Marketing efforts tend to be focused around images, videos, stories, and few people give a thought to the role of the humble full stop or comma. Alex Darmon explains why there’s more to punctuation than you might think.
Whilst measurement is a priority for almost 90% of marketers, content attribution is often a blindspot. Punctuation analysis presents a unique way to discover more about an author than was previously thought possible. Using this approach, brands are able to measure tone of voice across all channels, devices and platforms to create consistency of customer experience.
Punctuation analysis is in its early stages but is a first step in understanding how a person or brand can be identified through patterns in punctuation and logically this could be applied to language. A recent study from Fospha Marketing and the University of Oxford, into textual analysis via punctuation sequences, aimed to discover whether an author could be identified purely by the frequency and placement of their punctuation.
This technique is effectively stylometry, the quantitative study of written text for determining authorship, but without words. Punctuation was identified as a potential means of author identification because it is a non-standardised representational system that varies significantly between individuals.
Using almost 15,000 documents from The Gutenberg Project, the study extracted an ordered sequence of ten punctuation marks: full stops, commas, colons and semicolons, left and right parentheses, question marks, exclamation marks, quotation marks, and ellipses. Using machine-learning-based neural networks, six simple, punctuation-based quantitative features were identified and analysed. These included the frequency of punctuation marks within the document, their sequencing, and the number of words sitting between each mark. The results of the study indicated punctuation marks do encode stylistic information about an author, with the correct author identified in up to 72% of cases.
While the study in question focused largely on author identification, its results could have significant implications for marketing measurement. The common point between marketing and punctuation is consumer behaviour so, using machine-learning techniques to identify patterns in punctuation sequences, marketers can analyse marketing materials to determine which are driving the best consumer response as part of their multi-touch attribution (MTA) strategy. Punctuation analysis is simpler than word-based language analysis techniques currently in use, such as Natural Language Processing (NLP), and punctuation sequences are smaller and easier to store than larger word-based data sets.
When running ad campaigns, marketers frequently test different creative elements such as images, calls to action, backgrounds, and colours to determine which are driving the greatest response, and then optimise accordingly. In future they may be able to do the same with punctuation analysis, identifying which patterns and sequences are most effective at driving marketing goals across all channels. This information can then be used to maintain brand tone of voice and identity across fragmented channels, devices and platforms for a seamless customer experience.
These insights could be used to achieve a unified brand tone of voice across all campaign activity, and develop standardised punctuation patterns that work for their unique audience or marketing goals, before implementing them across marketing content in all channels.
Marketing measurement is developing rapidly and it’s about more than just the adoption of new tools and techniques. While embracing MTA models that take account of all consumer touchpoints is undoubtedly the first step, there are other complementary technologies continually emerging to help marketers measure and optimise content and brand. This research is just the start, and its future applications are very exciting.