You can read the post here When Social Media Strikes Post
Today, in light of the recent events at NetFlix (well actually over the past several months), I figured I would back track and analyze how social media struck Netflix and what mistakes they have been making by thinking their strategy can run faster than the social media speed of insight.
"On July 12, 2011 Netflix announced that it would separate the current subscription plans into two separate plans: one covering the instant streaming and the other DVD rental. The cost for streaming would be $7.99 while DVD rental would start at the same price." (http://en.wikipedia.org/wiki/Netflix).
Netflix added Insult to injury when...
"On September 18, 2011, Netflix CEO and Co-Founder Reed Hastings said in a Netflix blog post that the DVD section of Netflix would be split off and renamed Qwikster, and stated that the only major change would be separate websites for the services. The new service was to carry video games for an additional charge, whereas Netflix did not. Netflix subscribers who wanted DVDs by mail would have had to use a separate website to access Qwikster." (http://en.wikipedia.org/wiki/Netflix).
This event or some would say "strategic business choice" has lead to a major upheaval of what has been a very innovative brand over the past 5-10 years. They single-handedly changed how people rent movies and literally killed the on ground movie rental store. Blockbuster and Hollywood video were basically decimated and Netflix has been a leader in both the mail-order movie business as well as the streaming business.
If you look at the historic stock price of Netflix you will see nothing but upside. It has grown since it first hit the market from about $5 a share to its peak at $286.93 on July 11th. This is exactly the day before Netflix announced its now infamous price hike and then later in September split in how it would serve its consumers. Since that decision the stock has dropped to $108.66 in less than 3 months. They say you can take years to build a relationship but seconds to destroy. Evidence would suggest this is the case at Netflix.
So doing an analysis of social media would seem redundant. All the information sits in that stock chart right? It is...but how could Netflix have avoided its pain or at least reacted more quickly during this crisis to rescue some of its brand equity. In my experience, most brand equity is tracked using survey over longer periods of time. In fact, my current experiences have taught me that most companies are looking for ways to increase the speed by which they track their brand's health.
Below is an analysis I performed on Netflix. I started today and put about 5 hours into the work you will see below. I did this alone, with no help. In fact, I had knee surgery 5 days ago and decided to use our solution at NetBase to do this work to show how a single person at home at their desktop can analyze a problem like the one Netflix is having.
First, I looked at the overall buzz, sentiment and passion for Netflix over the past 12 months. To do so I used high precision sound bites are shown below. High precision is determined using natural language processing to get sound bites that are 80-90% accurate. As a scientist, having good accurate data is the key to doing strong analysis. You can read on how NetBase measures its accuracy here. NETBASE WEBSITE WHITE PAPER.
You will see there is a great deal of buzz (over 4,000,000 soundbites) over the last year and that Netflix has a very strong net sentiment and passion intensity. The definitions are listed in the slide below...
Looking closer, you can definitively see in the social media data that all was quiet on the social media front for Netflix over the past 12 months. At least, until they made their announcement on July 12th.
So social media can show us how people felt about their decisions and how it affected their brand versus before the decision. Let's go a little deeper. Below is a look at the social media buzz, sentiment and passion for Netflix before the announcement (October 11th, 2010 to July 10th, 2011) and after (July 11th, 2011 to October 11th, 2011). If we begin to analyze the numbers a bit, it gets even more intriguing.
What is the quantitative effect of their decision? You will the stats on their buzz, sentiment and passion before and after the announcements. Firstly, if we look at the buzz as percentage of the total, we see that 33.2% of the buzz generated has occurred in the last 3 months (7/11/11 to 10/11/11). So there is a definite, uptick in buzz since they made this announcement. So the buzz is up! But if you only know where and how much buzz their changes have generated, you are not looking deep enough. In fact, if you think about this some more (at the buzz level only) you will only say that there is a moderate uptick in the buzz and might not understand the meaning of it or if it is positive or negative.
Because of natural language processing we can go deeper to look at both sentiment (the positive and negative expressions of emotion) and passion (which is the intensity of that emotion...love versus like, hate versus dislike) because not all sentiment is created equally. To do this, I performed some very simple ratio analysis looking at the the number of positives/the number of negatives. This will give us a rudimentary way to track changes in both sentiment and passion. To make this meaningful in the context of the netflix announcement, I looked at the sentiment and passion data before and after.
Sentiment Ratio Analysis: Before the announcement on July 12th, Netflix has a sentiment ratio of positive to negative of 3.57. For every 1 negative expression of sentiment, Netflix received 3.57 positive expressions of sentiment. After the announcement the ratio drop to 1.85 almost dividing by 2. Essentially, in the last 3 months the number of negative expressions of emotion have doubled compared to the previous 9 months. This is a major difference.
Passion Ratio Analysis: If we perform the same calculations using the strong and weak emotion counts in the chart above we will see the same picture. Prior to the announcement, the positive passion/negative passion ratio is 4.92. At nearly 5 to 1, Netflix customers express positive passion for the brand. After July 12th, this ratio drops to 2.29. It drops in half. In fact, in the case of passion it drops a bit more than half. This again shows the dramatic fall of how people are feeling about the Netflix brand.
The question is why? In doing meaningful social media analysis it is critical to go beyond buzz (which we have been able to do) to sentiment and passion. The real value however, will come from understanding why. Below are some automatically generated charts from our Insight WorkBench. Again all data is using high precisions sound bites. The first are the top 10 "likes" for the Netflix Brand. The second is the top ten "dislikes" for the Netflix brand. Our method for "clocking" this precision can be found in our white paper on the subject. You can find it at the link on the NETBASE WEBSITE WHITE PAPER.
The why before the announcement...
Prior to the announcement people speak about things one would expect. They like streaming, the movies, and the service. They call it the best thing. As for dislikes, they also mention the movies and the service. They also mention that it just isn't for them or the selection. I would argue from a consumer perspective online, you can please everyone. They do mention an outage that occurred, but it is not dominating the chatter. Now let's look over the last three months.
Bringing into one view...
At NetBase, one form of analysis we often use what is called the Brand Passion Index. Attached is a short video that explains the chart and how to think about it. BRAND PASSION INDEX VIDEO LINK. Essentially, this view of social media data takes all three types of measures into the tracking of brand health. It uses the buzz (size of bubble), the sentiment (positive and negative expression on the y-axis) versus the passion (intensity of the sentiment (like versus love, dislike versus hate) on the x-axis) to give you an overview comparison of a brand's health. It can be created using competitors to see how your brand relates to theirs with consumers. It can be created over time to see the shifts (which I will show below). It can also be used to see how certain attributes contribute to the position of the brand. In this case, you take a dataset focused on a brand (Netflix in this case) and use verbal context to "carve" out a smaller data set from the larger one. Below is such a chart for Netflix.
You will see on Netflix brand passion index several different bubbles. You will see the whole year (complete brand look for Netflix) as the largest bubble. Overall, people were reasonably bullish on the brand. They show some intensity but the overall sentiment of that intensity is overall neutral. If we break up the timing by which the data set is created (using before 10/11/10 to 7/11/11 and 7/12/11 to 10/10/11), we see a different picture. Now we can see that prior to their announcement both the sentiment and the intensity of that sentiment is more positive. This demonstrates that things were pretty good for Netflix. When we look over the last three months we see a major downgrade of the brands overall brand passion. But if we look at some pieces of the data, we can see the key attributes that are contributing. The three smallest bubbles show subsets of the data. They included searching on things like price, announcement and Qwikster in the context of Netflix. In cases where key language was added to see how specific attributes affect Netflix, much of the negative chatter is due to these areas of discussion (as the pie charts above show). We can now quantify and see what parts of the social media discussion are driving the negativity of Netflix. This was done quickly and we only used the scalpel in a few areas, but with natural language processing you now have the ability to quantify what attributes of the social media data are causing either the pleasure or the pain.
So how could Netflix have taken action on all this....
Funny...I did a little thinking. How was this data predictive. It is clear if they had been understanding what was going on as it was happening, they may have made course corrections. But in the end it is about shareholder value. I did some very rudimentary trend analysis by doing my best to put the net sentiment over time charts against the stock curves available on Yahoo!.
Below are snapshots getting more focused in little by little...
You will see in each of the charts, the sentiment over time charts seem to show downward trends that occur prior to the drop in the stock price. This gives a glimpse of how sentiment seems to preclude the response in the market place. And how could Netflix taken action with this data. They could have watched after the first announcement at how the sentiment for the brand never recovered to original levels. They may have held off on the Qwikster announcement by watching if the sentiment improved and if the stock price began to recover. It seems the decision to move foward with Qwikster was absolutely disastrous to their brand and stock price. In fact, since the pull back on the Qwikster, it seems that the sentiment is improving. If we look at the stock price today is has moved upward again since yesterday. It has climbed so far today about 5 dollar to around 114 from 108 yesterday...
Maybe they will reverse the trend, maybe they won't, but one thing is clear...
When social media strikes and the company ignores the consumers...they do so at their own peril.