- Start But How?
- Posts
- Simple CAPTCHAs into a $42,000/Week Goldmine
Simple CAPTCHAs into a $42,000/Week Goldmine
Yo, welcome to this week’s ‘Start But How?’
Today we will do a deep dive into how ReCaptcha turned frustrating online security checks into a money machine making $42,000/week from one client alone.
It all started back in the early 2000s when Luis, a university professor at the time, invented CAPTCHAs. You know the ones - where you have to try and decipher some distorted text to prove you're human and then access web pages?
Yeah, those drove everyone crazy but they were a hugely effective way to stop bots and spam.
Fast forward 7 years, CAPTCHA has taken off to the moon. Luis realized that close to 200 million CAPTCHAs were being solved by humans every single day around the world.
He did the math - each one takes about 10 seconds to complete. This means humanity was wasting over 500,000 hours daily just typing out those stupid sequences!
He was like ‘What a waste!’ and started thinking, "Could we get useful work out of those 10 seconds where human brains are doing something amazing that computers can't?"
And then came his AHA moment. He realized CAPTCHAs were the perfect way to crowdsource tasks that still couldn't be replicated by machines.
And this is how his new company ‘ReCaptcha’ started based on his own invention ‘CAPTCHA’.
Two-Piece Puzzle
So, what did he do exactly?
One word (maybe a couple), ‘Digitizing Analog Documents’
He figured out a two-part puzzle:
Instead of conventional CAPTCHAs, visitors were given two words to write down.
One was that the system knows what the answer is, so if you get that right, you are not a bot.
Another was a word out of a document they were digitizing. If 10+ people give the same answer, that means it is the correct transcription.
This way, accuracy was higher than even professional transcribers.
$42000/week from one client
So now, it’s time to find customers. And they got one: The New York Times. NYT had a treasure trove of over 150 years' worth of back-issued newspapers spanning millions of pages that needed to be digitized.
Luis struck a banger deal where the Times paid ReCaptcha $42,000 for every year's worth of archived content they digitized. And they got to do one year’s transcription in just one week. That’s $42000/week from just one client.
Other major companies like Google Books and the Internet Archive also paid ReCaptcha good money to tap into their crowdsourced digitization pipeline. His approach was just a lot more cost-efficient than hiring professional transcribers.
Aftermath
The growth was mind-boggling. Within a couple years, ReCaptcha had digitized over 17.6 million books and documents, over 440 billion words in total transcribed across the 200+ million CAPTCHAs solved daily.
The rest is pretty well known. In 2009, Google acquired ReCaptcha for an undisclosed sum in the tens of millions of dollars.
Now, the guy who invented CAPTCHA and founded ReCaptcha seems pretty cool, right?
Well, later on, he went on to build Duolingo (A billion-dollar consumer app that we all probably used at least once in our lives.)
Now, we have seen this model of crowdsourcing being repeated over and over again such as, in Google’s image captchas and obviously - Duolingo.
But that’s the story of another day - another case study.
And that’s a wrap. Let me know in the replies how you liked this week’s case study.
Until next week, over and out.
Startup Growth Guy