Cutting through in the era of big noise: a case study of 3 US banks
(Maps viewable on desktop only)
In an increasingly cluttered landscape, brands need new ways to detect - and go - where the ‘narrative traction’ is. It starts with hearing: How do you understand the wider context in ways that neither social media monitoring nor generative AI can reveal?
Please allow up to 90 seconds for the map to load.
Step 1: Understand the full narrative landscape - simply
Move slider to see detailed view
What are your competitors and customers actually saying in the first place? The first step is an interactive map ofthe outbound Tweets of three major US banks, along with their investment arms: Wells Fargo, Citi, and JP Morgan Chase. High level narratives like “corporate citizenship” break into subnarratives like “women winning,” “racial wealth gap,” and “supporting the arts”. The map is interactive, with each dot corresponding to a tweet which can be read by hovering with your cursor. Is it more intuitive than what you are using now?
Step 2: Find competitive overlaps and gaps
Viewing the same map with the dots coloured by brand, we can quickly see the competitive footprint for each brand, and gauge internal alignment between sub-brands. Click on the legend on the right to select/deselect specific brands. Wells Fargo bank and Wells Fargo Investments appear to be at opposite poles - with little in common. By contrast, every one of the 6 brands is tweeting about "women winning", while Citibank is all alone in the concerts space. Would this be helpful in making clear, reality-based choices on what narrative sweetspots to target, and what ‘no fly zones’ to avoid?
Step 3: Hit areas of high ‘narrative traction’ to cut through
Based on the clusters in the above map, you can see which narrative spaces generate the highest proportion of likes or shares. It varies by brand, so you may find that what gets traction for you is different versus what gets traction for your competitor. Would this help you to improve your earned media performance?
Step 4: Set a clear direction for prompt engineering
The attached map - taken from a subset of the above data - shows how prompt engineering can miss the mark. In our prompt testing, ChatGPT generated a lot of tweets about gender-based inclusion as well as high-level D&I, but misses huge swathes of the narrative landscape, including in this case the important topic of racial inclusion. Comparing the map to the generated results make it very easy to gaps like this. Random prompt engineering won’t provide this objective clarity. While your competitors are flailing by subjectively ‘experimenting’ with prompts, you will be far better informed and productive in the application of generative AI, by starting with narrative analytics. How smart would that be?
Step 5: Better correlate trends with outcomes
As Jeff Bezos said, ‘your brand is what people say about you when you aren’t in the room’. Narrative analysis makes it easy to track that grassroots reality, because the ‘categories’ that appear in a map are never pre-set by an algorithm, model, or researcher. They are unique to that dataset, built from the true conversation. (something that isn’t likely true of your brand tracker or social listening approach, wherein most of what you see is based on pre-assumed categories or guesswork). Tracking the clusters in the map over time, we have found that narrative analytics correlates with business outcomes between +40% and +120% better than traditional approaches. Would that help you to explain the value of marketing to others in the business?