Table of Contents
1. Introduction
Internet has revolutionised and democratised the ability to express oneself, communicate and collaborate with anyone in the world, and conduct commerce beyond the limitations of physical distance.
One of the artifact of this democratisation is the emergence of influencers across various social networks. These influencers command a celebrity status and has created a system where niche content is being produced and elitist hegemony is being questioned everyday.
I decided to make a foray in this era of user content generation and decided to become an influencer.
2. What problem are we solving
So the key objective I set for myself was to become an influencer. To make the objective more concrete, I setup a target of achieving one million followers, 100K daily views, and 20K engagements with my content.
This is nothing new and I suppose every aspiring influencer takes a similar journey. However, what differs for me is the fact that I am a geek and in the best case scenario, a travelling geek who loves food. It should be pretty evident that I don’t have skills or looks for creating naturally viral content. But what I have is an ability to pose any problem as a mathematical modeling problem and solving it.
3. Are we SMART?
One of the first thing that we need to ensure before achieving any significant endeavour is to ensure that our goals are SMART. This basically means that the goals should be *S*pecific, *M*easurable, *A*chievable, *R*elevant, and *T*ime-bound.
So if we look at our goals discussed in the previous section, our goals are specific and can be measured. The only thing we need to make sure is to have a timeline and also be sure that these are achievable. In my case, I do believe that the goals are achievable as I am trying to reach 1M followers out of 2B users on Instagram. We can do more Fermi estimates to make ourselves doubly sure that this is something that we can achieve.
The key question that needs an answer is what is the potential size of our audience who will be interested in my content. Although I don’t have rich market research data, I can leverage various pieces of information made available by Instagram to estimate the size of potential audience, as discussed later.
Final question that remains is whether the goal is time-bound and yes it is. I want to be an influencer over next 180 days. With this timeline in mind, for all practical purposes it seems that the goal is unachievable. This is because if I want to achieve the 1 million followers target, I need to approximately 6000 followers per day for 150 days (assuming I get first 30 days as a bootstrap period).
4. How to Approach the Solution
4.1. Content Problem
To solve the problem we need to first identify the key components of the solution. Since we want to be an influencer, the most important component of the solution is the content generation. The content part of the problem can be further divided into two problems: what content to post and what is the presentation. We will address the what content and its presentation later in the series.
4.2. Audience Discovery Problem
The second key component of the solution is how to ensure that our audience can explore our content and we can drive engagement.
4.3. Algebra and Statistics
To approach a problem from machine learning perspective, it is important to establish the algebra of the solution and whether the statistics of underlying data ensures a sound solution.
4.4. The Algebra
In my solution, the abstraction that built is that we have a set of users denoted by 𝕌 and set of contents. Since I am primarily concerned with my content, I want to distinguish between the content generated by other users and content generated by me. I use 𝕆 and 𝕄 to denote the set of content generated by others and myself respectively.
To develop a mathematical model, I assume that there exists a bipartite graph between the set of users 𝕌 and set of content 𝕆 ⋃ 𝕄. An edge of this graph denotes the affinity between the user and content. We can associate different semantics to the edge for example, it could be the probability that the user will see the content, probability that user with engage with the content and so on.
It should be noted that although I assumed a bipartite graph between users and content, we will extend the formalism depending on the iterative solutions.
5. Next Step
In the next part, I will start with exploring the problem of audience discovery.
Created: 2022-09-19 Mon 04:23
No comments:
Post a Comment