USING SOCIAL MEDIA TO GUAGE CUSTOMER SENTIMENT

Almost 9 years ago I was involved in a startup that was building a community of college age people in England. Our plan was to virally market this and do what we called a “land grab” of consumers with the thought that at some point we would figure out how to make money mining the data from blogs, chatrooms, etc. We never did get funding for the venture, but kept the idea in my intellectual property (IP) toolbox.

Fast forward to about 2 years ago when I got back into this area by joining the virtual world called Second Life (SL). I spend a good year observing and interacting in SL looking for ideas on how to extract useful knowledge out of a chaos of information. In SL I was introduced to many other social media forums and eventually stumbled on Twitter.

I’ve been on Twitter now for a while and again looking for ways to experiment with the platform. During that time several world events were covered by micro-bloggers on Twitter which got me thinking (yes again). A world event will trigger a mass pool of observers posting tweets, and an even larger pool of carriers, virally re-tweeting these observations.

About 3 weeks prior to the day retailers in the US call Black Friday, I decided that it might be interesting to see if I can trigger such an event by asking my 500 or so followers to encourage their followers to tweet their Black Friday observations about crowds, spending, discounts and retailers with hot deals. By doing so I could mine the data and predict in real-time how the day was going.

So I enlisted several people I follow, including fellow blogger Brandon Mendelson who helped spread the word. I began to receive accounts from micro-bloggers beginning mid-day on Thanksgiving. This shocked me, but these were from people who camp out every year to be among the first in line for deals. A side note was that some of these people have the equivalent of a tail-gate party going each year so the wait is just one big party.

By 1AM my army of micro-bloggers said their goodnights and set the alarm for early Friday morning. Most stores opened at 5AM; however I didn’t begin receiving much in the way of information until 6 or so. The in store special deals appeared to be good, though certain stores had little or no inventory for these special items. Standard “bait and switch” Black Friday tactics were underway.

When the early deals were done, many items were still discounted though the competitive advantage went to stores who managed their inventories well and had good anticipation of product demand. Stores like Wal-Mart and Target were among the highest frequency in tweets all day.

I continuously posted tweets during the day with a summary of observations as I mined the data coming through. Several trends became apparent throughout the day:

The advantage went to stores with well managed inventories

Most people left without buying if they didn’t find the items they were looking for at one store. This was counter to previous years where people looked for alternatives even at higher prices.

People were willing to compromise selection at a lower price, but not if it was higher.

It became apparent that Black Friday 2008 would have higher gross sales for some retailers, but at significantly reduced margins due to consumers looking to stretch the money and only buying sale items.

Many high end items sat on the shelves even discounted with consumers waiting for better deals on Cyber Monday for the more expensive items.

Consumers were well disciplined, avoiding the careful marketing strategies of retailers that encourage impulse buying.

Early results from the traditional news agencies indicated a significant slowdown of sales on Black Friday. They determined this by the traditional method of interviewing people exiting the stores. When the real sales numbers came out late last week, many retailers showed an uptick in sales over last year, though at the cost of margins.

I believe the traditional method of interviewing people to project sales is no longer useful. Tools like Twitter can provide us with a more interesting view in that each person who Tweets can be an active observer and when Tweets are mined and correlated a more accurate picture is the result. Our experiment seems to prove this, though methods of mining the data are lacking at this point to do large scale consumer sentiment research.

Now we all know that the economy is suffering and most of us are looking to get more for our dollars. However I do see that in 5 to 7 years we will look back and see 2008 as the year that caused a disruption in the value chain of the retail industry.

I believe that going forward we will see a significant change in the way consumers shop and the Black Friday will become a term of endearment rather than that one day blow out sale. Make no mistake that retailers will continue to forecast their yearly profits based on a similar concept but that too will evolve.

I also see that Cyber Monday will morph as well. On-Line retailers will begin to use more intelligence in their special offers as well. Repeat customers will likely be aggressively pursued by e-tailers with similar promotions to their in-store competition (i.e. Friends and Family, etc.).

For the rest of the year I believe we will see retailers looking to clear out inventories by keeping discounts through the season and likely to offer a post-New Year’s sale.

Happy Holidays to all and good luck in getting those bargains!