Brand marketers have long used surveys to question consumers on their opinions. Now, via fast-growing social media channels, consumers are volunteering these opinions, unedited and in real-time.

So can this rich source of content be mined to complement, guide or even replace the traditional survey?

Media analytics company Commetric and strategic research consultancy Incite pooled resources to test this and look for correlations and contradictions. We surveyed 2,000 people for their opinions on five major high street ‘casual dining’ brands and compared responses with an analysis of three months’ Twitter discussion of the same brands.

We found that comparing social and survey data gives us more confidence in the accuracy of both techniques – and that it can give a useful guide to what is driving social conversation. Crucially, it also shows compelling correlations between measures of positive sentiment in each data set.

While the study suggests that simple Twitter ‘snapshot’ analysis cannot totally replace the traditional survey, it has opened up further interesting lines of enquiry.

Here are some topline conclusions.

Brand equity measures are possible using either approach

The number of people indicating that a brand was a favourite in survey data closely correlates with the balance of positive/negative sentiment about that brand in social media. This type of brand equity measure can therefore perhaps be carried out, for this sector at least, more quickly and cost-effectively using social media analysis, with a fair degree of confidence.

Consumers tweeted about brands they love and experiences they hate

The study suggests that if people are generally loyal to a high street dining brand, survey responses about their relatively recent experience of the brand (i.e. not immediately after consumption) are more likely to reflect their overall opinion of the brand. This is perhaps because the survey process encourages a more reflective, formal response or because a ‘one-off’ experience has simply faded. However, tweets are informal and tend to describe people’s feelings during or immediately after the most recent experience, with consumers posting at the venue or very shortly after. This difference between the psyche of the individual at the time – and in the context of the response – makes it harder to use a snapshot measure of social media opinion to predict overall customer satisfaction, at least as it is traditionally measured.

Social media can be the canary in the mine

If it is true that consumers are more likely to tweet about brands they love, and experiences they hate, then this study suggests social media could be used as a ‘canary’ in the mine, a warning of trouble to come, or of opportunities to exploit. Immediate analysis of conversations about experience and products could be valuable in helping marketers improve service – and tweak survey designs.

Survey data about social media is a good proxy for actual social data

An interesting validation of the survey method is that the number of people who claim to have heard about a brand on Twitter strongly correlates with the actual amount of mentions of that brand on Twitter. People aren’t simply making it up or mis-remembering. This suggests that either approach can be used to understand the presence of a brand on social media.

Conclusions and next steps

The study has numerous interesting observations for individual brands in the casual dining sector, ones that may well prompt additional lines of enquiry. At the aggregated level too, the very promising correlations between survey and Twitter data suggest that there is no doubt that social media studies can sometimes be surrogates for formal studies and that both types of data support each other, at least for this sector.

More work needs to be done, especially in looking at social media trends over time and at sub-elements of the ‘pillars’ which underlie the satisfaction model in the survey design, but this is a very interesting start.

If you would like to see the study in details, please click here (.pdf, 1,61MB).


A sample of 2,000 people, who had used at least one of the largest 5 UK high street brands during the last six months, were asked about their experiences in relation to food, service, ambience and value.

At the same time, over 5,000 English tweets, which mentioned at least one of the 5 brands during a three month period, were analysed to understand tone of comment and type of content.

Correlations were then established between the two data sources using Pearson.