Customer and Social Media Insights

I can spout off all the marketing speak on GameVoice that I can but at the end of the day, it’s about seeing what GameVoice can actually do.  Ok, down to brass tacks. Here are some surprising insights that we pulled for our July 2010 report.

• Platform Insights:  The Sony Playstation 3 generated the highest conversation volume of any platform with over 73,000 comments.  Microsoft Xbox 360 followed with 62,000 comments surpassing the PS3 in daily comments on July 21, 2010, when the Medal of Honor beta version went live on Xbox.  The Nintendo Wii rounded out third place with almost 34,000 comments, followed by the Sony PSP and the Nintendo DS.  The not yet released Nintendo 3DS, announced in June, had a decrease in comments in July to 6,462.

• More Platform Insights:  Although Nintendo’s DS and 3DS platforms are lowest in terms of total comment volume, the handheld devices rank highest in positive sentiment.  The Xbox 360 and Sony PS3 rank the lowest in the sentiment index.

• Publisher Insights:  Electronic Arts (EA) leads in online conversation volume with over 26,000 comments focused primarily around EA developer BioWare’s Dragon Age franchise and the upcoming Medal of Honor game.

• Released Game Insights:  With over 32,000 comments, Call of Duty: Modern Warfare 2 is the most talked about game in the month of July. Battlefield: Bad Company 2 is the second most talked about released game with about half as many online comments, and Red Dead Redemption, which led the conversation in May, dropped to fifth place in July with 10,000+ comments.

Can you name another social media analytics app for the video game industry that can do all of this?  Counting comments is one thing, but measuring sentiment with a high degree of accuracy?  GameVoice’s OpenMic software has got the magic as you can clearly see.

Demo GameVoice Interactive FREE for 7 days

I started my marketing career working for a magazine publisher focused on printing video game consumer rags for the hard core enthusiast.  During my tenure at the magazine, print editorial gave way to the rising social media wave:  forums, blogs, podcasts etc. Basically, any online medium to talk about anything and everything on gaming.

A few years ago, I moved into the brave new world of social media analytics.   All I can say is where were these social media analytics apps when I first started my career?  It would’ve been so much easier for me to quickly understand all these consumers’ conversations than just randomly scouring our gaming forums as well as competitor forums.  Of course talking to the editors of the magazine gave me some good intel on games and game industry topics but listening in on what gamers say to each other is a commodity that shouldn’t be overlooked.

Here’s an example of what I’m talking about.  I used to put on video game contests and tournaments as well as line up sponsors to provide prizes for these gamers.  It’s interesting to be able to cut through the gaming lingo and past the company marketing speak when monitoring these conversations.  As a marketer, we get so wrapped up in our message that we sometimes overlook the “voice of the gamer.”

What did this mean for me?  I put on some contests and executed some activities that weren’t all that appealing to some gamers.  And of course I read about it from comments posted on line.  Fortunately, my team of consultants and techies set me straight to avoid any huge marketing disaster.  What would’ve been useful is a tool that aggregated this buzz into one comprehensive and organized source to analyze.

Gamers don’t just dissect the product; they analyze the brand as a whole.  I’ve come to learn that gamers are a passionate group of consumers.  From fan boys to conspiracy theorists, they will take a game or brand and pull it apart, for better or worse.  The solution to monitor video game industry buzz?  GameVoice Interactive.

GameVoice Interactive – Main Dashboard

GameVoice Interactive Dashboard

For those that haven’t seen the press release yet, GameVoice Interactive launched on August 16th to compliment the insightful GameVoice Report. Besides being able to reference the industry benchmarks within the printed report, users can now experience on-demand social media insights with custom research and reporting options.

GameVoice Interactive – Insight Explorer

GameVoice Interactive Insight Explorer

Using the same application GameVoice analysts leverage to produce the original GameVoice Report, video game industry managers can now analyze their own insights on over 300 upcoming and released game titles sourced from over 35 of the largest video game communities. The 30,000 ft. view of the industry is only a start. It’s become apparent to us at GameVoice that managers have their own specific business objectives and because of that, they require more detailed information on social media data. We at Overtone can’t predict what everyone is looking for but we can certainly provide to the tools to aid in this discovery.

One of the fondest memories I have as a kid is learning to drive. I couldn’t wait to get behind the wheel of my ‘67 Shelby Cobra GT 350 and hit the open road. I could taste the freedom as I drove off into the sunset, free at last, free at last. Of course reality set in and I realized that I would be sharing part time driving duties with my sister in a ‘78 Buick Electra that had a 225 and an 8-track player.

Either way it was only a few short months until I had the keys in my hand. Well, I can tell you those months were anything but short. My dad, “the drill sergeant,” grilled me on everything from proper braking techniques, to adjusting the mirrors, to the pre-ignition checklist. I thought he was out of his mind. Why couldn’t I just get in and drive? How hard could it be? It’s a motor, 4 wheels and a few pedals. I just want to get in and go. Well, I can tell you that way of thinking can get you into a lot of trouble. Hindsight is always 20/20.

Implementing a social media strategy, in my mind, is very much like learning to drive a car. Social Media, it’s exciting and you can’t wait to be a part of it. Your mind is racing with possibilities, Facebook, Twitter, Yelp, forums, blogs and Foursquare.  Social media will set you free.  How hard could it be? All you need to do is collect some data and use a social media tool right? Wrong.

A social media solution, with social media content fueling it, is very powerful. And much like my dream car, the ‘67 Shelby, has all the performance and handling you could want. But, put it in the hands of the wrong driver and it may perform like a ‘78 Buick Electra. You can’t just get in and go. You have to learn about and understand the power of social media and how to align a strategy with your company, product, brand or service goals.

So how do you get the best performance out of a social media solution? As an organization you should start to:

• Define your goals and objectives;
• Identify the audience or community you want to engage;
• Define what success will look like.

As the frenzied quest to implement social media strategies continues, there will be more and more “drivers” on the road. There will be the guy trying to do 100 MPH in a Yugo, with no plan or direction; the driver doing 40 MPH in the Bugatti Veyron (Top speed is 253 mph for you car enthusiasts); and then the driver who knows when to go 100 MPH, when to go 40 MPH and knows exactly which road to take. This person you might imagine is driving the ‘67 Shelby Cobra GT 350.

Which one will you be?

The day has come!  Today Overtone introduced OpenMic version 5.3 and I get the privilege of talking about it!  We have spent a great deal of time and effort listening to our clients and prospects about what they want in a listening solution.  After all, what kind of listening solution would we be if we didn’t listen to our own clients wants, needs, and desires? The over-arching theme from our interactions with clients, market experts, etc. is that Overtone’s core value proposition is the ability to automatically and accurately categorize large volumes of comments for facilitating insights and action.  We want to continue to deliver on that core and promise, and have been diligently working on new features, functions, and methods that continue to keep OpenMic as the frontrunner in comment categorization and sentiment accuracy.  While our core remains accuracy and reliable information, we recognize that no listening solution would be complete without an intuitive and friendly interface.  As such, we have refreshed the dashboards, reports, exploration tools, etc. with various usability enhancements and fixed quite a few bugs, to make finding insights as easy as possible.

Here’s a highlight list of the major new features of OpenMic v5.3.  For a more in depth look at features and benefits, click here.

• Hybrid Statistical/Linguistic categorization engine: We’ve added phrases, stemming, and negation to augment our statistical engine for higher accuracy, but still maintaining the flexibility for any business user to train and track any category they want.
• Sentiment 2.0: We have increased sentiment accuracy, added flexibility in sentiment category assignment, and have enhanced the dashboard and report charts to offer further detail into breakouts of Positive, Negative, and Mixed comments.  And, of course, each of these widgets is only one click away from drilling down into comments and other sub-charts for deep analysis.
• Dashboard Custom Dates: Do you have events such as marketing campaigns or product releases that have occurred in the past, and outside of the Dashboard rolling date ranges? Wouldn’t it be nice to have the ability to enter in multiple custom date ranges in the Dashboard and not just Reports? We’ve listened to your requests and have added the ability to define custom date ranges in the dashboard!

1. Whatever your business, your customers are voicing their opinions, find out where the conversations are taking place and harvest them, use this to augment surveys, focus groups, and other forms of traditional customer and market research.
2. Make sure it’s obvious and easy for your customers to leave feedback in their own words, especially in each place that they interact with you (customer touch point), encourage them with hard and soft incentives, such as membership in a customer community, acknowledgments, and special offers or promotions.
3. Social Media is significant source of customer conversations, get on the bandwagon (not just communities and blogs), participate and actively listen – you will need technology to maintain coverage and to monitor this dynamic environment focusing your attention on where the conversation is most relevant
4. Capture both structured and unstructured customer-generated content. then organize it into categories that are meaningful to your business – those that represent your brand, how you want customers to perceive you, and topics that you can take action on
5. Make sure the cycle time to capture, categorize, analyze and  act matches the purchase cycle for your product -  in many cases this is real time, so accuracy and precision are also key – if your feedback volume is large, then this will have to be automated – text analytics is the obvious technology
6. The action starts as soon as you gather customer feedback – immediate acknowledgment, timely and intelligent responses and then deliver the insights to the right people inside your organization who can act upon them, then ensure those actions are taken.  There are strategic actions, tactical actions and just plain listening that will improve your customer experience – some of these actions can be automated with the right technology in place
7. The departments who take their cues from the feedback need to have a quantifiable structure to evaluate the feedback and give them the mandate to act– use on-demand reporting and business intelligence tools to deliver those insights

Surprised by What You See in the Social Media Mirror?

If you take on a social media analysis project, you might be surprised by what you find.  What your customers, pundits, and competitors say about you online only reflects, and in some cases amplifies, the shortcomings of your brand, product, service.  And, most importantly, reflects the overall customer experience. Yes, you’ve heard that before, but can your social media strategy really change things about your online reputation?  To quote an astute customer at a large telecommunications company when asked how his company handles online reputation management – “We don’t give them anything to complain about.”  This customer just happens to top the list of just about every annual “Best Customer Service” poll.  In fact, they run their business by listening and methodically responding to customers based upon feedback they receive directly from their customers via their mobile phones.  This level of customer feedback analysis is rare. In meetings with CMOs and VPs of Brand Marketing of well known brands, I hear the same response to the question – “Do you analyze all the direct feedback you receive in email, web forms, and open-ends in surveys?”  The answer is in most cases, “Well, sort of, we sample some of the feedback, but mostly that’s really customer support’s area.”

Sure, analyzing the black hole of email and other direct customer comments you receive is not as sexy as “engaging” in social media, but it really highlights why some customers feel they need to speak about their brand experience to the public.  The real solution is to best understand and gain insights from both direct customer feedback as well as social media.  Direct customer feedback can be collected at every touchpoint with simple, “Tell us how we did,” comment boxes presented on your website, in store kiosks, mobile phones, or even voicemail boxes that are common with retail and hospitality chains.  Yes, you’ll get 1000’s of comments a week or even a day, but that’s why companies have turned to OpenMic so that every single comment can be analyzed and the correct response can be triggered.  Once you can truly understand what your customers are actually saying both good, bad, and ugly, then you should be comparing it with what’s being said in social media. In most cases within online communities that you’ve sponsored or at least those websites that best represent target customer segments.

So, where do you start when your CEO or CMO is telling you to do something about social media?  Well, start with listening, but include in your project both direct feedback and social media analysis so you have the whole picture.  Forrester Research, in their recent report Peer Influence Analysis (April 2010; www.forrester.com) said that only 16% of online US adults create a majority of postings and impressions that drive the perceived online reputation of brands for the rest of us.  There are two things to consider.  First, do the 16% really reflect the customer experience of all your customers? And, are they really your customers?  Regardless, your analysis of social media can only be justified if you compare it to feedback collected from all your customers.  And, if your goal is to deliver a better customer experience, your best shot at identifying and quickly responding to a customer is through listening and responding to direct customer feedback.  Every once in a while, you should put the social media mirror down and look at your customers in the face; they have something to say.

Managing your brand is hard enough offline, but managing your brand amid a social media landscape that continues to evolve and change at such an incredible pace can be overwhelming. Conversations about your brand, product or service are happening all around you. With the advent of social media, the voice of the consumer is growing louder and louder.

Listening to the right conversations can provide a wealth of invaluable information about how consumers are experiencing your company, brand, product or service. But unless this information is organized, processed, and managed properly it can be a burden, not a benefit.

We live in a world where bigger is better and it’s better to have more than less. It seems that this same mentality has been adopted by companies trying to leverage social media. But just because there are conversations about your brand, does that mean you should listen? Software makes it very easy for anyone to participate in and contribute to online conversations. But because anyone can be a part of the conversation, companies should do a “listening audit” to evaluate the value of those conversations and determine how useful they will be. If your car breaks down, who are going to listen to, your best friend or your mechanic? Consider not only what’s being said, but who’s saying it.

Like many people today, I often times will search the internet for information about a product or service that I’m interested in buying. For example, the other day I did a search for; Men’s Nike Running Shoes. My search yielded over 1 million results. My first thought was there is no way that I can mine through all this information. It was information overload. I also thought, do I really need this much information to make a decision? The answer was no. I only reviewed the sites that I knew, trusted and believed would provide me with the answers I sought. Out of 1 million search results, I reviewed relevant sites, and was able to make a better informed decision.

As an individual, in most situations, we don’t need or want to review every single piece of available information before we make a decision. But for some reason, companies have adopted this behavior when deploying a social media strategy. This notion of consuming as much social media as possible can leave an empty feeling. Because having a lot of information doesn’t mean it’s the right information.

Rather than taking a superficial, high-level view of everything, focus on the conversations that really matter. Prioritize the most meaningful conversations and don’t get caught up in the chatter.

The goal should not be to consume as much social media as you possibly can. The goal should be to create an effective social media strategy that allows you to listen to the most important conversations and then make informed business decisions.

Have you been handed an initiative to listen to and analyze social media messages for your product, brand, or service?   Were you given a few high level corporate initiatives to leverage social media messages to validate and understand your business trends?  If so, you should be asking these questions:

1. Why is customer attrition on the rise?
2. What do customers like and dislike about you and your competitors?
3. Why did your latest product launch fail to increase usability?

You perform a little research and come across a wealth of tools that allow you to enter in keyword searches. You sign up for one of these services, define your keywords, and begin to analyze. For example, to track customer attrition you enter searches of “cancel” “leaving” “and “switching”. You get high level reporting on how many messages mention each of your keywords, but you are struggling to extract insightful information to validate and understand the root cause behind your initiatives because due to the shortcomings of simple keyword search returning false positives and also only including the exact words or phrases you define.

By defining exact keywords, you limit your search to only those queries. Looking deeper into actual messages, you’ll find that the ways people actually mention an attrition risk can literally be hundreds if not thousands of different word/phrase combinations. Further, just because a message has the word “cancel” in the text, does not necessarily indicate an attrition risk. For example, someone could say “I had to cancel my credit card because I lost it, could you please call me so I can change my billing information.” That message would be included in your attrition risk keyword search, even though the customer does not actually want to cancel your service. This problem persists even further when you go to perform sentiment analysis on a topic, as there are just too many variations on how a customer can mention positive or negative things in a post (see http://eqi.org/fw_neg.htm for an overwhelming list). So, how can you minimize the number of false positives you receive, while also effectively catching the many variations, without defining 1000’s of keyword queries?

It’s time you looked into Natural Language Processing (NLP), or NLP.  NLP centers on the ability to extract author intent from their message. This science enables machines to discover and match author intent to a virtually limitless set of words and phrases. Further, in the application of Statistical NLP these technologies assign weights to words and phrases, as opposed to simple exact match keyword logic. In the attrition risk example, the word “cancel” will most likely produce a strong weight, and be a good indicator of an attrition risk, however, the other words that appear with it (e.g. would, like, to, please, me, subscription, service, etc.) would also be assigned a weight.  With NLP based processing, other words bubble up that the user would have never thought of to enter in as search words, for example the words “provider, assistance, help, hold, hours” are strongly weighted for attrition risk. The attrition risk category would then receive a machine generated threshold to determine whether or not a message has enough weighted words to meet the threshold for that category. The net result of these NLP approaches produce a much smaller % of false positives, and also increase the amount of recall (# of messages your category returns) due to the inclusion of other words you would have never have thought of to enter into a keyword search.

You’re probably wondering how a machine can automatically create these expansive categories. In a small sample (a few hundred), a human will categorize the messages with a higher degree of accuracy than any machine. However, when you give a human 100,000 to categorize, they will suffer from fatigue, thus greatly reducing the accuracy. Systems such as Overtone’s OpenMic have been built to leverage a small sample of human categorization, but then take that sample and use machine learning to categorize the rest. This hybrid approach using humans to train a machine produce categories that are much more accurate (minimizing false positives) and find more mentions (maximizing recall) than simple keyword searches.

Due to the highly accurate and highly expansive categorization results from the NLP based machine learning approach you can create categories that really center around a theme, as opposed to just a set of keywords. Categories like Positive Sentiment, Negative Sentiment, Customer Service Praise, Customer Service Issues, Loyal Customer, Sales Opportunities, Usability Issues, Usability Praise, etc. become accurate and expansive. The possible business insights from theme based categories are impressive, especially when combined with customer demographic and segmentation data. Tracking these categories over time, and spotting trends, and drilling down into the subthemes or clusters and comparing to demographics can greatly enhance insight discovery and root cause analysis.

Keywords do of course have their place. For example, tracking categories that don’t really contain a theme (e.g. competitor names, product names, etc.) are best handled by simple keywords. That is why OpenMic also supports Keyword analysis. But when it comes to accurately categorizing the theme of a message to inform trending and root cause analysis, NLP is the way to go.