Understanding UK Women via social media: Learnings from a usage and attitudes study
By Gaelle Bertrand, Head of Brand Insight, Kantar Media
We recently conducted a study of women’s drinking habits in the UK by analysing social media conversations to derive insights around potential preferences and what influences women in their choice of alcoholic drinks.
What was interesting about our approach is that we integrated image content, gathered using logo recognition technology initially, and analysed the images in depth to decode the language of these visuals and derive insights around consumption moments.
Visual Intelligence: bringing in additional content
Image-based social media networks now play a huge role in documenting people’s lives but text- based searches (what most social media listening tools offer) only capture 20% of all image content. Almost 80% of visual organic brand content on social media is untagged, meaning traditional, text- based methods miss a significant part of earned brand mentions.
With the increasing prominence of Instagram as a platform for in-the-moment posts about FMCG brands, we wanted to test the additional insights that could be derived by including untagged brand images.
Currently image-recognition technology is limited to simple concepts such as logos and brand properties. Complex logos such as Starbucks are much easier to find than simpler logos like the Nike swoosh but overall the technology is quite effective.
In FMCG categories including alcoholic drinks, as in our case study, brands often define categories so logos are a good proxy when looking at images.
The importance of targeting relevant content
In addition to brand images found through logos, we gathered text-based posts about alcoholic drinks on Twitter and Instagram using words evoking all types - beer, cider, cocktails, wine, spirits, sparkling wine, champagne - and ensured these were consumer conversations by linking these terms with affinity language such as ‘I drink’, ‘I love’ etc…
We selected posts authored by women, using automated gender segmentation first, but we also manually checked gender attribution as automated methods are only 70-80% accurate for Twitter posts where gender is attributed (only 50% of the content), based on our research. Over a period of 3 months we found almost 6,000 relevant conversations.
Women’s drinking moments
Once we had gathered our data, we started by determining which types of alcoholic beverage were driving the most conversations online. But we did not just look at the overall volumes only. We wanted to distinguish whether mentions expressed a drinking moment, the need for a drink or more generic attitudes and perceptions towards a type of beverage or brand. We assessed the key drinking moments and where they took place for each brand and type of beverage.
We found there were different patterns in the content of conversations, depending on the type of drink mentioned, with in-the-moment posts dominating conversations, while general attitudes and perceptions accounted for just over a third of all content.
Thinking back around the concept of a traditional usage and attitudes (U&A) survey, what we found were drinking moments or the social media equivalent of usage occasions, we identified attitudes and perceptions, the U, in U&A. But we also identified conversations around cravings and upcoming events. That last part is the social media cherry on the cake; the part that any other type of consumer research may struggle to find.
More often than not it will validate or match what other sources have uncovered. But what it can provide is an in-the-moment and uninhibited (no pun intended) perspective which other sources of research would never be able to provide because of interviewer intervention.
Validating our findings
We delved further and our in depth analysis of conversations enabled us to map individual drinking moments for example, the planned, ‘proper’ night out where partying was the key motivation, and relaxation, taking some time out on your own or spending time with friends were the main drinking moments to emerge, accounting for two thirds of moments mentioned.
We validated our findings around these moments and the drinks consumed against our TGI data
and found that they mapped closely. What our social media analysis added was granularity around the micro moments, which social media is ideally placed for consumers to document.
This piece of research provides an interesting perspective around women’s drinking habits and offers avenues to shape communication messages and plan digital communication campaigns around specific moments. Understanding the specific (moment x location x choice) equation offers a considerable opportunity for brands to engage consumers at the point of consideration by leveraging their own social media channels.