Don’t trust all the trends

abhijit maheswari
5 min readJan 30, 2021

Because what you see could have been hand crafted. With right mix of talent, imagination and effort, one can attempt to create that social media ‘trend’. When someone tells you an event or a person is trending, ask in return — is it ‘organic’ or ‘hand-made’ ? With lot of science and a bit of art, a data scientist should be able to tell the difference.

Pockets of influence

Do you know companies are training people to be influencer on the social media ? Like an university, they are churning out ‘skilled’ digital influencers. As a matter of fact, there are online courses which have popped up to create this labor force. The brands are also encouraging this new supply with their insatiable need to have swarm of people talking about their product. Why am I talking about it ?

For one, I work with social media data for large brands and when you start listening to some conversation to build a context, you get this funny feeling that something doesn’t feel right. It doesn’t look random at all. Building a model of what people are thinking is challenging but highly rewarding. One of the impediment is the mix of influence (paid or directed) in the social media conversation leading to bias in the data. One can explain some insights that come out as work of ‘some’ digital agency — their hand prints all over. You will be surprised how apparent it can be. To give you an example, I was once analyzing sustainability (around single use plastic) conversations on social media and found a company which showed up everywhere. Shall I say an over zealous agency did its job too well and manufactured the trend ? Often we will make a mistake of treating this noise as signal and social media is filled with it. And influencers along with the paid promotion are big part of this problem. Imagine for a second, a scenario— 100 largest digital agency from around the wold coming together and coordinating their strategy. Do you think ‘the social media’ would be compromised ? If yes, you will stop believing any insights from social media.

If not already, artificial intelligence will create the ‘next’ trend. Having worked with graph theory in my research role at autonomous vehicle startup, I see how this can also be achieved in practice. One needs to have just 50–100 ‘right’ social media accounts and together with some coordinated action with optimum sequencing among them, it can snowball into a trend or controversy or a viral thing. To remove the human judgement, reinforcement learning (learning by trial and error) or graph network can be used to build a learning model. The selection of the ‘right’ social media users is akin to identifying the right pressure point in the building which you want to bring down with explosives while leaving neighboring structure intact. There is nothing random about the selection.

Second is more behavioral, why would a consumer trust a ‘trained’ influencer ? It beats the purpose. Influencers in an ideal world are people who have made some contribution to the world, have built a following for their work and what they stand for. So they have built a trust. With influencer academy, we are creating a trained machine which is peddling from everything — cookies, juice, shoes, mobile phone or noodles. And will switch their allegiance on a hour’s notice depending on the fee. The fact is that these influencers might not know the product or care about the brand any which way; they are mercenaries who have been hired to point the gun at anyone, even their old benefactor. Would you seriously consider them influencer? It might be working, but will it continue to work as supply is ramped up. When everyone is influence marketer than no one really is. The music will stop sooner than later. Remember the link sharing in the early days of internet to get your website traffic before google changed the rules of the game. What will trigger this change is anyone’s guess.

The concept of influence marketing took shape with the rise of social media in the last 10–15 years. Belief that someone leveraging his following and lending his voice is more credible than direct marketing like banner ad or TV campaign. It’s true. Influencer’s persuasion is considered indiscernible and audience is nudged rather than pushed. It did work, audience did get influenced. In the early years of social media, marketers, brands and agencies were also experimenting with the new medium in creative way. Today, more money is spent on digital marketing and a good mix of it goes to influence marketing. Every digital agency has ‘influence marketing’ as part of their sales pitch. Digital agency have overplayed this and brands have played along. Or is it other way ? Irrespective, brands will soon see that ‘successful’ influence campaign not translated into more sales. They will start seeking answers. It will demand the audit of the influence network. On a positive side, it is still an army of human influencer and not bots.

Where do we go from here, especially with influence marketing ? It will not be as easy because you have an army undergoing training. Preparing to be deployed by the motivated agency. There could be fatigue with influencer when they see too much of them. Consumers once start realizing influencers role and become more aware of their interest, might become less prone to their persuasion. Social media might be forced to label influencer (or paid post from them).

Question begs to be asked. Who are these ‘new’ influencers ? All of them are not your celebrity with their fan base but are housewives, men, boys, girls who are lured with the promise of passive income. These incomes are coming out of brand’s budget. Their efficacy will be known in time. This web of players and their machinery has become twisted and complex; its now a question of trust. Your friend might talk about new chocolate he loves and you might later learn that he is an influencer. Who do you trust in social media ?

We are so in love with the ‘trends’ on facebook, twitter, instagram, that knowing or unknowingly a story of trend can create a self-perpetuating trends. And the truth is, we start believing this to be true. An analysis which doesn’t see trend is no good. Like every new strategy, if you overdo it looses its prowess.

This story also appears here



abhijit maheswari

I like to decipher patterns whether it’s from robot sensor or digital footprints left by humans. AI, Robotics, Mathematics & Poetry interests me