In today’s globalized market driven economy, analytics is driving a transformational impact. Companies are using advanced data science algorithms to mine publicly available unstructured data and to drive faster and impactful decisions. It is increasingly essential that businesses draw insights from what Key Opinion Leaders (KOLs) are saying. But who are they? Key Opinion Leaders are people or organizations that have such a strong social status that their recommendations and opinions are listened to when making important decisions.
Welcome to ‘Social Listening’
Organizations are actively using advanced text mining algorithms on social media data to analyze customer preferences. Social listening is helping companies not only to derive insights from unstructured data but also in transforming those insights into customer focused strategies.
Some of the key reasons of why social listening is gaining widespread momentum are:
- Coverage of sentiments/expressions that might be missed in traditional Primary Market Research
- Huge amount of social data available that is quick to analyze and help companies take corrective actions on a real time basis
- Significantly cheaper than performing primary market research
Academicians: Gear up for a Tech. Product!
A hi-tech company recently launched a product for academicians and scientific researchers. The company knew that academicians and professors would be the early adopters of the product and their opinion would be crucial to the success of the product. Since the product is based on a new technology, the company wanted to identify key online influencers among these academicians who can influence customers.
Using Data Analytics to Identify KOLs
In order to address the problem of KOL identification, we proposed to identify social circles and top influential academicians through network analysis.
1.) Data collection :
We extracted Twitter conversation data from the period January to June 2015 with keywords that consisted mentions of product and the technology
2.) Data Cleansing:
Once relevant social data with relevant keywords was collected, the data had to be cleansed. Social data contains a lot of non-relevant content as well. So, data cleansing is an essential step that needs to be done before the data is ready for any analytics.
Typical steps for data cleansing include :
a. Noise removal by keyword searches
b. Removal of web-links, web-addresses and sentence tokenization through algorithms
3.) Author identification:
The next step involves identifying the tweets from professors and academicians. We applied academic focused stakeholder segmentation algorithm to identify professors among various other authors. We used feature generation algorithms on the Twitter metadata to segment metadata with academic specific keywords. In this way we were able to separate out tweets posted by academicians & professors vs other stakeholders like media, advertisers, etc.
Fig. 1 (Illustration)
4.) Network analysis: In order to identify online influencers, we had to develop a metric to measure and compare influence of twitter authors. We believe that retweets indicate the flow of information originally posted by the author and influence is the share of relevant content that makes an effect on people’s psyche.
Hence, capturing the Tweet-Retweets converts into capturing the flow of influence. Networks based on retweets reflect the level of endorsement one party has from other’s original thoughts. Endorsement networks make sure that only relevant (highly endorsed) information is considered.
After establishing this hypothesis, we created a network based on the retweets among the professors identified in step 3 using a proprietary network algorithm.
A visualization of the author network map is shown in Fig 2 where:
a. Each node represents a Twitter author handle
b. Size of the node is proportional to the author influence in the overall network
c. Color regions specify author communities (i.e. the subset of author more connected through frequent endorsements online). Nodes of same color represent a social community
d. Connections among the nodes define the shape of the overall network
Fig. 2 Author Network Map
Potential of Analytical Technique in other Realms of Business
Once the KOLs and their influence circles were identified, the hi-tech company was able to formulate messaging strategies (emails) focused on these KOLs and their pain points as stated in the tweets.
However, the point to be noticed is that this analytical technique of identifying social circles and influence of opinion leaders can also be applied in a number of other industries like Fashion, Pharma and Politics where consumer behavior can be heavily swayed by the key opinion leaders.
This analysis coupled with topic identification and sentiment analysis can help companies further categorize the conversation and understand the reaction of key consumers.
It is our constant endeavour to engage Business Leaders from various industries, who are serious about leaving behind a legacy. Kindly support our endeavours by using the 'Comments' section to tag people in Fashion, Pharma, and Politics, and help us reach out to them. Thank you for your kindness.