One of the most important discipline areas at iStrategyLabs is buzz monitoring. While we’re technology solution agnostic, we’ve been using ScoutLabs a lot recently because they’ve created a product that is much less expensive than their competitors (something our clients certainly appreciate right now) and it works quiet well. There are a number of ways to conduct conversation monitoring, competitor monitoring, influencer identification and outreach etc., but this post will focus on just one facet: the comparison of two brands from a sentiment perspective.
Brands include in this buzz monitoring exercise:
Google and Microsoft
- Image by noworks? via Flickr
Buzz monitoring report range: February 6th – March 7, 2009
First, lets take a look at all blog posts mentioning “Microsoft” and see what the positive and negative sentiment of trend looks like.
On average there is more positive buzz generated about Microsoft than negative. Digging deeper into the numbers, we find that on average there are 5.4 more positive posts than negative posts about Microsoft. This could begin to serve as a benchmark for the brand, and since this is the first time we’ve done this analysis, there’s no way to know if these that a typical ratio for them.
Now lets take a look at all blog posts mentioning “Google”.
On average there is more positive than negative buzz for Google as well. Digging deeper, we find that on average there are 4.3 times more positive posts than negative posts about Google.
Next, we should see how their positive and negative sentiment stacks up against one another:
Volume comparison of positive posts for Microsoft and Google
Volume comparison of negative posts for Microsoft and Google
Lastly, let’s take a look at the share of voice between the two brands. This is a measure of who has a greater percentage of buzz about them on the web compared to the other. Keep in mind that if you look at the two graphics above, they’ll look pretty even. What you’re not seeing are all the “neutral’ sentiment posts that account for the bulk of content out there:
With positive, negative and neutral posts accounted for, you can see that Google is talked about nearly twice as much as Microsoft. This is perhaps irrelevant at the top brand level, as Microsoft and Google do not compete head to head selling “Google” and “Microsoft” but they do compete in selling things like “Google Apps” and “Microsoft Office”, as well as “Azure”, “Google App Engine” and another product “Amazon S3″ rounds out the cloud computing space.
Share of voice for “Microsoft Office” vs. “Google Apps”
Clear Microsoft Office dominates the conversation, but Google Apps has a 13% share of voice when they’re market share is probably 10% of that (I don’t have those figures).
Share of voice for “Microsoft Office” vs. “Google Apps” vs. “Amazon S3″
Azure seems to be the most talked about cloud computing application. I wonder if my data is correct, as I believe Amazon S3 has been around for a lot longer…
What do you think? What comparisons would you like to see? I may include your requests in my next buzz monitoring report!
![Reblog this post [with Zemanta]](http://img.zemanta.com/reblog_e.png?x-id=a72b7f75-b4e5-43be-abe6-a76aec1a9036)
Comments
I’m not familiar with the intricacies of Scout Lab (and these things are probably proprietary anyway), but how does the program accurately measure positive or negative buzz about a product? Does it recognize sarcasm? How accurate are the results against a control sample, e.g. of pre-counted positive/negative web content? Does it measure degrees of negativity, or the importance of the source? I’d think that a strong endorsement of a product in the New York Times should have more weight than mild praise on an average citizen’s personal blog. Perhaps this is all included in the equation, however.
Secondly, stepping beyond brands, could this be used to track results of public diplomacy or strategic communications campaigns? For example, track positive vs. negative buzz about the United States across content posted from servers based in the Middle East. Does this already exist? Is it possible?
Thanks for posting, Peter–good food for thought!
Peter,
Interesting piece but this still does not solve the problem of identifying the key influencers once the “buzz” topic or entity has been identified using these burst type services. It’s a very tough problem to solve but we think we’ve nailed it at Jodange, where we focus on extracting opinions and their source (opinion holder) from all online content.
Katherine also brings up a very good point on the other elements that need to be factored into the analysis of tone/sentiment.
Nice work.
Katherine: You bring up good points. Scout Labs looks at the sentiment for each post without any weighting for the prominence of the source it appears on. Here’s my recent blog post about how we produce sentiment:
http://www.scoutlabs.com/2009/02/26/how-does-sentiment-work-and-how-accurate-is-it-anyway/
We absolutely benchmark against human beings and our algorithm, while solidly on the valuable side of the equation, is not as good as college educated humans. It is terribly inaccurate at irony and sarcasm but it does do a good, directional job of assessing opinion- especially useful when you are dealing with volumes for brand like Google and Microsoft! Of course you can change the machine estimate on any content if your judgment is different than a purely grammatical read can supply (for example, all mentions of “unlocked ihones” are negative because people should not unlock their iphones). Do please give us a ping if you’d like to test drive the application for free!
Great Post, I’d like to refer to it in my next post. Here is a link to a series of articles describing IBM, Microsoft, Apple and Google. This is the main article with links to the related articles that have been published. the first one is Here, overview and description of market position.