By William Nee
Senior Director, Market Intelligence
I. Can Twitter Help Marketers?
How should we, as marketers, view Twitter? The vast majority of the information it contains is often irrelevant to us. For example, one of the most popular Tweets of 2018 was a video of an iguana falling off a table that was viewed 26 million times!
Shifting through Twitter data may therefore not seem to be worth the effort. And yet, the sheer volume of Tweets (500 million per day), as well as Twitter’s influence on business, government, and indeed society, make it an attractive source of consumer data that could yield many potential insights to marketers. The primary challenge is sifting through Twitter’s immense repository of information and identifying and extracting what’s relevant to us.
Traditionally, as marketing professionals, when we’ve looked at potential data sources (e.g. marketing lists, consumer behavior data, etc.) we’ve rightfully focused on extracting value from the data itself. But Twitter is different in that it’s the metadata associated with the content that often has the most value, not the content. Additionally, analyzing trend information associated with the sheer cumulative mass of relevant tweets can provide us with much useful information.
II. What Twitter Can Tell Us
So what can Twitter tell marketers? Well, it turns out to be plenty. For example, the popularity of brands, products, trends, competitors, etc. can be found by analyzing the number and types of tweets that mention them. In fact, aggregate customer attitudes towards a brand or trend can be determined by running sentiment analysis algorithms on the tweets that mention them.
For example, suppose we’re a marketing manager for a particular brand of sneakers sold by a U.S. shoe manufacturer. An analysis of Twitter data tells us that the number of tweets that mention our company in Texas has been flat for the past few months. However, tweets that mention our brand have soared in the past week. Ordinarily, this might be good news, but by running sentiment analysis algorithms against tweets from Texas, we discover that 60% of those reflected negative sentiment. We have a suspicion that these negative attitudes might be due to the recent price increase that we implemented, and can get confirmation or refutation of this hypothesis by running a query to see how many tweets mention both your brand and “cost” or “price” in the same Tweet. Since Tweets often indicate intent rather than an actual purchase, by acting quickly we could perhaps forestall any negative material impact on our brand.
In another example, say we’ve released a new brand of sneakers endorsed by a popular NBA player from Turkey who plays for the Celtics. As we would expect, the number of positive tweets based on our sentiment analysis are high in both Boston and Ankara. But in something of a surprise, the number of positive Tweets originating from Berlin, London, Paris and Amsterdam are also high, something we can, upon reflection, possibly attribute to the large Turkish expatriate community in each of those Western European capitals. We could therefore increase our marketing efforts in those cities to capitalize on our spokesman’s localized popularity.
III. Using Third-party Tools to Analyze Twitter Data
Practically speaking, there are two ways we can analyze large amounts of Twitter data. The first is through the use of a third-party tool, and the second is through utilizing the Twitter API. Let’s start by taking a look at the third-party option.
There are many tools available on the market today that have a social media analytics component, with just a few of the more well-known ones being SocialInsider, Tweet Binder, Hootsuite and Sprout Social – the list goes on and on.
These tools weren’t specifically designed to be marketing tools per se, since their intended function is much broader – they’re primary function is to act as a platform where an organization can monitor and control the entirety of their social presence. For example, a common function seen throughout all these tools is the ability to schedule posts to all social media platforms so that they all appear simultaneously.
But most have a social analytics capability in one form or the other, either in the base product, a purchased option or through integration with third-party solutions. For example, HootSuite offers its paid subscribers the ability to view dashboards, collect metrics and generate custom reports that allow its customers to track their social media presence. Additionally, if a customer feels that it needs greater analytic capabilities, it supports integration to other solutions like Google Analytics. A full comparison of social media management platforms goes beyond the scope of this article
If you already have access to one of these tools through licenses purchased as part of your company’s social media effectiveness and monitoring program, you should obviously first check it out to see if any of those meet your needs.
But if you don’t have a license you can leverage or feel that the solution you have access to isn’t able to meet your needs, then I’d like you to consider another option, and that’s building you own custom program to extract and analyze Twitter data through the Twitter API.
As you may know, API stands for “Application Programming Interface”. APIs are essentially a programmatic way to interface with an application in the same manner as an end user does via its user interface. They’re generally the most powerful and flexible way to retrieve information from any application, and in fact, this is the mechanism with which all social media management solutions obtain their data from platforms like Twitter.
Writing a custom program to retrieve information through an API may seem somewhat daunting since it does require some fundamental technical skills. But the work involved is offset by this solution’s low cost (effectively zero) and the fact that you’ll be getting an application that’s specifically tailored to your needs. In the 2nd half of this article we’ll walk through a concrete example to see how easy it actually is to write your own program using the Twitter API. We’ll find that most data retrievals require just a few lines of code; indeed, most social media management vendors would probably not want you to discover how easy writing your own program is!
William Nee is a Senior Director of Market Intelligence at Oracle Corporation. In this role, he performs analyses on a variety of software industry topics for his company’s leadership, including extracting competitive insights by applying machine learning against social media Big Data. Previously, he held senior competitive intelligence roles at Oracle and SAP, where among other functions, he performed detailed functional and market analyses of an array of enterprise tech vendors.