How Brainly's social learning model is changing education through community - and algorithms
- Summary:
- I'm sick of reading about startups obsessed with hyperconvenience. But building an online community to change learning and help students stay on track is a venture I have time for. Here's what i learned from Brainly in New Orleans.
Too many young people get caught in the gap between our flawed educational systems and the "narrow hiring" of corporate employers. On diginomica, I've focused more on critiquing talent practices, but that doesn't mean I'm giving schools a hall pass for the social intimidation, bullying and archaic curricula that too many students lose themselves in.
I'm always on the lookout for "education tech" startups that can provoke me from despondency over this state of affairs. Such is the case for Brainly, a social learning startup specializing in community learning for middle and high school students.
At Collision 2016, I had a compelling talk with Brainly's Katie Sagan, Head of Strategy and Operations, and Erik Choi, User Insights Resource Manager. When I pushed them for funding news, Saga and Choi were elusive, giving me a wink-and-nod. Soon all was revealed, with Brainly's May 12 announcement of a $15 million Series B Funding Round.
Then, on June 16, Brainly acquired fellow peer to peer study help platform OpenStudy, which has 54 million learners. Add that to Brainly's 60 million monthly users across 35 countries and 12 languages, and you have a critical mass of online learners. At the time of our interview, Sagan and Choi were amongst eight New York based Brainly staffers. The company, originally started in Poland, also has 70+ employees in Poland.
At the time of my chat with Sagan and Choi, they were doubling down on their U.S. presence, hoping to move from two million monthly U.S. visitors to ten million by the end of the year. The numbers aren't pipe dreams: in Russia alone Brainly has an estimated 70 percent market share in peer learning.
Peer learning - how do you surface the best content?
Peer learning isn't as simple as it sounds. Brainly has pulled it off with its 1,500 or so community moderators, all volunteers. But even Brainly's volunteers couldn't keep up with a system that claims to provide an answer to a student's question within ten minutes. And how do you ensure that the answers are accurate and not wrong or intentionally misleading? That's where gamification and algorithms come in, something Choi's team is toiling over. Other big issues include Brainly's monetization model, as well as ensuring safety of young students in a sadly not-safe world.
The first question is the obvious: why bother with a peer learning network when search is ubiquitous? Sagan:
Why do students come to Brainly as opposed to going to Google or Wikipedia is a question we get. The reason goes back to the community: students tend to learn better from each other. They speak each other's language. A seventh grader generally learns better from somebody within the same age group.
The next question is the kicker. The right answer turns you into Google, the wrong, into Yahoo: how do you surface the right information to a student's query? Brainly tackles that through three aspects: moderation, voting/gamification, and, increasingly, machine learning.
The volunteer moderators, all from the Brainly community, are the backbone. The best part? As Sagan told me, a lot of them found success on Brainly they might not have gotten in school:
They're all volunteers - they just love Brainly. A lot of students have really gained a certain confidence in being leaders on the site. It's really helped them thrive. Where they might have not been so cool in school, they're actually really cool at Brainly. People want their attention, they want them to answer their questions.
But with Brainly's volume, moderators would get swamped without automated elements. Students can upvote the best answers; their own activity is gamified into points they can acquire within the system. In turn, all this site activity now feeds an algorithm that helps identify the best answers.
As of our talk, the system worked like this: when an answer comes in, it's given an algorithmic score. If that score happens to be, for sake of this example, a 9 out of ten, it gets published directly to the community. But if the score is more like, say, 5, it gets held up in queue for the human moderators to review. Currently, 70 percent of the answers go through automatically. And of those 70 percent, Brainly's research finds that the vast majority (80 percent of those approved) are considered good or best answers.
Though the algorithmic part is in the early stages, Choi's team is pleased with the initial results. They built an algo based on twenty different data points, including community activity, history on Brainly, quality of prior answers, and your points and ranking. Those questions inform the algo's filtering of which answers get automatically approved and which are sent to moderators for review.
The algorithms also filter out profanity. Interestingly, correct answers can also be flagged, if they don't provide any detail. If you just give the letter to a multiple choice question, without explaining why it's correct, your answer won't get through.
The wrap - on business models, safety, and an inspirational surprise visit
Brainly has work ahead. Though they have clearly solved the adoption problem, they don't have a monetization model - yet. Two things they do know: they will always provide a free community to support learning, and the monetization model WON'T be advertising. We discussed monetization ideas, including paid curricula from professionals, matchmaking services for students with topical needs/interests, and social graph data. Sagan is open to possibilities, as long it doesn't change the core:
We absolutely want to add something cool, because we're very mission-driven, and so we want to provide high quality access to information for every student in the world. That's the end game here. We really can't walk away from that mission, so we have to find something to add.
I'd add one more thing: mentors. While it might not lead to monetization, an "ask me anything" approach to bringing in mentors/special guests to advise on careers would be a fantastic add. This gets back to an ongoing piece of the diginomica skills discussion, with Den advocating for the lost art of apprenticeships.
Or, as I said to Brainly: "My immediate stress is that I don't know how to finish my homework tonight, and Brainly can help there. But my long-term stress is that my family life isn't that great. I'm not really sure what I'm ultimately good at in this world. If I had motivation that I was heading somewhere, it might give me a little more boost to finish that assignment." Eventually, those "ask me anything" mentors could be Brainly alumni coming back to where it all started.
We had an extensive talk about safety. I won't get into that here, except to say that Brainly clearly takes the safety of its students very seriously and is always working to bolster safety further. There are some limits: you can't do background checks on the teens who volunteer, for example. But there is plenty you can do, and Brainly works closely with parents and teachers to get their buy-in. As they push for monetization, they'll need that buy-in.
Startups have bad days. But one bad day came with a surprise. Brainly got a call from a moderator in Texas:
They called up our office in New York and said, "My dad and I are coming on vacation, can we stop by?"
So the student, who is home-schooled, showed up at Brainly's New York City office with his dad:
We said, "Yeah, absolutely come by." He's about 13 years old; he's a superstar on the site. Erik and I were sort of talking to his dad off to the side... His dad said, "I just can't even tell you what Brainly has done for my son."
And:
Katie: I get the sense that he's sort of an exceptional student. He's very, very smart.
Erik: He's been answering more than 1,000 questions on Brainly.
Katie: His dad said, "As a father, to see your child succeed at something, you just can't even imagine what that feels like."
Erik: We almost just broke down.
Katie: I mean, the both of us were near tears.
Erik: It was a bad day.
Katie: It was a bad day. Hearing that, it was like "Okay, I know why we're here every day."
Yes, that will turn your day around. It goes without saying I am rooting for Brainly to solve their business model and remain a part of the schooling solution. No, they probably can't fix the educational system. But that's too much to ask of any startup. We've already got plenty of startups fighting over funding to incrementalize convenience with apps that let you order razors while you're in an elevator heading to a client. Give me a startup trying to solve the monster problem of education anyday.