On the surface, lexical analysis seems like something only a literary academic would love. It breaks down textual content into its basic components–e.g., verbs, nouns, adverbs, adjectives, or compounds–and provides an analysis of the usage of those components. While interesting, marketers and writers might see no obvious practical application for their trades.
Recent research at Stony Brook University shows that lexical analysis can be used to predict the success of literary works. The research performed a lexical analysis on more than 5,200 books across eight genres. The goal was to determine whether that analysis could discriminate between successful and unsuccessful books. The result: The analysis identified the successful books with 84% accuracy.
Now, what if you could use a similar type of analysis on, say, marketing copy on a web page or a blog post to predict how well it engages readers? Combined with standard web analytics and elements of sentiment analysis, you can build predictive models for web copy based on lexical analysis. For example, you could look for correlations with engagement metrics such as time spent on a page or exit rates for web a page. This is similar to what the Stony Brook researchers did.
Sentiment analysis extracts phrases and words from text and categorizes them into different sentiments, typically strong or weak positive, neutral, strong or weak negative, major problems, and minor problems. With lexical analysis, you can use the sentiment counts to look for relationships with the engagement metrics. A page with 15% strong positive sentiments might get 30% exit whereas a page with 7% strong positive sentiments might get a 65% exit rate.
Once you have identified examples of successful marketing copy, blog post, or other categories of online text, you can then start to build a model based on each item’s lexical makeup and associated sentiment that you can either use to revise poorer performing text or create new text with a higher chance of success.
MimoLex from Mimosasoft has combined lexical and sentiment analysis with web analytics to help create such models. Here is one example of how the process using MimoLex might work: Marketers spend a great deal of time revising promotional copy and product descriptions, testing again and again to find the wording that maximizes conversion rates and consumer engagement. MimoLex can streamline this process by helping marketers create models for copy that is closest to the mark.