Cookies remember you so we can ensure to give you the best experience possible. By using this website or closing this message, you are agreeing to our cookies and policies

Do not show this message again
UK Insights

How do you measure the quality of social influencer impact?

Kyle Findlay

Director of Data Science Innovations, Insights Division

Social 07.08.2018 / 09:00


Brands are funnelling more budget towards influencer marketing as the media landscape continues to fracture and ad blocking becomes more prevalent. How then should campaigns be assessed?

We recently spoke to vlogger Lucy Moon about influencer marketing for our Future Proof podcast, and she explained the difficulties of ‘guaranteeing’ any sort of short- or long-term success for brands she worked with. “When people pay me to promote their brand or products”, she explained, “they have already done the maths – they know my follower numbers and the average number of views a video will get”. She doesn’t get remunerated based on those stats – she takes a sum up front as a content creator, and a second sum on completion of the work, as a platform.

But when brands use Lucy to spread the word about their goods and services, they aren’t simply paying for an audience of 316,000 followers to hear the news. They are hoping that quality conversations will be sparked; that the association with Lucy – or any other prominent influencer who has built an organic following of like-minded fans, and who has shared enough of their personal lives and opinions to be seen as an authentic and credible source of ‘good taste’ – will rub off on them.

What is influence?

There are a few ways in which we talk about influence. To be influential, an influencer must cause a perceptual change in others that ideally leads to action. In addition, influence is contextual: someone might be influential in the area of skincare, but less so in the area of footwear.

Given these principles, the best way to measure influence is to look at the extent to which an influencer causes actions in others. In a social media context, this means looking atactiveengagement (number of likes/favourites, retweets/shares and comments) rather thanpassiveaudiences (number of followers). And, within the context of active engagement, not all types are equal. It takes far less effort to like or favourite something than comment on it. A rough hierarchy of engagement might look something like this:

1. Likes and favourites = low-level, passive engagement on content that a user agrees with; good for awareness building and maintenance

2. Shares and retweets = a slightly higher-involvement version of the above

3. Replies, quotes and comments = highest form of engagement; best for changing perceptions and attitudes.

To illustrate the different forms of engagement, consider the informal ‘ratio metric’, which relates to a specific post or tweet. It works by dividing the number of comments you get to a post by the number of likes and shares combined. If you got more comments, this means that you caused some discussion or controversy. If you got more likes and shares, we might reasonably infer  that people agreed with what you said.

This kind of ratio is necessarily a very crude measure but it does make some sense, since people tend to like and propagate content that they agree with (although some research also shows that people will propagate content that makes them angry or fearful). This is a more passive form of ‘System 1’, auto-pilot-type engagement that occurs when the world aligns with our expectations and confirms our existing beliefs. Conversely, they they are more likely to switch into the higher level of ‘System 2’ engagement that a reply requires when they feel passionately enough about it. There are A LOT of assumptions in this interpretation and it lacks nuance, but it feels broadly right.

The point of all this is that engagement is a good measure of influence, but not all engagement is equal. Sure, you could focus on more complex measures of influence such as the network metrics behind spreading behaviour (like betweeness centrality and similar), but, in our experience, these tend to lead to the same conclusions as a simple focus on engagement does.

Finally, on influence definitions, something else to take into account is the different strengths of each social media platform. You could argue that Twitter is more about perceptions, and Instagram more about the moment of consumption, with Facebook sitting somewhere in between.

The shape of the conversation

Assuming we’ve now identified our influencers, how do we measure them? It’s one thing to look at the number of likes and retweets that an influencer gets but, as just discussed above, not all forms of engagement are equal. We often look at the different types of engagement across entire campaigns, but the principles can be applied just as readily to individual influencers’ impact within a campaign.

Let’s start with an example of shallow engagement (favourites and retweets). Budweiser’s 2015 Super Bowl campaign was the most retweeted ad in the Super Bowl. This is what the sharing pattern looked like on Twitter:

Influencers -01-budweiser

However, despite being successful in terms of the volume of retweets and favourites, Budweiser’s marketing team still banned the use of cute puppies (the main theme of the advert) in subsequent adverts because they didn’t help sell beer. It would seem that simply retweeting a campaign-related tweet is a relatively low involvement, passive behaviour that just confirms existing beliefs but is unlikely to change anyone else’s (in this case, Budweiser was trying to get more women to drink beer). Indeed, Kantar TNS said as much at the time.

Conversely, UK retailer John Lewis regularly comes out top (or near the top, depending on the metric you’re looking at) in the UK’s annual Christmas retailer advertising bonanza. Their campaign networks tend to be far messier and organic with lots of two-way conversations between consumers, which aligns well with the strong survey metrics their ads receive.

Influencers -04-john -lewis

In this case, the campaign (Monty the Penguin) won big at Cannes.

The Budweiser and John Lewis campaigns had very different sharing patterns or “network shapes”. We set about trying to quantify these different shapes and came up with our Network Shape Metric. The methodology puts network shapes on a continuum and can be applied to any kinds of network: discussions of categories, brands campaigns or those generated by influencers.

Influencers -06-shape -continuum


Initial research and development is looking very promising in demonstrating a link between the shape of a network (i.e. how organic it is) and the success of an advert. It’s based on the pre-supposition that successful adverts create organic, rather than forced, engagement.

What’s great about this methodology is that it gives us a metric with a value between 0-1 that puts each network onto the above continuum. This means we can quantify the quality of a conversation and trend this over time.

The shape of influencer networks

While we’ve discussed the shape of networks in the context of campaign footprints, it can just as well be applied to the activity of influencers in order to gauge whether or not an influencer’s content was pushed to their audience in a passive manner (less impactful) or whether they created true engagement (more impactful). As already discussed, there is arguably room for both approaches, depending on whether or not a brand is trying to build or sustain awareness with existing users versus converting new users and trying to change its positioning.

As an interesting example, one telecoms brand launched a new campaign, and they engaged the support of several influencers in order amplify the campaign. This is what the campaign footprint looked like on Twitter:

Influencers -07-case -study

Without knowing anything about the above accounts, the network looked quite organic. However, once we realise that the accounts labelled as “Influencer” acted as seed nodes for the campaign content, the picture begins to look decidedly less organic.

As we can see, each influencer managed to get their communities to retweet the content but they did not engender much two-way conversation between community members, implying a relatively passive level of engagement. For each influencer, we could work out the shape of the conversation around them. In this case, we would expect to see a hub-and-spoke pattern. Indeed, once we take into account the seed nodes in the overall above campaign network, the entire network was quantified as a hub-and-spoke shape, implying that the influencers were only able to generate relatively low-level engagement from their audiences.

Your brand must consider whether they have the right metrics for assessing reachandthe quality of the conversation (like the shape of the network). These metrics could be added to existing models. One might even be able to get to the point where, given enough data, machine learning/AI is employed to start predicting the quality of content and messaging in advance.

Source : Kantar

Latest Stories

The latest grocery market share figures from Kantar show year-on-year supermarket sales grew by the fastest rate since November last year, at 0.7% over the past 12 weeks.

As an industry, we like to make annual predictions… but what are we getting wrong? Matt Muir tells us what to ignore, and what we’ve missed.

Jane Ostler and Margo Swadley discuss the trends and changes in media, including TV, radio, social and esports.

Discover what makes a strong brand in 2020 with the new Best of BrandZ compilation.

The latest figures from Kantar show the grocery market achieved modest 0.3% growth during the past 12 weeks.

Related Content