“If consumers say something bad we don’t know about, it will eventually impact the business negatively.”
At a global consumer electronics firm (top secret), their ML product lead is building a central nervous system for online feedback.
“Reviews influence consumer choice heavily. We’re making a web based platform to analyse ALL reviews.”
I find it interesting that the demands to be heard - treated independently - have fuelled the automation of listening. 65K+ products, millions of comments, many countries. NLP came to the rescue.
“Take all reviews from fragmented review boards, create a database. For each product, we have key characteristics. A shaver has 200 characteristics like cutting, battery, etc. Insights extracted to create interactive dashboards.”
He explains NLP is “actually quite complicated” and not ‘just because Maths’. He tells me sarcasm is tough.
“Sometimes consumer write poems! We need to do contextual analysis to see if they’re actually happy. Sometimes people are happy and unhappy!”
Different teams are interested.
Research? Performance, NPD.
Marketing? Pricing, sentiment.
“There’s almost a use case per department.”
“We pretty much now have our finger across consumer voice.”
Consumer feedback as a sense, always-on.