Crimson Hexagon provides a useful report, The Nine Essentials of Opinion Monitoring - Measuring brand equity in the age of big data. As they note in the introduction, 41 percent of bloggers frequently post about brands they love or hate and their views are shaping the perspectives of their readers. So listen to what the bloggers are saying and the conversations in other web formats such as consumer communities, forums, discussion groups, and product reviews. They also note that “although the numbers are often disputed, one estimate puts the number of bloggers worldwide at 184 million and the number of readers at 346 million.” Whatever the number, 41% of it is very large. So you need scale to look at these opinions and you need to be able to filter out the noise.
I got a lot of false positives with Google Alerts but the total number is small enough that I can simply scan the total. But what if you are a big brand and get thousands of mentions in a day? Simply counting mentions is not good enough but sorting through them for true positives and then understanding what is being said can be difficult when the returns are high.
The paper states that simply discovering positive or negative brand opinion is simply not detailed enough. You need to ask the right questions. They offer some areas to look at: your brand attributes that are most important to the public, what motivates customers to buy, features most and least liked in your product, problems people are having with your product or service, and the most pressing customer service issues in online communities devoted to your industry.
They go on to write that you also need to understand the authority of the writers who cover these issues. In some instances, you will want to hear from the largest possible online sample to make sure you are seeing broad opinion trends. In others, you will want to “weight” results as needed, and to understand if certain sentiments are emerging only from the influencer community or more broadly. You also need to follow these trends over time to see if opinions change and the weight of writers evolves.
Next, they cover what online opinion monitoring tools should do. I will let their words describe the possibilities. “More advanced online opinion monitoring borrows from these principles: machines and humans helping each other. Human readers prime the software by choosing an indicative range of text as examples of negative and positive opinions within each category of interest. This indicative text can include slang and grammatical errors, thereby teaching the software to identify opinion expressed through many language variations. The algorithm then extrapolates this base knowledge to thousands or millions of documents that would be impossible for even an army of human readers to scan.“
The report goes into more detail. Crimson Hexagon provides an online opinion monitoring tool but the report offers product neutral advice.
Comments