With word of mouth and personal endorsement carrying ever greater weight in our purchase decisions and consumers increasingly willing to share details of their lives and preferences on social media, many brands are asking themselves:
- Should we be using social media as part of our research and insight strategy?
- How can we use it effectively?
Arguably the principal concern brands have with incorporating social media into their research and insight strategies is whether social data (i.e. the comments made on social media and statistics derived from them) is representative of their customer base and target market.
Social data is often not representative of a target market, at least not to the same level of accuracy or reliability as survey data. However, the common presumption that data is only useful if it is representative is, I think, flawed. Brands that dismiss social data as suitable only for use as an early warning signal or for anecdotal quotes miss out on the benefits social data can bring.
If used effectively, social data can add value unavailable from other datasets, such as surveys, which are traditionally used by Insight teams.
Is social data representative?
Even where the demographics of social channels appear to correspond to a target market, social data is unlikely to be representative of that target market because:
- Not everyone is talking. Freely published statistics on social media users are a starting point for brands seeking to match demographics for social media channels to those of a target market. However,
a. some social media users just watch, they never post comments themselves;
b. other social media users vary in how much and how often they post; and
c. user demographics do not generally contain information as to the amount (or lack) of data generated by individual social media users within a demographic.
It is therefore difficult to know whether demographics of a social dataset reflect user demographics, and, accordingly, whether a social dataset is representative of a target market or not.
- Not everyone is talking about your brand. Because social data is unprompted (unlike, for example, data from scripted survey responses) it will tend to reflect the views of those who feel they have something to say on the topic (whether a particular opinion, experience or general interest in the subject) and are therefore more likely to comment on it – online or offline. In practical terms it would probably be incorrect to assume that because 20% of social media comments about a product are negative, that 20% of the market is negative towards it.
- Social data cannot be weighted. In most cases subscribing to a social media account requires only a name and email address. By contrast, survey respondents are generally asked to provide some additional background details (for example age and salary bracket information). These additional details allow the data to be adjusted to reflect the differences between a respondent base and a target market.
Why is social data useful, if not representative?
Although unlikely to be representative, if used effectively, social data can add valuable insight not available from other datasets.
In particular, unlike other datasets social data is:
- Real-time: Social data can be collected in real-time and collated relatively quickly compared to survey data. Brands are increasingly leveraging social data to adjust campaigns in real-time to increase their effectiveness.
- Low cost: Relative to other datasets, the investment required to collect and analyse social data, which is publically available, is low
- Specific and detailed: With social media users often referred to as the world’s largest focus group, the qualitative nature of social media messages contrasts with the tick box approach of surveys and polls, where open-end fields tend to be limited by budget, time and resource. This qualitative detail in social data can help to convey reasons behind statistical results, making them more actionable than would otherwise be the case.
- Unprompted: Social data shows what matters to the consumer. The data has not been generated because someone has been stopped in the street and asked a question; the data is there because it’s something the market wants to express an opinion on.
- Frank and unbiased: Social media is a platform where consumers write frankly on all manner of topics. This gives brands access to unskewed data which can offer unparalleled insight.
How can social data be used effectively?
The key to using social data effectively is to accept that it is unlikely to be representative, and to use it as either:
- a complementary source of information alongside other representative datasets; and/or
- a discrete subset of information from which accurate trends and insight can be drawn
Both approaches address social data’s unrepresentative nature, rendering it a resource from which valuable insight can be safely and reliably drawn.
Using social data as a complementary source of information alongside other representative datasets
Datasets such as surveys can give an accurate representative view of a target market, but can also be limited in the detail they provide and how quickly they provide it. Used in conjunction with representative datasets, social data:
- Adds the detail and qualitative information to validate and supplement findings and make these findings more actionable.
- Fills the time gaps associated with traditional datasets: If a survey is fielded quarterly, there is no need to wait three months to find out how the market has moved. Social data can be used to track changes continuously and in real-time.
- Increases the relevance and use of representative datasets by providing guidance on the best questions to ask in prompted research, and the best language to use when asking those questions.
Using social data as a discrete subset of information from which accurate trends and insight can be drawn
Social data can also be used as a stand-alone source to show:
- Trends in consumer conversation, attitudes, and/or behaviour by analysing how social media conversation changes over time relative to its own baseline. For example, LOCOG examined social data trends in the year leading up to the 2012 Olympic Torch Relay to gauge the impact of related campaigns and messaging, and used the findings to support product development, increase messaging effectiveness, and understand expected attendance.
- How consumer conversation around a topic compares to and benchmarks against conversation around related topics, for example, how conversation around a new product compares to those around previous and competitor products. During the 2012 Olympic Games, LOCOG compared social data relating to individual venues within the Olympic Park against one another to ascertain, if, how, and why spectator experience varied across venue, and used those findings to then implement changes on a daily basis to improve spectator experience.
Given that social data tends to be relatively and directionally accurate within a particular topic, it does not necessarily matter (for these purposes) that the data is unlikely to be representative. It provides Insight teams with the information they need to identify trends in consumer attitude and behaviour to inform a range of areas from market and competitor analysis and product development to marketing content and campaign effectiveness.