IMAGE: Venimo — 123RF

What if the key to freemium models lay in machine learning?

This is a subject I have been exploring for some time and one that is leading me toward creating a classification of business models based on their potential for exploiting data through machine learning.

What if the key to many freemium business models — based on the combination of a free and a premium offer in exchange for payment — was in developing that service through analytics and machine learning?

Imagine, for example, a newspaper: could it be possible to offer news to all users for free, but also to create an additional offer of services in exchange for a payment to those users who see value in them ? How many users would be willing to pay, for example, for a system of alerts really based on their interests, based on everything they read regularly, or that brought new things to their attention? What would be the key allowing a newspaper to offer these alerts accurately and in the certain knowledge that what is being recommended is really what the user is interested in? The answer could be by generating and analysis of data, something a free model should be able to offer. At what point do Amazon book recommendations cease to be perceived as an incitement to compulsive shopping, and begin to be a valuable or even fundamental service to an expert on a particular topic?

Would we pay for a a music service able to create a play list for a particular occasion, ensuring that the chosen songs are to our tastes? Would we be willing to cough up for a Twitter that separates the signals that interest us from the rest of the noise, or even make the appropriate checks and prevent us from retweeting something dumb? With the right database, there is no doubt that this is something that a good machine learning algorithm would be able to do well. How many apps would improve their value proposition to users if they properly processed data? Some apps, unable to deal with the analytical procedures involved, sell their data to third parties. This could change thanks to the progressive development of applications like MLaaS, or Machine Learning as a Service in the cloud that allow for infinitely simpler management without having to hire dozens of data scientists.

How many of the free services we use today could incorporate a payment tranche with products built from analytics? Obviously, not all or for everyone, but would such an approach sweeten the premium portion of some of those services? To what extent could a business model based on freemium benefit from an adequate knowledge of the possibilities that analytics and machine learning offer from their data?


(En español, aquí)