Traditionally, language learning software like Rosetta Stone has been very expensive. Duolingo offers a free language learning service. As students improve/advance, Duolingo starts to use real sentences for teaching / translation. Because the other side of their business offers language translation services, the app can surface client's text to students and once enough students have attempted to translate it, their software compares and returns the best match to the client. [1]
Equally interesting, its founder previously started Captcha, which cleverly took the "prove you're a human" opportunity to have humans OCR books and street signs. Since the Google acquisition, I think ReCaptcha is now primarily used to improve Google Maps data.
I believe Duolingo has started exploring other revenue opportunities as this hasn't been as profitable as originally expected, but the idea is/was clever.
[1] https://unicornomy.com/how-does-duolingo-make-money-business...
An unusual business model I've always liked but never got around to trying would be to have a stock trading game on iOS or Android where people bet on the directions of hypothetical charts (that are secretly derived from historical data) and compete for the best returns. Give the game away for free with no advertising, mine the results for outsized performers, and if you find any, then secretly feed current market data to them and trade off their bets. I'll bet there are lots of other possibilities for getting useful work done for free by gamifying it.
I'm still not sure wether this is outright genius (most people overestimate the number of times they will actually go) or really dumb (your heavy users train for free).
The lesson: a good business model is to come up with a subscription product that's aspirational, where people will feel like they're giving up on their dreams or giving up on becoming better if they cancel, because many people will continue to pay you month after month rather than face the sorry truth.
I have a meta-model I use to determine the mechanism of how a given company makes money. Simplest is the service model. Most complex is arbitrage. There are 3 in between. Variables are the scaling factors.
Question isn't "how does it make money," but rather, "of possible transaction types how must it make money and how do its factors scale?" There is a short list of ways it must, and then finite factors of how it scales.
Most clever to me is Renaissance Technologies who appear mainly to be an arbitrage factory. Ad-tech markets are closely related. Clever in the sense that their business is about buying clever and applying it to a limited class of increasingly esoteric arbitrage problems. From what I can tell, investment banks are essentially institutional casinos where you take a set of street level hustles and conceptually walk them back until they fit into a regulatory framework. Beating that game would be a lot of fun.
Market makers like uber, airbnb, ebay and AMZN are the next smartest. A company like sharetribe is a good example of smart market making, I hope they do well.
FANG companies are essentially publishers, and their business is about demographic pandering and selling viewers to advertisers/marketers.
The other categories have different structure. Separating the business model from the problem and solution gives insight into just how clever they really are.
The product (individual domains) advertises the service (domain backordering).
Sort of reverse-engineered marketing.
