Research by IIITB says no.
Social media is ubiquitously used as a platform for discussion, discourse, opinion, and information sharing. Certain individuals in social networks are very popular and influential. These individuals can alter the opinions and ideas of their followers, and thus possess marketing potential. Moreover, information travels quickly through social networks, resulting in advertisements reaching many people in a short span of time.
Due to these advantages, social networks are increasingly being used by brands to gain exposure and advertise products. For example, Selena Gomez’s ‘Share a Coke and a Song’ photo endorsing Coca-Cola garnered over four million likes and became Instagram’s most-liked image at the time.
From the perspective of a brand, planning an advertisement campaign is not simple. Influencers are expensive; an Instagram post by Beyonce is worth a million dollars! It would be unwise for a brand to make such an investment without understanding the outcome which depends on a variety of factors including the behaviour of the individuals in the network (which has a degree of uncertainty) and the structure of the network itself. In addition, there are different kinds of influencers in terms of the number of followers they have, e.g., celebrities and nano-influencers.
Are celebrities always more successful than nano-influencers? Recent studies say, no. Nano influencers can perform quite well, especially when several of them are advertising the same product together. Thus, a brand is also faced with the questions of what kind of influencers to use and how many. As a result of the varying nature of the factors of an advertisement campaign, there isn’t a one-point solution or an explicit algorithm that can predict its outcome.
An influencer marketing survey by Mediakix showed that 61% of the marketers surveyed agreed that it is difficult to find the right influencers for campaigns. This is where modelling and simulation methods prove helpful. By designing a model that approximates real-world factors, it is possible to simulate and observe the behaviour of large groups.
Researchers at the International Institute of Information Technology Bangalore have taken such an approach. Ronak Doshi and Ajay Ramesh, under the supervision of Prof. Shrisha Rao have formulated an agent-based model that can simulate influencer advertising campaigns and subsequently help study their outcome.
The model is graph-based, i.e., the individuals of a network, viewed as vertices of a social network graph, are modelled by agents. More specifically, an individual’s behaviour is modelled by its parameters. These include the individual’s interest in a product, willingness to pay for the product, social influence, and social engagement. With this setting, a propagation algorithm simulates an advertisement campaign starting from an influencer.
Based on the above-mentioned parameters, an individual decides whether or not to purchase the advertised product. The agents could also propagate the advertisement to their connections in the network. In the real-world, this whole process corresponds to an influencer posting content endorsing a product in a social network, the influencer’s followers viewing this content, engaging with it (through comments, likes, etc.) and possibly purchasing the product.
These followers might in turn propagate the advertisement to their followers (through shares, retweets, etc.) and so on. The social influence of one individual over another thus plays an important role in the whole process. The central notion of the model is that every individual of a social network is an influencer to some degree.
In an effort to understand the significance of different kinds of influencers, a comparative study of the performance of different kinds of influencers in a social network under varying circumstances was undertaken using the formulated agent-based model. Simulations were performed using real-world datasets such as social graphs obtained from Twitter and GooglePlus.
The performance of influencers was measured using two metrics – customer acquisition cost and conversion ratio. Customer acquisition cost is the cost of winning a customer. Conversion ratio is the ratio of the number of customers to the total number of individuals to whom the advertisement has reached. Better performance is attributed to low customer acquisition cost and high conversion ratio.
The results reveal that the conventional expectation that celebrities are always the best choice to advertise a product is incorrect. In fact, the performance of different kinds of influencers depends on the circumstances. For instance, simulations show that as the nature of the product varies from luxury to non-luxury, the performance of celebrities declines whereas the performance of nano-influencers improves.
In terms of the customers’ interest in a product, we find that the performance of nano-influencers declines with the decrease in customers’ interest whereas the performance of celebrities improves. In addition, a complete sweep of model parameters facilitates visualization of the performance of different kinds of influencers in varying scenarios.
The domain of influencer marketing is relatively new with the advent of social media networks such as Twitter, Instagram, etc. It has indeed become a promising platform to advertise products. This work shows the effectiveness of model-based techniques in gaining practical insight into the progression and outcome of advertising campaigns, and subsequently developing good advertising strategies.
In order to simulate a specific advertising plan/campaign using the model, a social network of choice in the form of a graph, the corresponding customers’ interest and willingness to pay, the influencer hiring and engagement rates of the network, and so on can be provided as input for simulation. In a similar manner, several different scenarios can be simulated and compared to decide on a good advertising strategy. The model can thus also be used as a base for further studies in influencer advertise.
