The Oxford dictionary defines marketing as, “the action or business of promoting and selling products or services.” Implicit in that definition is that this business of marketing, of promoting and selling products, is made possible through the effective use of communication.
While communication through images and video is important in the world of digital marketing, the success of most digital marketing initiatives hangs on strategically using written language.
As the overall size of the internet sails beyond 1 million exabytes (10^24 bytes), the need for organizations to process written language at a massive scale has never been more pronounced than it is today.
This is where software-based natural language processing (NLP) has changed the game.
NLP is a branch of computer science that allows computers to understand, interpret, manipulate—and in some situations, produce—human language. There are a variety of methods, software development kits, APIs and approaches used to achieve this.
Put in its simplest form, NLP allows a computer to ingest and understand in seconds the same amount of written language that would take a human being years to read, comprehend and categorize.
NLP has left its mark on nearly every aspect of digital marketing, and while this primer is not entirely complete, below is a faithful representation of the ways in which NLP impacts the work of a digital marketer on a day-to-day basis.
Search engines like Google heavily incorporate NLP methods in two basic ways:
The goal of a search engine is to enrich the understanding of both search queries (what people search for) and search engine results (the web sites and content that Google returns to the searcher). This allows Google to consistently produce the best possible answer for any search.
What does that mean for marketers?
Search engine marketers still need to do keyword research to determine which keywords matter for potential customers, and they still should use a deliberate approach to creating new content and optimizing existing content around keywords and keyword themes.
However, the SEO industry is on the verge of a pretty dramatic change, caused by the application of extremely sophisticated NLP methods to search engine algorithms. The net effect of the change is that exact keyword use matters less today than it ever has in the past.
In the new world of on-site optimization, the only thing that really matters is doing the basics and creating original, useful and engaging content that serves potential customers. The challenge for search marketers today is to understand the customer journey better than their competition and to create, measure, and improve content over time that aligns with that journey.
There are no durable shortcuts, hacks or tricks left to exploit. The heavy use of NLP in search engine algorithms is a nail in the coffin for black hat SEO—brands must now be useful to searchers, or they will not be found.
NLP can allow marketers in charge of a large digital web presence to better understand the effectiveness of web content at scale by enriching each piece of content programmatically. NLP can look for trends in content based on attributes that the computer model gathers from the content with high degrees of accuracy. Some examples of those attributes include:
The richness that can be added by NLP approaches allows marketers to understand what types of language lead to success metrics like time on page, page abandonment rate and conversion rate. It allows marketers to understand what thousands (or millions) of pages of content are about and find insightful relationships between language and marketing success using a quantitative method—no more educated guesses.
Search engines and social media platforms are financially motivated to serve ads that generate the most engagement. The thought is the ads that receive clicks, visits, and purchases are helpful, preserving a good user experience and creating revenue from ad clicks in pay per click formats.
To determine the relative quality of an advertisement or landing page, paid search platforms rely increasingly on sophisticated NLP modeling to better predict that the content of an ad, landing page, and offer will truly be relevant to a user based on their demographics, search behavior, interest category, or some combination of these things.
Google Ads is the platform that most explicitly uses this type of modeling. Google shows advertisers a quality score for every ad run in their auction, which is a composite score generated by NLP modeling of content and weighted scoring of the actual behavior of users interacting with ads (mainly click through rate, time on landing page, and abandonment).
The Google Ads auction gives advertisers with high quality scores two advantages in their marketplace:
Facebook and Instagram use a component of content quality and user engagement to also incentivize advertisers to produce great ads by allowing high quality ads greater reach and lower effective cost per impression, though they are less explicit in publishing how exactly the quality of an ad set is determined.
Another interesting, common use case of NLP for the digital marketer is computer processing of call transcriptions from call tracking and call recording software. As digital marketing becomes more and more accountable to creating revenue, call tracking software has become almost universal in the field.
Call tracking allows digital marketers to understand the amount of call volume being driven by digital efforts, and now that those calls can be transcribed to text by a computer, it allows digital marketers to model the text-based version of those conversations.
Some use cases for this kind of modeling are:
Some incredibly complicated modeling is now being done completely behind the scenes by the big players in the call tracking world.
The reality of all of this advancement? It is possible for today’s digital marketer to extract impactful data from massive amounts of call minutes in just a few clicks, thanks to this application of NLP.
An incredibly exciting emerging use case for NLP is unmanned or artificial intelligence chat and SMS text communications. In either medium, NLP provides a mechanism for a computer to understand the request of an end user and to select the most appropriate reply to guide that user to a desired outcome.
The sophistication and “intelligence” of these platforms and solutions varies, but there are solutions that exist today that actively listen to end user input and supply an answer conversationally. The systems have the ability to learn the most effective responses to language patterns on their own, sharpening their usefulness and effectiveness autonomously over time.
Another interesting advantage of even simple unmanned solutions is that the memory of a computer is nearly perfect. Even a simple text bot has the ability to remember all of the prior conversations with an end user with nearly perfect clarity. It can then produce information from those past conversations or from related systems in fractions of a second.
The computer can choose when to produce the details, allowing the bot to move conversations forward effectively and efficiently, which creates obvious operational and marketing advantages.
While not nearly complete or exhaustive, this list outlines the absolutely gigantic influence that NLP has on some of the most common digital marketing functions.
Despite being a “behind the scenes” technology for many marketers, having an understanding of the mechanism, influence, and implication of these modeling techniques on day-to-day digital marketing is critical for creating effective strategies today and in the future.
The influence of NLP on marketing will only increase—which is good news. As this technology becomes more advanced and accessible through API and SDK from organizations Like Google and IBM, NLP stands to make digital marketing more useful and efficient for customers and more effective for any organization using these technologies strategically.