Narrative analytics: An example based on banking app reviews
(Maps viewable on desktop only)
We took a small sample of 2,669 banking app reviews for 6 UK banks, and ran them through narrative analytics to understand what drives the experience of banking app users.
The maps can be built with any type of unstructured language: Research verbatims, social media, news, and more. This technique does not pre-set the categories: The axes and clusters are based on the language in the app reviews and are unique to this dataset, from trillions of possible combinations. This customized clustering helps to avoid the bias and blindspots of pre-set categories.
Every dot in the map below represents a single review. Hover over the dots with your cursor. Note the slider on the lower right of the map below. The 4-cluster view shows the high-level axes of the map. The left side of the map is about the functional deliver of the app. The right side is about the bank, and the feelings and experiences of the user.
Now move the slider to the right, to the 24 cluster view in the map above
The narrative clusters are more than functional/rational ‘topics’ – they are experiential. Clusters like “Superlative bank: Modern ideal” capture a strong experiential driver - showing that a large number of customers are experiencing a big change in their expectations. There are many ways people choose to express this experiential concept, making it difficult to measure through existing text analysis tools or topic models. Narrative analytics is unique because it can measure these seemingly abstract themes.
Finding the 'hotspots' based on the app rating score
The below map is exactly the same as the 24-cluster level of the above map. The only difference is that the dots are coloured based on the app rating score. What does it tell you about the drivers of negative vs positive experiences? When a customer or employee rating score changes, narrative analytics is a way to understand the "why". Why are people giving low scores, and why are they giving high scores? The answer is easily found in the map.
Mapping the customer experience "footprint" for each bank
The below map is also the same, with the dots coloured by the bank name. By clicking on the legend on the right side, you can select specific banks. How is the customer's app experience different for Revolut, compared to Starling?
Detractor analysis: What is driving detraction for different banks?
The bar chart below is based on the 9-cluster level (the middle level) of the top map. We isolated the angry app reviewers: the ones who gave the lowest possible rating score of 1 on the 5-point scale. The number of detractors is similar for each of the three banks in the chart below. However, the story is very different when we look at the drivers of detraction for each bank, based on the map. For HSBC, App UX is the driving 58% of detractors. For Lloyds, Bugs and gaps are driving 57% of detractors. And for N26, 62.7% of detractors' reviews are about service shortfalls.
This illustrates the simple actionability of narrative analytics. The score alone is not always actionable. When we know the score along with the narrative, we understand what to fix: quickly, easily, and clearly. In the HR space, Phrasia does the same type of analysis to understand the drivers of employee engagement, ENPS and churn (for example).