Fake reviews in the news
Fake Reviews (1).png

We love reviews because they are therapeutic and help consumers make sense of their experiences (previous post) and give brands a ton of information that can be used to improve future customer experiences and business performance. 91% of shoppers read reviews, and 84% of people trust them (thanks Craig Bleom), which means that we must ensure that the reviews are real.  

 Thankfully, the Federal Trade Commission has been fighting fake reviews. Back in February, they announced that "the first case challenging a marketer's use of fake paid reviews on an independent retail website." Andrew Smith, Director of the FTC's Bureau of Consumer Protection. "Posting fake reviews on shopping websites or buying and selling fake followers is illegal. It undermines the marketplace, and the FTC will not tolerate it."

 Yesterday (October 21, 2019), the latest settlement with the FTC involved cosmetics brand, Sunday Riley. According to Jon Fingas with engadget, "...ordered employees to post fake reviews on Sephora's website in a bid to boost sales. Managers and Sunday Riley herself reportedly created fake accounts to post reviews between 2015 and 2017 and urged employees to do the same. They also asked staff to dislike negative reviews to get them pulled, according to the FTC, and even resorted to using VPNs to mask their identities after Sephora spotted earlier fake reviews."

 Thankfully review management technologies take this issue seriously. For example, Trust Pilot has a four-tier approach to dealing with fake reviews. They engage a compliance team, tech team, enforcement team, and help from their community. (source

The device matters - Effects of consumer reviews on phones VS desktops
mobile desktop.png

Anywhere there is research on the thoughts and feelings of consumers will get our attention. A recent study looked at over 275k reviews to see how reviews composed on mobile compared to those written on laptop/desktops. The researchers sought to understand the value of reviews for the reader, not the writer. "Mobile reviews were associated with 10-40% fewer likes than the reviews generated on laptop or desktop computers."  

What I find interesting in this study is that if you want a better understanding of that customer's experience, then you want a mobile review. The team found that "...content is more effective, more concrete, and less extreme when created on mobile devices".  

For the person writing the review, "Writing reviews may be therapeutic and help consumers make sense of their experiences." Extending this thought to consumer feedback, it would seem that we want to give consumers an easy mobile experience so that we get the clearest picture of their experience. 


Based on the work of Sam Ransbotham, Nicholas Lurie, and Hong Liu (source: link), "Creation and Consumption of Mobile Word of Mouth: How Are Mobile Reviews Different?" 

Science Daily (source: link ) 'Consumers: Online restaurant reviews are not all equal - Consumer appetite for restaurant reviews depends on how they were generated' (source: link ) 

Let’s look at groups: the same issue is experienced differently

In our last post, we talked about identifying Topics. Today we are going to explore the importance of looking at how those topics change with different groups of consumers. You can define a Group by many factors, including; geography, purchase frequency, age, gender, and many more. A deeper understanding can be achieved by looking at segments and exploring how different groups behave and how these differences impact their perceptions.  

An excellent example of group perception is that of an electrical issue that we found for a client that sells motorcycles. One of the Topics looked at how customers talked about electrical matters. We explored the impact that electrical issues had on the overall satisfaction scores and found that there was a small reduction in scores. When we looked at the customers by generation, it became evident that younger, millennial buyers' perceptions of electrical issues significantly reduced their satisfaction.  

The difference between the two was stark. With an overall brand NPS in the low 9s, the NPS for all customers with electrical issues was still about an 8, which while not ideal is hardly something to stop the presses over. When running a regression on the overall population, electrical issues were not a significant predictor of a deviation from the brand level NPS. 

However, when looking only at millennial customers the average NPS for someone with electrical issues drops to a 5, and the topic becomes the number one predictor of a detractor score among millennial buyers based on a regression analysis. 

This finding led us to take a more in-depth look into how different age groups perceive electrical issues. We discovered that although both groups experienced the issue, older buyers were more tolerant, for them, it was part of the experience. Younger buyers, by contrast, viewed the problems as a significant flaw.  

A deeper understanding of the customer's perceptions is developed when you explore how each group talks about their experience. Remember, behind every comment is an experience that matters.  

How are you currently segmenting your customers? 

Behind every comment is an experience that matters

TL;DR Creating well-defined topics is an essential step so that you can get the information you need from comments to inform better decisions.

Contained within each comment provided by customers are essential topics impacting your business.  Do you know what they are?

Direct to consumer or two-sided market companies find it hard to organize the high volumes of customer comments. The TOPICS that your customers bring up are the ones that are most important to them and a valuable source of insights for you. Understanding what your customers' are talking about at scale gives you a business advantage that only the best have put into operations.  

Defining the topics that are important to your business is a process but not a difficult one. To get started, it helps to list the steps in the customer journey. Doing this will give you a big list of Topics. Now we need to define those topics with the words and phrases used when talking about the subject. 

Below is an example of a topic and how it is defined.

Topic: Exchanges and Returns

Topic Definition: exchange exchanged exchanges exchanging refund refundable refunded refunding refunds returnable "non-refundable" "return policy" "return policies" "return it" "return them" "returns it" "returns them" "returning it" "returning them" "returned it" "returned them" 

A collection of topics with their definitions, we call a dictionary. Managing and updating dictionaries is important in on-going analysis. PROTIP: Dictionaries are essential in the study but knowing where to route findings is critical to turning insights to actions. A best practice is also defining what department and WHO in that department will be the liaison. Leading companies have established insight channels to provide actionable insights to the relevant department. 

Contained within each comment provided by customers are essential topics impacting your business. Let's find out what they are?

What story is your customer satisfaction data telling ABOUT YOUR BUSINESS

The Variance Mountain

TL;DR When looking at good companies the shape is a slope and in bad companies it looks like a mountain.

An often-overlooked dimension of this is the shape or distribution of the variance across topical areas, and the scores your customers have given you. When you look at the data, through this lens, things become apparent. For influential companies with happier customers, the shape of their variance is a slope while it looks like a mountain if your customers are unhappy.

4th Image Smaller.png
1st Image Smaller.png

Looking at this, it makes sense at a basic level. As Tolstoy proclaimed, "All happy families are alike; each unhappy family is unhappy in its own way." People that leave good reviews need to say something, and invariably it will be a lot of the same, things that go right go right for the same reasons. Whether through excellent service, a good product, or meeting a commitment, people will be happy in the same ways.

When it comes to things going wrong, though, they tend to go wrong differently.

Take a topic like Delays. Intuitively we know many people will be unhappy about a delay, but we also know that the negative of a delay will correspondingly provide a company with a chance to create a good impression by solving the problem. One high scoring comment read,

"Check-in was easy; the flight was delayed by about 20 minutes, so we assumed we would be arriving in Orlando about 20 minutes late... HOWEVER, our pilot put the engines in "afterburner," and we arrived 22 minutes early and about 5 minutes ahead of some severe thunderstorms!" 

This is a perfect example of a negative experience giving a company a chance to make a positive impression. It drives variance higher in negative topics and creates the upper left to lower right trend we see across industries and data sets. This trend is also present in our dataset assembled from over 5000 hotel reviews. This is true when charting the average CSAT score against the variance score.

This again depicts the sharp upper left to lower right trend, indicating that lower-scoring NPS topics have high variance and high NPS topics have low variance values. One high variance topics in the data was 'booking related comments,' and it's easy to see why booking problems can be an opportunity to overcome or fail, and why there is a wide discrepancy in scores.

"We booked a suite, and it was never given to us. They were so rude and refused to act as if they had a suite. After 6 hrs of travel, they did not care about their mistake. I would never stay here again."

"My wife and I arrived about 2 hours ahead of the check-in time, and they were completely booked, we had a reservation, but we were early. They found us a room, and it was amazing. They have valet that's owned by another company they are really expensive but worth it if you're driving like we were."

These examples highlight the natural variability of customers discussing booking related issues, and while all businesses strive to turn problems into opportunities, and it is natural not to bat a hundred percent on those opportunities. There is, however, a second possibility, when you have topics with low NPS and low CSAT scores that also have low variance. It would represent a more negative hypothesis that your company has issues where they are consistent and bad. This sort of chart would look different, more like a mountain than a ski slope. With this in mind, let's take a look at a quick plot of NPS for a telecom companies product (left chart).

This shows the telecom data with a standard OLS regression that basically tells us nothing due to the scattered nature of the data. However, when I break the data set in half at an NPS score of 5 and run two regressions, the results are starker.

This chart clearly shows their most negative topics in terms of NPS are also tied with the most positive topics for being the lowest in their variances. Not only are they doing things their customers dislike, but they are also consistent and repeatable with these actions.      

It represents the worst type of opportunity, a squandered one. 

 When plotting your variance data, it is essential to realize that high variance isn't always a bad thing; it can be a positive sign that the steps you have taken to address known problems are working. Remember, a low variance is not always good either; the only thing worse than being bad at something is being consistently bad at something!

OdinAnswers with Happy Market Research

Jamin Brazil of Happy Market Research discussesAnswers with Andy & Tom

At this year’s Insight Innovation Exchange (IIEX) in AustinTX, we had the pleasure of chatting with Industry stalwart Jamin Brazil of Happy Market Research. Jamin asked about our recent rebrand and the exciting new way that busy customer experience professionals and marketing researchers can leverage intersections between key metrics, customer segments and comment data to quickly get the most actionable answers (White Paper).

Follow links below to listen to the podcast interviews.


[Republished from Happy Market Research]


Andy Greenawalt - OdinAnswers

Rebranded Odin Text. They pivoted from "what goes in" to "what goes out". Odin Answers is the intersection between the text and segments that produces a just-in-time business insight. They specialize in servicing large Digital first companies. 

Learn More: https://happymr.com/iiex-na-2019-andy-greenawalt-odinanswers/


tom podcast.png

Tom H. C. Anderson - OdinAnswers

This guy is the pioneer of text analytics. I've been a huge fan of his and his company. We even co-presented a few years ago at the CEOSummit. 

In this episode he talks about how important it is to triangulate truth through multiple data sources while including the customer voice. 

Learn More: https://happymr.com/iiex-na-2019-tom-anderson-odinanswers/


As always feel free to post or email us your comments and questions.

We’d love to show you how your team can use OdinAnswers to be more effective. Feel free to request an exploratory call or live demo here.

Summer is here. Don’t let your team waste another minute waiting for a data scientist, or trying to connect thousands of customer comments, segments and business metrics. OdinAnswers can help.

Tim Lynch
Beyond Text Analytics, OdinText is Now OdinAnswers

Text Analytics Isn’t Enough - OdinAnswers Easily Connects Thoughts and Feelings to Metrics That Matter

I’m very excited to announce that OdinText is now OdinAnswers. Here at OdinAnswers, we have long been about more than text data. OdinText was the first text analytics firm to realize that customer comment data never exists in isolation, and that any interesting data is a combination of thoughts, feelings, knowledge about where these come from, and how they affect your business. Our earliest thinking and patent claims reflected this fundamental truth.

OdinAnswers, which we are introducing today, is a celebration of the fact that looking at either hard metrics or human thoughts and feelings in isolation only gives you a partial picture. Our new customer analytics platform is designed to easily give you the best possible answers across all your relevant data.

You can read more about why text analytics alone doesn’t provide sufficient understanding in our white paper here (Text Analytics Isn’t Enough: Moving From Text to Answers).

The Problem With Traditional and Text Analytics

Let me take a step back. All analytics tools, whether general or specifically for text, have limited users in two important ways: by their focus on one type of data, and their difficulty of use.

While there are countless providers in the space, businesses continue to make decisions based on a fraction of the available data. Moreover, current text analytics options, whether open source or third party, are so difficult to use compared to their added value that even Fortune 500 companies continue to struggle with human reading and annotation of customer comment data in Excel.

The Opportunity Is Tremendous

Think back with me, just a few years ago, to the CD players of the past. Sound and music fidelity had reached its peak (or so we thought).Any analyst would have told you that the market had been fully realized.Ironically, looking at Gartner and Forrester adoption curves for text analytics, one might be led into thinking the same about text analytics today.

What happened of course was the iPod. The iPod changed the market completely and forever.

This is the current state of analytics. It’s about more than just AI and machine learning. We are at a pre-iPod moment in history. The key is to think about data differently, more inclusively, and as importantly, to simplify the access to this data and analysis. What we want is better answers, not more complexity.

Exploring the Possibilities With OdinAnswers

In conjunction with our renaming, we are introducing two new products, Answers, our flagship product for those who need a continuous understanding of their customer-business relationship (at an intervals that suit them best), and Explore, a brand new friendly and easy-to-use interface with improvements in speed and functionality that help you quickly get the answers you need.

OdinAnswers provides real-world actionable answers to important questions such as:

  • How can I improve customer/employee satisfaction?

  • How can churn be decreased?

  • What are the key drivers of revenue?

  • Which new product should we launch and why?

  • What advertising copy will have what impact?

Amazing digital-first companies from Google and YouTube to Uber have told us that no other analytics tool helps them make these decisions better than OdinAnswers. While easier to use, OdinAnswers also allows them to consider the thoughts and feelings of their customers, together with other key data, faster than anything else currently on the market.

Still, we’re not satisfied with where we are today. To understand why I’m so excited about where we are going next, you’re welcome to download our new white paper here.  

We look forward to working with more digital-first clients large and small, as well as research and data partners, as we change and disrupt the current analytics paradigms in pursuit of the valuable answers you need.

Tom H. C. Anderson
Founder & Chief Research Officer


About OdinAnswers

OdinAnswers is a customer-experience analytics platform for digital-first companies to better understand their customers by finding the hidden relationships between customers’ thoughts and feelings and business performance data. OdinAnswers uses natural language processing, advanced statistical modeling and machine learning to help its customers win the insights arms race. To learn more, visit http://www.odinanswers.com.

Tom H. C. Anderson
Why I Joined The OdinText Board

Last month, I was honored to be invited to become a Board Director for OdinText, and gladly accepted. But the reasons I am excited to be a part of OdinText go all the back to 2002. That was the year I entered the insights field, in the qualitative part of the industry. I started out as marketing and support staff for my wife’s ethnography consulting agency, then developed as a researcher pioneering online qualitative research, and then founding the Revelation in-depth qual platform in 2007 acquired by Focusvision in 2014, and then as a C-level executive for one of the larger insights tech companies in the world offering both quant and qual solutions. 

All through that journey, I felt there was opportunity being left on the table. Incredible amounts of amazing rich textual data would be collected, filled with insights and game changing answers just waiting to be discovered. But the effort required to extract that insight was immense. Coding tools and early text analytics tools often proved to be more effort than was worth it, especially when working on rapid commercial market research timelines. It was frustrating to know that while researcher and decision makers were being served well, there was more to be had.

The most frustrating thing is that I could see where things needed to go. We needed to put machines and humans in position to do their best work. Machines are able to find and surface patterns in immense data sets, something humans aren’t well suited to do. Humans are key in assigning meaning to those patterns. But we just didn’t have the tools.

As a new generation of text analytics tools emerged over the past 5 years, with OdinText being on the forefront, and processing power has reached a critical point, I saw that the puzzle is finally being solved.

The key breakthrough is to drive the understanding soft data - thoughts and feelings expressed in textual responses - through the lens the hard data and business metrics. The power of statistically-driven text analytics enables decision makers have not just “the what”, but also “the why” at decision scale. This is OdinText’s sweet spot.

Over the past 18 months I’ve seen the beginning of an amazing evolution at OdinText. Anchored by the pioneering expertise of Tom Anderson and the superlative analytics engine he developed, OdinText added experienced serial founder and CEO Andy Greenawalt who sports a tremendous track record of tech startup success. 

As I looked OdinText entering into 2019, I saw transformative tech, developed by one of the industry’s leading thought-leaders, now being focused and elevated by experienced executive leadership - all at the service of a vision to create a new kind of understanding offering. It’s a strong combination that made it easy to want to be a part of OdinText.

There are a lot of exciting developments on the way for OdinText in 2019, so please watch this space!

Steve August


Do You Know the Top 10 Slang Words for 2019?

Drip and Tea, two words you may yet know?

Here at OdinText we’re all about understanding sentiment, emotions and opinions and linking these very human feelings to business performance. In so doing, the deeper meaning of the language customers use is an important and fascinating topic to us. Of course no part of language is more dynamic than slang.

In our annual post on trending terms we decided again this year to skip political oriented buzz words such as “fake news” and “snow flake” (the term impeachment has been on the rise), and refocus this post just on slang.

 “Lit” remained in the #1 spot once again this year, followed by “Yeet” which had moved up from #10 last year.

Of greatest interest this year are two brand new words we just began tracking this year, and yet they’ve already made it onto our top 10 list! At #8 we have the term “Tea”, and in #3 we have “Drip/Dripping”, a fast moving term which may end up being one of the more popular for 2019. Do you know what they mean? [You check your answers in our definitions below]

Top 10 Slang Words for This Year

Text Analytics Slang 2019.png

*Bold terms are new this year


“That party was Lit yo!”

#1 Lit Holding at #1 third year in a row, Lit (like on fire) literally remains “cool” for now. It has climbed from 4th place back at beginning of 2016, it may be that its position gets challenged by something fast moving like Drip, but it would probably need to be a term with as general a meaning as cool.




I’m so excited,yeet!

#2 Yeet Surprisingly, a term that was tied for 10th last year has jumped to #2, no small feat. It’s original popularity came from a new dance move and subsequent internet video meme. But as is common with slang it can transform and take on multiple meanings. By morphing into an expletive meaning excitement, it has increased in popularity.



Wow, you drippin!

#3 Drip Drip and Dripping came out of nowhere. Drip is closely connected to Swag, Swagger, Bling and Ice, and it is quickly replacing these. It’s popularity can be traced to various rap artists who began using the term [see genius hip hop lyrics chart below showing use of Drip VS Swag etc. in lyrics].

Slang text analytics 3.PNG

If you’ve got enough money, gold and jewels, then you have Drip or are Dripping.



“Clean your room”


#4 Bet Bet moved from #18 to #8 last year, and is now at a solid #4. That kind of movement almost always has to do with new usage and/or inclusion into some popular lyrics or meme. Moving from a simple term indicating agreement, e.g. “want to go to the movies?” “Sure, Bet!”, bet has been changing to just mean “yes”, and then, more importantly and ironically the total opposite of agreement, meaning doubt and sarcasm or simply the opposite of what someone wants or No. “Yo can you help me clean my room” “Bet (leaves walks out of door)”. It has even come to be used as a sort of replacement to Yolo., but the newest and most popular meaning currently is as the opposite of the older meanings, a negative sign of disbelief. Basically a sarcastic "No".



“That’s so fetch!”

#5 Fetch Fetch climbed from 6th place back at end of 2015 and is basically maintaining its place at 5th this year. This one is for white girls it seems, and is among the most female scewed slang term we track. This term was popularized by the movie Mean Girls and means cool/chic.


“Dope dude”

#6 Dope Dope can be used a number of ways (see 2016 definition), including as a synonym for Lit, basically high in quality or mind blowing


“What up bro?”

#7 Bro It seems “Bruh” which reached a high of #10 back in 2015/2016  has been flat and/or morphed back to the more common “Bro”. While there have even been female variants like Bra (in part promoted by advertising related to breast cancer - someone who supports you when you have breast cancer). The meaning of Bruh has been changing for sometime from a term of endearment (brother), to Bruh?! Meaning  “Oh no… why did you do that?!”, and now it seems, back to the more simple term of endearment, Bro.


“Spill the tea yo!”

#8 Tea Here’s another brand new up and comer we just began tracking quite recently. If someone asks you for the Tea, they’re nt talking about a hot beverage, they want the juicy gossip!


“That deserves a dab!”

#9 Dab The meme-able dance move known as ‘dabbing’ spawned and gave way to the term with similar origin “Yeet”, which has taken on more meaning and popularity than dab. While this term continues to slowly wane in popularity, its use together with marijuana may help it morph and remain in use. More something you do than say, after a win or achieving something like a touch down in football, the player might dab to celebrate the awesomeness of it.


“Wow, I’m shook”

#10 Shook Last year we mentioned Shook as a bonus term you may want to keep your eyes on. It was a popular meme, and the term’s meaning varies slightly depending on the context. Generally, “shook” refers to a state of fear or of being shocked or stunned. It can also refer to a state of being deeply affected by an experience (implicitly traumatic) or even the way one might be momentarily struck by the beauty of a romantic prospect ala Elvis Presley’s “I’m All Shook Up.” Of course it can also be used sarcastically, when one is definitely not surprised.

In any given year there are usually about 50 or so slang terms that appear often enough for us to consider active and track for our Top-10. Other than geographic differences, there are usually differences related to gender, age and other demographics. For instance while “Fetch” and “Bae” are far more likely to be used by women, Yeet and Woke are more popular among males. Terms like “Lit”and “Yeet” skew younger, the term “Dang” is rather Southern, whereas “lit” and “Bae” are more popular in the North East.

That’s our top 10 list starting us off for 2019. If there is a term you’re curious about and wondering whether it’s just popular locally, or is getting a broader foot hold let us know and we’ll look into it.


Top 5 Text Analytics Tips of The Year

Happy 2019 & Top Posts of the Year

Thank you all for your readership in 2018. We’re starting out the New Year with some changes to our website, so please bear with us as we migrate our older blog posts over and get things updated.

Our first post of 2019 will be our annual post on the changes in popular slang, a favorite among trend watchers and those following Millennials and Gen Y.

In the meantime in case you missed it, here are the top 5 posts of this past year ranked by popularity.

#1 A New Trend in Qualitative Research

Almost Half of Market Researchers are doing Market Research Wrong! - Interview with the QRCA (And a Quiet New Trend - Science Based Qualitative).


#2 Trend Watching +OdinText

How Your Customers Speak - OdinText Indexes Top Slang and Buzz Words for 2018 


#3 What You Need to Know Before Buying AI/Machine Learning

7 Things to Know About AI/Machine Learning (Boiled Down to two Cliff Notes that are even more important).


#4 Advertising Effectiveness +OdinText

Ad Testing +OdinText, a Review of the 2018 Super Bowl Ads

#5 The State of Marketing Research Innovation

What You Missed at IIEX 2018 – 3 Takeaways


Closely tied for 5th place were Market Research CEO’s Summarized and Text Analyzed (via the Insights Association CEO Summit), and Trump’s Brand Positioning One Tear In (Political Polling + OdinText)

Wishing you an exciting and prosperous 2019!

Your friends @OdinText