Disruptions are happening across all industries today. Companies like Google are returning near accurate search results in a fraction of seconds, digital assistants are becoming smarter by learning usage patterns, cab sharing app companies are dynamically forecasting demand and e-commerce companies are shaping demand of the consumers with personalized recommendations.
The first mover advantage comes at a cost of expensive education, steep technology curve, and significant business risks. There is enough evidence to support both of the paradigms, the first mover or fast followers. For instance, AWS introduces bleeding edge industry first IAAS products every year, risking business and getting success in almost all of the products. Swiggy, on the other hand, stayed nimble, focussed on championing customer service and building solid technology infrastructure for buttery smooth food ordering experience while the first movers in the industry eventually exited the scenario.
However, it’s paramount to understand the fact that no one remembers the guy who gets a head start in a sprint or is leading a sprint, everyone remembers the guy who crosses the finish line first. The right question here is to contemplate is how to gain momentum in the efforts to reach a breakthrough in a disruptive idea and build a product.
The hyper-growth startups like Uber, Paytm, Swiggy have the real advantage of moving fast, breaking things, learning and iterating over products. When some market follower tries to replicate a working idea in a niche market, they are practically creating a spinoff. When the companies keep innovating the gap between the competition widens and the first mover gets the advantage. Whereas the fast follower has the luxury of learning from the competitor’s experiments.
The engineering efforts across the tech companies can be classified into two major parts, one is building rule-based systems, other is learning based systems otherwise known as AI/ML systems.
These AI/ML systems use a primitive called models, which are mathematical expressions that sit in between the conjunction of computing systems and prediction of the happening of a real-world event.
Running machine learning models in production for a hyper-growth company is a technically complex and excitingly challenging at the same time.
Building industry first and new machine learning models require significant investment in engineering cycles in the human effort as well as a product release.
Hyper-growth companies with significant cash reserves for R&D and established product portfolio and customer base can have the luxury of experimenting building AI/ML-based products for mass use and become the first mover. Even if the product fails, the business has a product portfolio to fall back to.
Startups with a crunch in engineering bandwidth will be prone to business and financial setbacks if a product fails after bandwidth draining development cycles expended for building AI/ML-based product.
Being a fast follower enables companies to enjoy the benefits of battle-tested AI/ML algorithms and models, understand the market demands better and build functional superior and robust products at competitive pricing models.
While it’s great to be able to build bleeding edge AI/ML algorithms and models, it clearly needs added resources, persistent engineering efforts to succeed. It is equally great to use the existing algorithms and models invented by others and solve a customer pain point that the first mover is not able to.
Further read – My experience interacting with Datathon participants organized by Yes Bank.