Listen to your customer wherever they are

x Sponsored content on Research Live and in Impact magazine is editorially independent.
Find out more about advertising and sponsorship.

There’s nothing like a Twitter storm. The indignation of the individual, ideally involving a few carefully crafted sentences, eagerly jumped on by others – perhaps with the benefit of a hashtag – and, before you know it, you have a full-on public relations crisis for a brand comms team to try to avert, calm and resolve.

With social media offering such a public platform, easily shareable and able to rack up ‘likes’ and ‘shares’ in minutes, not just days, it’s not hard to see why companies spend so much time monitoring their online sentiment. Just because social media is easy to measure, however, doesn’t mean it’s the only, or most important, platform for customers’ opinions.

Last year, Kantar signed a partnership deal with social intelligence and analytics firm Engagement Labs with the aim of analysing all aspects of consumers’ conversations – online and offline. The TotalSocial platform collects data from social networks and Kantar carries out continuous surveys of a nat rep sample of the British public, to achieve a more balanced view. Sometimes, online conversations will mirror offline ones, but word of mouth often follows a different trajectory. Both sides matter to businesses, but just paying attention to one can give a skewed picture.

Brands can be assessed on four measures: volume or amount of conversation; net sentiment (subtracting negative conversations from positive ones); degree of brand sharing – how often people share a brand’s marketing in their online and offline conversations; and how engaged influencers are with a brand.

Online and offline

One example of the difference between online and offline sentiment cited by Kantar is when Welsh gardener Mike Armitage spotted a white film on his compost heap. It was a plastic residue from the teabags he was composting, and this raised awareness of the fact that plastic was used in teabags; about one quarter of a teabag is non-biodegradable. Armitage launched a social media campaign, which gained more than 200,000 responses, and PG Tips removed the plastic from its teabags, with other brands following.

For the tea companies, social media sentiment during this time was not good. Yorkshire Tea suddenly found the level of negative conversations equalled good ones – it normally enjoyed positive social media sentiment – and it looked like the brand was in trouble. Offline, however, positive sentiment remained constant and, one year on, most British tea companies have addressed the plastic concerns, and none has suffered the short-term drop in sales that social media conversations suggested was likely.

By analysing almost 400 brands in the UK, Kantar has found that the tea example is not unusual; there is a relatively low correlation between online and offline trends. Of those brands ranked, overall correlation was 52.5% when it came to average net sentiment over three years – where a perfect correlation between offline and online sentiment would be 100%, and 0% no correlation.

Survey data 

The online data sources used include Twitter, YouTube, Reddit, blogs, forums and customer reviews, with Kantar aggregating this through NetBase so it can be harmonised with the offline data.

Offline conversations are measured with surveys. Matt Dodd, managing director, analytics, media & digital, UK & Ireland, Kantar, says: “We conduct 26,000 surveys a year, nationally representative of the UK population, and we ask them to fill in the questionnaire based on their conversations over the past 24 hours.
“We’re capturing, on a weekly basis, the conversations around 16 categories that we prompt – ‘did you have a conversation about financial services’. We do not have a brand list; it’s for the respondents to say the brand.

“The interesting aspect is the evolution of messaging apps and measuring dark social. The only way you can do that in a consumer-friendly, compliant way is through surveys,” says Dodd.

Kantar asks survey respondents if they have used messaging apps, such as WhatsApp and Facebook Messenger, in the past 24 hours. “None of the digital titans will allow access to what’s being said within the messaging context. As GDPR and legislation gets more important to customers, surveys still have a valuable role,” he adds.

So how do online and offline sentiments vary by sector? Beauty tops both for sentiment. The top five categories for offline sentiment for the past
12 months are (in order): beauty, food, drinks, household products, and supermarkets. For online sentiment they are: beauty, household products, children’s products, drinks and fashion.

Natural influencers 

Handmade toiletries retailer Lush remains the fourth-placed brand (out of 410 measured) for offline sentiment, despite pulling its Facebook and Instagram pages. “It clearly shows that you can build a strong online presence organically,” says Dodd. Lush is second in offline sentiment.

“There’s still an element of sentiment being driven by people talking to people about beauty and cosmetics. Brands that have rushed toward super influencers shouldn’t forget that – on their CRM systems, for instance – they’ll find people who are highly engaged with the brand. They could be going to the website regularly, and they can naturally use those influencers to regularly boost their offline sentiment too.”

Using social media data for market research has gone through an evolution. A more limited group of people tweet, blog and post, with the conversations often being more passionate at different ends of the spectrum. People’s social media interactions are more of a broadcast, often motivated by social signalling. But the sheer volume of interactions and ability to measure them, along with AI and ML techniques, have improved the data significantly.

“We talk about turning social data into research-grade fuel; we take a lot of time making sure the topic relevance is at a high enough target range and that the sentiment accuracy is all in the 80-90% levels,” says Dodd. While there may still be some online skews, he adds, clients are often satisfied with what they consider to be “good enough, if it gets me to 90%”.

So, from Kantar’s TotalSocial work and this sentiment analysis of brands, what is the most important lesson for brand marketers? “Don’t forget that you have a lot of advocates who are not social sirens, but are talking about your brands and are open to communication,” says Dodd. “Don’t rush to the online orbit and forget that you’ve got a very useful reservoir of people talking to people.”