Explore - Experience - Excel

Adding value to your ad campaign through meta data analysis of Twitter – Dr. Shampa Nandi

June 15th, 2021.

Advertisement plays a paramount role in brand promotion and is a prime instrument to reach the target audience. According to American Marketing Association, Advertising is a paid form of non-personal presentation and promotion of goods, services and ideas by an identified sponsor. Advertising has evolved and transformed with time- started in an ancient time where seller used to shout and inform people about their products/services to the very recent transformation to digital marketing and social media advertising. Statista projected that by 2021 more than 50% of the total worldwide expenditure on advertisements will be generated from internet ad rather than traditional advertisements. Advertising in the digital territories started its baby steps when Google initiated native google ad in 2005 and Facebook introduced Facebook ads in 2007. Though technically, the dynamic ad was first introduced by Double Click in 2000 which was acquired by Google later. 

The digital platform is cluttered now with Facebook, Twitter, Google, Instagram, YouTube Linked In and many more social networking sites to place advertisements. Among them, one of the most popular microblogging sites is Twitter where viewers post their short messages called Tweets. There is a huge scope for the Marketers to take the advantage of Consumers’ feedback on any particular product, service, or brand after analysing the sentiments of tweets or retweets. Along with sentiment analysis, the metadata analysis of Twitter gives lots of insights about any viewer. Metadata analysis is a very cost-effective and quick way to gather information on the popularity of a media before posting an ad, gain viewers’ demographic profile, measuring the popularity of celebrities or brand endorsers, identifying trend or trending brand or event. Twitter metadata analysis has become an essential tool of social media analytics. Twitter hashtags help in following a conversation and Twitter API is easily accessible. Short and crisp messages in the form of tweets made Twitter data a very popular aid in data analysis in the domain of social media. This article will restrict its discussion on the application of Twitter metadata on enhancing the impact of an ad campaign. 

There are 90 columns available on any extracted tweet on which we can get information on the user who tweeted the text. These are called metadata. Some of the important metadata generally considered for analysis are-

  • Unique id, Screen name- Basic information on users
  • Timestamp- information on when the tweet is posted
  • Hashtag- Basic information on tweet and discussion theme or topic
  • Followers- who follow a particular tweet
  • Retweet- trending tweets show the popularity of the tweet
  • Latitude and Longitude- location from where the tweet is posted

How the metadata can help to build an effective ad campaign is discussed in this article with an example. A company is planning to create and launch an ad for a consumer product. Here the product is Air Fryer, a very popular and trendy kitchen appliance. This product helps in preparing fried snacks without using oil and has become a very common gadget among health-conscious people.

The most common and feasible ways to use various types of Metadata analyses of a Tweet to promote any product/brand are discussed below:  

Targeted Promotion- Advertisement or brand promotion makes more sense if it is targeted to the right people. 

Extract tweets on R-studio –> save it as a .csv file in the designated location –> Read file in python Jupyter file –> Get most frequent tweets.

Rtweet is a package based on R-script, and it is easier than Tweeypy as anyone having a Twitter account can scrape tweets up to 18000 in one shot. In Tweeypy, a python-based package, one has to be a Twitter developer to get approval to scrape tweets. In R-Studio “search_tweets” is the command used to get the tweets and ‘Screen_name’ will fetch the data on the people who have tweeted on ‘Air Fryer’.

Steps to be followed:

Popular Media using followers’ counts- Before placing an ad on any visual media an advertiser can make use of followers’ count to find out the most popular media. Taking the same example of ‘Air Fryer’, a Television news channel could be a good media to place the campaign as targeted customers are health-conscious educated men or women who are regular viewers of news. In case the company wants to place ad in aa commercial channel the same steps have to repeated. Identifying the most popular media through the maximum number of followers count help to generate maximum visibility to the ad campaign. The function called “lookup_users” is used in R-studio to get the users of all news channels like NDTV, or Republic tv. The exact name of the news channel should be given while extracting the information on channel users. In python, ‘Screen_name’ and ‘followers_count’ are used to get the number of followers of any news channel.

An advertiser should select the channel with a maximum number of followers to ensure the maximum reach of the advertisement.

Identifying influencer and tag your ad- Followers count help to identify influencers in social media and company can tag their ad campaign to them to get more visibility. For example, to endorse your brand of Air Fryer by the most popular cricketer, again folllowers_count helps.  

In the above example, it is observed that Virat Kohli is the most popular cricketer in recent time with a maximum number of followers among all the other popular cricketers namely Dhoni, Raina and Rohit Sharma.

Promotion of brand through popular retweets and creating brand message following tweets- Retweet_counts is another metadata that helps in the promotion. Retweet helps in identifying the trending topics as people tweet on some controversial incident or recent incident. By tagging a popular tweet, a brand can be positioned and promoted. Interesting content in an ad campaign builds a pool of followers and continuous retweet helps to generate brand awareness and brand recall. The most popular retweet message should be used in creating a brand message. As our product is targeted to the health-conscious people, they might be a follower of “Fitness”. Therefore, extracting retweets on “Fitness” helps in finding the most trending tweet on ‘Fitness’ as well as the most attractive post on “Fitness” can be found out.  

Once of the most trending messages from the above retweets is captured, the same can be used in two ways- message creation and tag the post to promote the brand of “Air Fryer”. For example, the most popular message in a tweet on Fitness was extracted and the message is “There is no alternative to exercise and one should exercise to avoid all bad habits – as fitness cleans your mind soul and body”. It is retweeted 1070 times and a good post to tag.

Conclusion– Metadata analyses of Tweets are very easy to conduct and often come as handy tools for gaining insights into viewers’ attitude and create suitable ad campaigns. Time series analysis on the times a brand appeared in tweets over a period of time is used to measure Brand Salience. Here the metadata used is “created_at”. Brand salience is the presence of any brand in the consumer mind at the time of purchasing a product or service. Location of the tweet is another metadata that helps in identifying the presence of any product/idea in a particular geographic location. Along with metadata analysis, Sentiment analysis is the most widely used technique done on tweets to gain an in-depth understanding of the sentiment of customers or messengers.  


Leave a Reply

Your email address will not be published. Required fields are marked *