If Timeline entries are selected to flow into CX Center, the algorithm currently adds keywords to CX Center if they are frequently mentioned, despite not having those keywords configured in CX Center settings.
This in particular affects the word map displaying sentiment against certain key words.
One example from our instance is that in one customer’s timeline, there are 25 emails logged that have the word “Ryan” in it because that happens to be our point-of-contact’s first name. Because of the frequency that the word Ryan is used - CX Center thinks that it is an important word to add as a Keyword, which is great in theory, however problematic when it can’t manually be ignored/omitted.
My recommendation would be to implement one of two things (or both):
- Some logic in the Keyword configuration to omit specific words from influencing CX Center AI
- Updates to the algorithm which specifically look at Timeline entries and, in the case of emails, ignores salutations and signatures. Perhaps only looking at word groupings longer than 2-3 words? I’m not sure how this would be accomplished specifically.
Hi Alex, (@TurkOfAges ) thanks for the feedback; this is a great suggestion.
This capability hasn’t been scoped in our near term roadmap. We can pick this up based on the number of asks, priority and alignment. Do keep a tab on this thread for more updates around this.