Hitmetrix - User behavior analytics & recording

Are You Feeling ABM? Analytics Corner

ABM has emerged as a terrific starting point to tailor dedicated marketing content and associated service to prospect accounts. But marketers can get even savvier in their understanding of how potential and actual clients appreciate that dedication by applying data science techniques.

One technique that is a straight-forward fit for accounts where feedback is digitized is sentiment analysis.

Sentiment analysis can help account teams identify how digital content is perceived, and make adjustments to improve the engagement rate for important accounts.

Digital engagement by an account has traditionally been tied to the activity related to the content – a click on a button to download, or page views. But drawing conclusions on the reception to that content was limited to volume metrics – the number of social media shares or downloads associated with white papers, for example.  With sentiment analysis marketers can drill a bit deeper into the feeling around the content by examining what words are generally used when the content is mentioned.

To start a sentiment analysis, marketers need to ingest the responses to be analyzed into a statistical programming language, like R programming or Python.  The responses are sourced as data, but the data can be sourced from any digital media.  For example, earlier this year I created a sentiment analysis on consumer reactions to IHOP’s faux name change to IHOB, as part of their introduction of a new burger line.  Well-known data scientists like Julia Silge pioneered the analysis by examining text from classic novels.  In my demo I used tweets imported into a R programming script, since Twitter provides fertile ground for examining sentiment on a consumer product, service, or brand.

Once uploaded into an R script via a file or API, the text is placed into a container variable called a corpus.  Punctuation and special characters are then removed so that contextual bias is minimized.

Finally the text is compared to text within a lexicon by applying a join, then visualizing the comparison.  A lexicon is a scoring list for trigger words registering positive or negative sentiment.  There are several kinds of lexicons, each with a scoring structure that cover degrees of positive or negative expression.  For example, the NRC Word-Emotion Association lexicon associates words with 10 sentiments: positive, negative, anger, anticipation, disgust, fear, joy, sadness, surprise, and trust.  Many lexicons are designed to examine one word at a time, but others can examine a phrase.  A graph representes the words that contributed most to a sentiment.  The graph typically shows how balanced or askew a sentiment is around a chosen word or phrase.

Marketers can use sentiment analysis to raise a consensus on how to approach an account.  Different lexicons and various visualization methods exist.  Within R programming, there are several libraries that provide variations of sentiment analysis. Applying sentiment analysis should create a dialogue about analysis assumptions, revealing how an account team chose its approach for understanding sentiment interaction, and how its analyses are tieds to an account’s presence online.

Sentiment analysis can also help highlight feeling towards your company, possibly related to specific client-facing initiatives or activity.  Technologies for ABM have made client-facing service more effective with a dedicated approach.  The tone of client response online can be analyzed, and even correlated against data about customer success.

While I conducted a sentiment analysis on tweets, the same technique could be applied to documents that have been digitized.  If accounts are to provide a feedback report with a description, those reports can be placed in a digital file, and then ingested into the sentiment analysis engine.

Imaginative marketers now have a means for quantifying whether an ABM initiative is contributing to positive or negative feelings, especially among target accounts.The key is to gather the best sources where accounts express their gratitude orconcerns. Analyzing those sources can help manage accounts’ levels of happiness.


Related Posts