLP), the process of analyzing text and extracting data. But have you heard of the inverse, Natural Language Generation (NLG)?

Natural language generation (NLG) is a software process that automatically transforms data into written narrative.

It starts with data.

The main requirement for implementing natural language generation is ownership or access to data. In order for any natural language generation software to produce human-ready narrative, the format of the content must be outlined (through templates, rules-based workflows, and intent-driven approaches) and then fed structured data from which the output is created.

For example, our Wordsmith platform accepts structured data via uploading a CSV directly, passing the data to our API as a JSON object, or through one of our integrations that connect to the API like Tableau, MicroStrategy, or Zapier.

NLG software in its current state will not pull unstructured data and magically generate written text without some human guidance.

NLG powers narrative creation in many industries.

While NLG is useful wherever there is a need to generate content from data, some of the most common implementations include:

  • Written analysis for business intelligence dashboards
  • Personalized customer communications via email and in-app messaging
  • IoT device status and maintenance reporting
  • Individual client financial portfolio summaries and updates
  • Ecommerce product descriptions and category landing page content

Narratives written with NLG are designed to read as though a human wrote each one individually. The specific insights, writing style, and structure of the narrative varies depending on the audience, as well as the context and intended purpose of the content. The output of every current NLG solution is powered by the narrative design (also referred to as “template”, “intent”, or “narrative type”), which is constructed by the end user of the NLG solution or by the provider of the software.

Within this narrative design are rules, also referred to as “conditional logic”, that trigger different outputs based on the data set behind the content. The most powerful NLG solutions allow users to edit these rules and narrative structure to best fit their needs and have those changes reflected in the output instantaneously.

This ability to customize the conditional logic that’s powering the content creation is why natural language generation is flexible enough to create data-driven narratives about so many industries-ranging from finance and business intelligence to e-commerce and sports.

Why invest in natural language generation?

Extensive Personalization at Scale

Natural language generation enables you to generate complex personalization at scale, creating improved customer communication and experience with your organization.

A financial services firm, for example, can deliver portfolio summaries to thousands of clients with each summary using the customer’s unique set of information to speak directly to the individual. In other industries, using each customer’s unique set of data can also improve user experience and boost retention rates. For example, the Orlando Magic improved communication for their loyalty ticketing program using NLG.

At a group level, narrative can be adjusted based on the role or function of the viewer as well, highlighting certain areas of focus and key metrics that matter to that specific group of people.

Make Data Understandable and Insightful

NLG also provides data experts with an efficient way to automate the “translation” process that occurs when they need to explain their findings in clear, concise ways to clients or to others within their organization who may not be data experts.

For example, one rapidly growing use of natural language generation is written analysis for business intelligence and analytics platforms. While charts and graphs are incredibly useful for depicting the health of a business, dashboards with complex visualizations can overwhelm viewers who aren’t data scientists or experts on the specific data set they are viewing. With tons of information to parse through in a dashboard, it’s hard to confidently know what matters or exactly what you need to do about it. On top of that, allowing dashboard intimidation to spread through your organization can create a lack of engagement with your analytics, causing inefficient and ill-informed decision-making.

Manually writing reports summarizing and explaining key insights from the dashboard is a daunting task for many analysts. With natural language generation, these experts are able to automate written analysis that speaks to each individual dashboard viewer through insightful commentary, without analysts spending so much of their valuable time explaining their findings.

Natural language conveys expert-level analysis and advice in concise, insightful terms that engage and fully inform each reader-regardless of data expertise. With NLG, analysts are empowered to go back to their higher ROI producing tasks while maintaining an efficient method of dashboard reporting.

NVIDIA Tableau dashboard
An example of natural language generation used in a BI dashboard to provide a narrative summary.

Our Wordsmith NLG platform, for example, is integrated with Tableau and other BI platforms so analysts can also generate custom, responsive narrative explanations of their data visualizations within a dashboard. The narrative can be customized to use a business’s unique terminology and is also responsive to dashboard viewers drilling down within their data, regenerating narrative that is pertinent to the displayed visualization.

Speed and Scale of Content Creation

With NLG, you can produce thousands of unique narratives in a fraction of the time it would take to write them manually.

For example, the process of writing hundreds or thousands of product descriptions is time-consuming and costly-whether that’s stealing valuable time from in-house writers or hiring an army of freelancers to complete the task. With natural language generation, the inventory specifications for each unique product in a catalog can be transformed into a unique, keyword-rich description.

The speed and efficiency of creating narrative with NLG is beneficial for any industry or use case that requires large amounts of unique content, whether that’s for the benefit of SEO, customer relationships, or internal communication.

Get Started with Natural Language Generation

In the past, natural language generation implementations could take several months of work by professional data scientists, software developers, and solutions architects behind the scenes. “Black Box” NLG solutions, in particular, would result in businesses investing six figures in a technology packaged as pure artificial intelligence, then not actually receive any content for several months.

Today, our Wordsmith platform makes it easy for anyone to upload data and start automating their own consumer-facing stories. Wordsmith users can also take advantage of many integrations to quickly implement NLG in their existing processes. Our services team can also assist with the implementation of NLG for a company’s specific use case.

And don’t be fooled by the nomenclature-“template” NLG is just as complex as “advanced” natural language generation. The content produced by both methods will be as intricate as the writer constructing the software logic makes it. Check out our Myths of NLG white paper for more on that topic, as well.

Request a demo of Wordsmith to begin implementing natural language generation in for your company.