Difference between Social Media Listening and NLP

What are social media listening and NLP technologies? In today’s video, we are going to define these two approaches of communication and discover what can we learn from them, their accuracy and their data sources.

Today we talk about social media listening and customer feedback analysis, or generally, NLP technologies. I get often asked about the differences between these two solutions, so I decided to make a video about it.

It is the process of identifying and assessing what is being said about a company, and its products and services on the Internet. The key data sources are social networks, such as Facebook, Twitter or Instagram…but also blogs play a role.

Social listening tools like Sprinklr, Sprout Social and Talkwalker can scan texts for specific keywords on social networks and blogs. Essentially, this type of software transposes specific words or phrases in unstructured data into numerical values.

It means that you can ask the tool to count how many times people mention a specific keyword or concept on the internet. You can also ask most tools to tell you if people mention these keywords in a positive or in a negative way.

In fact, most systems work just by looking at words in isolation, giving positive points for positive words and negative points for negative words.

You can get quantitative insights, knowing the main pros and cons of your products or brand. Which means that you still make learnings, but those would be superficial.

Generally, these social media monitoring tools do not produce very accurate results…mostly for two reasons:

-Number One: the scraping capabilities are limited, so a lot of noise is processed together with relevant information. Furthermore, matching specific products with unique URLs is complicated, and creates problems of disambiguation

-Number Two: text analysis technology is usually based on statistical algorithms, which implies that is not tailored around the corporate language, and not around the different product verticals. It shouldn’t be surprising to have results with accuracy scores below 60%, and consequently not valid from a decision-making standpoint.

As a plus point, we can say that social listening software is usually not very expensive. On top, most of these tools offer customer engagement capabilities, which make it possible for employees to interact with customers directly.

It is the process of extracting the full meaning representation from free text. For full meaning, I mean to understand who, when, where, how and why something is happening around a company, its products, and services.

The key data sources are usually divided between online and offline, and between public and private. You can think of product reviews, social media, open questions from surveys, call logs from the customer service etcetera.

NLP tools can scan texts for specific meanings on all kinds of data sources. Essentially, this type of software not only transposes keywords into quantitative information but extracts the “why” behind the messages.

Achieving such a deep understanding of a text requires linguistic configurations and customizations that are not cheap, but once they are in place the result can be as good as having humans interpreting a text.

Can you learn from this? You can get qualitative and quantitative insights, with higher resolution and actionability. You can now ask the data questions that historically required extensive market research.

In short: Social media listening quickly tells you how much, NLP gives you the background, the who, and most importantly, the why behind. So, Social listening software is to have a screenshot of the public sources fast, while NLP software is the necessary decision-making tool.