By Brynn Smith-Raska - October 15th, 2015

Article Event Banner

"These quants need to watch Pixar’s 'Inside Out.'"

Tony Bodoh is CEO of Tony Bodoh International, LLC. He is responsible for building analytics programs for many of the biggest companies in the country. We recently sat down with Tony to get his advice and hear his personal experiences with text analytics.

Brynn Smith-Raska: Is financing still a concern for analytics programs or is executive buy-in more of the norm?

Tony Bodoh: While there is more funding available, I think financing is still a concern from two different perspectives

1) Executives are willing to fund a limited number of projects for experimental purposes (i.e. generating leads from text-mined comments on social media) but they do not have the funding for operational support for those projects. Often the project fizzle out after much hype because they are unsustainable; and,

2) Many companies put their money in the technology but fail to adequately fund team building and training. They seem to think the technology is push-button and easy enough for anyone to do. I have clients take customer service agents off the phone and put them in analysis roles side-by-side with PhDs who conduct market research and user testing.

These former agents have lots of knowledge related to solving a single customer’s problem, but they have limited perspective on the strategic approach to analyze the customer base, much less, the research background to provide recommendations that have statistical validity. Without adequate funding for proper team-building and training, these programs will struggle.

Brynn: Has data volume become too large? Can systems handle what’s there or is narrowing the focus difficult?

Tony: I don’t think data has become too large. I do think the mentality that “we have to use all data” is the real challenge. Often clients want to capture all of their data and build a data model without first doing some experimentation to see what types of things they can learn with a narrow focus.

Those who start with the experimental approach generally realize that much of their success comes from a limited data set. Over time, they can build a robust model that integrates most of the data in a proven approach. If they go after too much data, they can brag about how much they have captured, but likely have little to show for its usefulness.


Incite “State of the Industry” Survey, June 2015 

Brynn: Are off-the-shelf solutions ready for the market or is in-house development still the way to go for proper text analytics?

Tony: From a customer experience analysis perspective, the off-the-shelf products have come a long way and I believe are very useful for the text mining components when focused on mining a comment or a document. I have not seen enough practical, operational success with clients who have internally developed solutions to make it worth recommending.

However, the off-the-shelf solutions need serious work in the area of integrating data structured data and a data architecture that is more aligned with the traditional BI tools that are focused on structured data analysis. And, the text mining tools NEED to go beyond the comment to incorporate the conversation across time and the conversations in a customer’s life cycle.

Brynn: Do you still face challenges in hiring and forming an analytics team? Have skillsets changed as technology eases (automates) the creation of metrics and data modeling?

Tony: We face challenges in the customer experience arena. Many leaders underestimate the skillset required and the time it takes to build the skillset.

Some believe text analysis should follow traditional BI and be quantitative so they focus on hiring those skills. They ignore that the comments they are analyzing are really often expressions of intense emotions. Quants focus on reducing emotion to a sentiment score so they can use one number to show improvement. They fail to understand the complexity of human emotion and the various qualities of human emotion. Unpleasant emotions are not necessarily a bad thing for people to feel and it does not always mean that the customer will leave. These quants need to watch Pixar’s “Inside Out.”

On the other hand, if you have a qualitative analyst, they often are uncomfortable when they have to cross into the quantitative territory. They will gladly tell you how a customer feels, but they hesitate to make any estimation about the ROI of fixing a problem or creating a positive experience.

I think the real answer is a dose of quant and qual in the same person. People with a background in experimental behavioral psychology or behavioral economics are great candidates. I believe that over time it will become easier due to advances in technology and with the expansion of the number of analysts with appropriate skills.

Brynn: Are there any emerging trends that you’re making use of (such as sentiment analysis)?

Tony: We have been using sentiment analysis extensively. But we go beyond that. One of the key things that helps our clients is the use of text mining to identify the conversational points that correlate with particular customer behaviors.

As an example, we helped one company more than double their sales because we helped the company redesign their marketing and sales language and process to mirror what they best buyers in the previous campaign were saying. In another case, we looked back in time at conversations customers had 3-6 months prior to terminating their relationship with our client and we identified key themes that could be used to trigger a new action path. Effectively, they could start the process of saving the customer before the customer knew they were thinking about terminating the relationship.

Using approaches like this, companies can focus on reducing the data they are harvesting to critical points that matter. But, they need the breadth of data to actually conduct the original experiments.

Brynn: Are social media or real-time analysis part of your work? What about the future?

Tony: Yes, we are currently working with clients that do real-time analysis. This will be a growth area, however I am cautious about the investment at this time. I think it is far more hype than reality. I have seen a limited number of case studies that proves this is a better and a more sustainable approach to specific solutions (i.e. customer engagement, lead generation, customer service) than other proven approaches.

Brynn: How does the customer come into play for your text analytics?

Tony: We focus exclusively on text mining customer feedback and researching the customer experience. It is unique because of the depth of emotion that we deal with.

There is a huge gap between mining non-emotional language of contracts, claims, etc. and the mining of customer conversations. What we learn from customer feedback we can then use to study the customer journey and even use to improve our focus group or other research efforts.

Brynn: Are they part of your BI and storytelling or is that separate?

Tony: Yes, because we link the customer conversation and the customer behavior (i.e. sales, cancelations, calls, etc.) we absolutely help our clients integrate the data and analysis from their BI, CRM and CEM systems in the process of analysis and storytelling.

The customer is always center and we teach our clients how to tell a story about the customer’s journey and experience rather than about the internal processes that need to be fixed. Many times the executives have never actually had the real customer experience because they don’t use the product or have expedited support due to their role in the company.

Brynn: What piece of advice would you give someone who is starting or expanding their TA program?

Tony: Think with an enterprise scope. Build deep relationships across the company. Prove yourself by analyzing the feedback related to one customer behavior at a time. That will build your skills, prove your mettle and provide you with confidence. And, the executives will love the improved metrics. 

Incite “State of the Industry” Survey, June 2015 


Tony Bodoh will be speaking at the upcoming Incite Text Analytics Summit: West. Taking place on November 5th and 6th in San Francisco, this event is the premiere gathering of the best and brightest minds from across the industry. Find out more here:

Incite Text Analytics Summit: West 2015

November 2015, Hotel Nikko, San Francisco

Learn to tie your data to business goals from over 25 of the brightest minds in text analytics. The 15th Text Analytics Summit: West will show you how.

comments powered by Disqus