Hi everybody! First time poster. I'm curious if others have run into this.
I provide KPI analytics for customer service in the social media space. There is a popular desire for text analytics, but cost prevents us from hiring big guns to provide this in a proper way. So I volunteered to tackle with homegrown efforts, and decided that Tidy Text would be the best way to enter into this world.
Fast forward. After a few weeks of late evenings, weekends, and thousands of code lines ... voila! ... I created my first text analytic report. Let me tell you, I pulled out all the stops. Coding in R - I felt like I was finally rolling with the big dogs, and off the Excel puppy porch. It had brigram breakouts, sentiment pyramids, polarity trends, Markov graphs, relative happiness by product, sentiment arcs, team sentiment boxplots, mood pies, wordclouds of negative themes, volcano plot of what's changing, competitor twitter scrapes, and other things.
In addition to learning R/Tidy Text, a goal was to tell the story of where things are not going well for our customers using their words through text analytic methodologies. So I hit them with some common breakouts and graphs you see people doing with Tidy Text.
After a handful of meetings, with all types - from Managers to fellow analysts, it's gone down like a lead balloon. There's been a complete disconnect, and people (even so-called friends!) are running for the exit.
All that work for all that silence.
When there is feedback, it's "this is interesting, but what do we do with it?".
Despite guided explanations, it's largely been a bust.
I see where some of the disconnect is. One idea i know to be true is that my crowd is conditioned to KPI analytics. What's going up and down? Who can I go beat up to improve the business? They like their data fast and hard. I ignored all of this because text analytics requires slowing down, reading, and thinking differently.
Also, even though they are asking for text analytics, they either 1) do not know what that means or 2) their expectations of it competes with what I taught myself using the internet. I assumed too much and regret not understanding my audience's expectations better. I've worked with these fine folks for years, and still made a rookie mistake.
I learned a lot and sense there's more to it and don't want to abandon it all together. I'm challenged to find what translates to actionable insights, and not just cool and interesting.
Wish me luck. But has anyone ran into something like this - specific to text analytics. Have you lost people despite using common approaches and graphs with text analytics? Just me?