The explosion of consumer reviews on the internet has provided many businesses with a much greater insight into what their customers think and feel about them before, during and after they engage with a product or service. But how can businesses, and those who market for them, best interpret this vast pool of commentary to get a clearer insight of the overall sentiment?
Numerous tools exist to assess the overall strength of a target audience’s perception of a brand. For years, marketers have used a range of traditional brand perception and promoter scoring mechanisms to gain insights into customer experience measures and the strengths and weaknesses of specific brands.
More recently, measurements of consumer sentiment have become popular with marketers looking for another way to understand what consumers think of their organisations, brands, products and services — essentially, the ‘user experience’. This has led to the development of the Net Sentiment Score (NSS) and the natural language processing (NLP) behind it.
The advantages of the NSS process
The concept of consumer sentiment opens a window into the minds of consumers: how they perceive, experience and interact with goods and services. An NSS can provide an organisation with real-time, rapid feedback on which parts of its brand or services are loved, hated or not being talked about at all.
However, until recently, assessing sentiment was a largely qualitative and selective process and open to interpretation. Trying to assess the huge volume of reviews also proved difficult and time-consuming, and the often-varied nature of these reviews made it hard to identify and summarise consistent trends in the commentary.
This has now changed, thanks to the use of the machine learning algorithms inherent in NLP. L.E.K. Consulting uses these in its proprietary NLP solution to analyse reviews and assess consumers’ sentiments as expressed in their language, making the millions of customer reviews publicly available on the internet much more accessible as a source for NSS ratings. The process can be applied to a range of organisations — from consumer-facing fast-moving consumer goods (FMCG) businesses through to service providers like hospitals, government departments and financial institutions.
How does NSS process work in practice?
L.E.K.’s process uses algorithms to qualitatively score the reviews, opinions and commentary of consumers. The analysis examines the language used in reviews to determine how positive or negative a review is, alongside the linguistic ‘drivers’ of the review, to better understand how reviewers are commenting on their experience and assign a negative or positive sentiment score to each review.
The NLP approach can examine very large quantities of reviews and commentaries from a range of websites and apps, delivering real insights and a clear interpretation of trends from these often varied and unstructured reviews. Essentially, the process is to:
Define what the organisation needs in order to understand the users’ sentiments regarding their experience and determine the right question set to use in the data analysis process
Source the reviews and other data and run the algorithmic process using thousands of individual data points from a wide range of sources
Determine a sentiment score and identify any overarching linguistic drivers of the reviews; give a summary of findings
The algorithms also allow for bias in these reviews through their ability to examine the language used to determine whether the review is a genuine experience, or a more generalised comment on the product or service. It can also monitor reviews as they change over time, to identify any corresponding sentiment changes along the way.
The entire process can be very quick — faster than other traditional brand perception measures — and can provide real insights into brand or organisational strategy as well as the impact that consumer feedback should have on these key business drivers.
For a more detailed case study on how NSS can be applied, see below.