By Michalis Michael
Over the past few years, it has become apparent that social listening and analytics is steadily becoming an integral part of market research and consumer insights. You could say that there are three main data sources used to generate consumer insights:
- Asking questions – intercept surveys or online focus groups
- Tracking behaviour – in the form of purchases, website visits etc.
- Social listening – analysing public online data posted by consumers
This is the Triple S Integration: Surveys + Sales + Social, with surveys representing asking questions, and sales representing tracking behaviour.
Although integrating data is nothing new in principle, unsolicited opinions such as those analysed through social listening, tend to be missing from the equation. The reason behind this is that social data is unstructured. This requires significantly more effort to analyse, especially if it is to be compared to, and correlated with, structured data from other sources. This is where technology comes into the picture, specifically machine learning (ML). ML is the means to get to Artificial Intelligence, through the use of which we can analyse unstructured data in an automated way.
Conducting accurate social listening & analytics that works in any language, any country, and for any topic is challenging. However, it is doable. Furthermore, there is no reason why this analysis cannot be integrated with more traditional data sources to synthesise unique consumer insights!
Traditional market research methods such as face to face or telephone interviews and physical focus groups, are neither adequate nor fully representative when it comes to gauging consumer perceptions. Even “traditional” online surveys are inappropriate, unless you think relying on consumer memory is a good idea.
There are so many ways to engage consumers online, ask questions and co-create digital content, why would an organisation resort to a lengthy and probably dull online survey? Online communities offer the possibility to mix and match data collection methods with the same objective. For example, starting out with a poll, followed by a bulletin board discussion including open ended questions, a photo or video diaries exercise, and lastly a group discussion for the most engaged consumers.
Integrating this analysis of various sources, including social listening, with brand health trackers and retail sales data enables the synthesis of otherwise unobtainable insights, such as a correlation between consumer sentiment and product sales. In one study, the R square between sentiment and sales was 0.81, with the beta coefficient (level of causality) for positive sentiment and sales double than the equivalent for negative sentiment. In another, correlations were observed between traditional survey KPIs and social listening metrics. Taking into account that social data can be as granular as we want it to be in terms of timing, it could even be used as an early indicator to track monthly brand health or NPS.
The web is the biggest focus group in the world. It is accessible, and it is always on. Consumers will continue to talk about products and brands online whether we want them to or not. Therefore, it is in every organisation’s interest to “listen” to these online conversations and use them to their benefit. As a matter of fact, ignoring these conversations can only make it worse. Some people believe that only negative people post online, therefore online consumer opinions are biased by default. However, this isn’t the case. The predominant sentiment expressed is actually neutral. Even if the above belief was in fact true, the few who post negative opinions have an impact on other consumers’ perception of a brand and ultimately their purchasing behaviour, so they are definitely not to be ignored.
ML makes it possible to analyse text, audio, and images in a fast and affordable automated way. Online communities enable agile and on demand research. What if we put it all together? Then we will be able to ask, listen, and observe, collecting valuable customer information resulting in otherwise invisible business insights.
By Michalis Michael, CEO, DigitalMR