Quora was once upon a time supposed to be the next $10 billion company. With an extremely accomplished founding team including Adam D’Angelo (former Facebook CTO) and Charlie Cheever (lead on Facebook Connect) and exponential early growth, it seemed destined for inevitable massive success.
Moreover, the cultural impact of the company outstripped even it’s monetary potential. Early Quora seemed to be the platform for intellectuals to congregate and debate trends, and for a curious person to get a series of well-thought out viewpoints on almost anything. It was an almost a mix of modern Tech Twitter and what ended up becoming Medium. Even today, there’s archives of these wonderful early days; Keith Rabois and Chris Dixon duking it over which VCs are best or Sizhao Yang describing the mechanics of what made Farmville successful (it was 2011, Farmville was really important).
And yet very few people talk about Quora today, even as both growth and content levels have exploded. It’s not Foursquare; the valuation has gone up, albeit at a slower than expected rate to about $1.8 billion during it’s recent Series D. Users have grown from 100 million in early 2016 to about 300 million this year while it racks up nearly a billion monthly views. And yet when you talk to most people in tech, Quora is considered an afterthought. As I was writing this, I saw a tweet that probably explains this best, ironically from recent round Quora investor and VC Kanyi Maqbuela of Collaberative Fund:
In 2011, Quora was the logical next step to creating a more sophisticated Internet, an Internet that would remove global information asymmetry and create a personalized education system. In a way, the disappointment of the unrealized potential of the Quora mission started to outweigh the quantitative growth results of the platform, with the economic potential remaining unfulfilled for almost 8 years. The perception of Quora also reflected the shift from optimism to pessimism for the net value that connected networks on the Internet could bring about.
At some point, Quora even started to vaguely resemble the platform that it started out to replace, Yahoo Answers. The overall quality of content started to decline, early thought leaders abandoned the platform in troves, and it turned into a distribution channel primarily powered by a strong SEO presence.
So what went wrong at Quora?
Confusing Language Segmentation for Cultural Segmentation: Quora started to grow internationally really fast, especially in large markets such as India. And having a connected global Q&A platform open to anyone was fundamentally part of Quora’s goal. Internationalization with single Q&A base that everyone could contribute to would lead to increased engagement, growth numbers, and contribute to the mission. But at some point, cultural differences became a net negative: language barriers led to grammatically flawed low-quality answers, popularity of US irrelevant questions, and a whole lot of noise. Going to the “in different language” regional then globalize approach instead of vice-versa might have worked better. Quora wasn’t Facebook or Instagram; high level text conversations didn’t scale globally like pictures or statuses.
It’s honestly not for lack of of technical controls and in the moment of “optimize for engagement” social sites, a relatively rational decision. But retroactively, the decision to build a single pool of questions that anyone could or read/write instead of building regionally was probably a mistake.
For example, if I search “how do I get an internship at Google”, the first result is a great answer from a Google Recruiter. But the next six results are from Indian students about how to move from IIT into a job at Google Hydrebad. Tangentially related, but not useful for me.
Too Many Questions, Not Auto-Filtering Some: Quora actually does a great job of filtering on a micro-level. It made a decision early on to use ML to essentially remove repeat questions, by merging them to make questions as general as possible and filtering out spam, or extreme down-voted answers. Also, the PeopleRank algorithm is a good filter rank for distributing in-question influence among answers and the community filter, both through moderators and the implicit voting system.
On a macro level, there’s just too many unanswered questions, and too many dumb questions on the site. It’s easy too hit a wall when trying to stay engaged once on the site, either by hitting two blank questions in a row (which there are a lot of), hitting a question with out-dated information, or low quality questions. Personally, I think putting an expiration date on unanswered questions is warranted; the % chance someone answers a 2 year old question vs. the number of people who hit that wall must be pretty lopsided. Even a design solution of assigning a basic /10 quality score on possible click-through related questions might significantly improve the bounce rate.
In a way, I’m not even sure the overall degradation in quality of questions at scale is something that was at all Quora’s fault. It’s pretty hard for a machine at scale to quantify that the question “What’s the Scariest Thing That’s Ever Happened to You?” is a dumb question as long as engagement levels are high. If you ask users to rate the question, what’s perceived as “low-quality” will vary greatly across a heterogeneous viewing audience. And it’s a shame, because the personalized recommendation system is technically excellent; it generates high-level general content for the user, but there’s too much noise on the platform.
Shitty Embedding: This is something pretty minor, but has always bothered me about Quora. Content sites will essentially reprint great answers from Quora as articles all the time, with minimal branding from Quora. Quora presumably gets some type of affiliate fee every time this happens, but why not create an easy embeddable API like Instagram and Twitter which increases branding and encourages users to click-through to see second-level responses or different perspectives? I would argue the long-term growth boost probably justifies the short-term revenue.
Giving Up on Blogging/Publishing: Quora launched a blogging platform that easily integrated with the rest of its site when it had a strong distribution advantage in early 2013. It was actually brilliant in theory; you could create a brand through your blog, and then have the option of scrolling through a bevy of other relevant posts through the same topic-based ML algorithm applied for recommending questions. Once you had built a brand through blogging, you could extend your brand through shorter-form answers when people requested you for questions. And it turns out the demand was there too. It’s simple, Quora just didn’t pursue it hard enough when initial results were off because of some solvable product flaws (i.e. you needed a Quora account to even sign-up when it had <50 million users), and got beat to the punch by Medium.
But it’s not like Quora is a hopeless cause; there’s still the unfulfilled potential to become a giant. In other words, the company is more Kristsaps Porzingis than Michael Beasley. The company optimized for mobile early and the product design is at an A+ level, a critical mass of the world’s population have accounts on Quora, and the distribution advantage is here to stay. Yet as the company moves into the second stage, there’s more questions than answers (no pun intended).
Key Questions About Quora:
What’s Better About Our Ad Platform? What’s the Back-up Monetization Plan? Quora’s monetization attempt through advertising is going to face the same questions that Snapchat and Twitter have, which is what monetary advantage is this giving me over doubling down on Facebook or AdWords? There’s a second level to this as well for Quora; when 85% of your traffic is coming from search engine referrals (aka Google), what’s the value of paying a higher CPC when the advertising on the original search query might cannibalize some of the highest quality potential customers?
How to best approach International Growth? Quora has now started to drastically expand internationally using a regional approach, by building a series of local sites with a shared feature base but independent content. And with treating each market individually, it gives Quora the opportunity to learn from it’s past mistakes regarding content control through updated ML systems to dominate each region, but without the advantage of the existing network effects that it has in the US. It’s something Quora has been working on for a couple of years, and only time will tell how it turns out. Given that the US is still a relatively small market, successful int’l expansion combined with decently effective monetization could help Quora reach new heights.
What’s Quora’s Acquisition Floor? Unlike most companies, Quora probably has a massive floor on what it’s worth even if it fails to monetize effectively and user growth stalls out. First, as mentioned, the high-level SEO for millions of different topics is extremely valuable and can be effectively vertically integrated into a different social or content driven site. It’s a do-now, worry-about it later acquisition problem at the right price.
Secondly, and more importantly is that Quora is a perfect acquisition target for a BigCo that falls behind in the artificial intelligence rate. The biggest impediment to a great machine learning algorithm is getting your hands on labeled, relevant real-world data to act as a cog for your neural networks. Quora is sitting on probably the most (sans Google or Facebook) labeled data of any company in the world. For years, it’s been using machine learning to label user text among other functions: once someone posts a question, it has an automatic system of detecting quality and then making question-topic labeling decisions based on the text. Given the sheer volume and various use cases of a system of labeled text data, it’s supremely valuable.
It’s kind of like how Foursquare was able to turn things around even when it started to fade as a consumer-product because their DaaS for location services was so useful.
How to Get Back From Distribution Channel to Community/Platform: Quora’s per visit engagement levels aren’t public but since a majority of their traffic comes via search instead of direct traffic, (8% according SimilarWeb) it’s safe to say they just aren’t a platform any more. There’s some mechanisms the company uses to encourage engagement loops (Daily Digest, A2A, Top Reader/Writer awards, etc.), but it pales relative to the hooks of personal-network validation offered by alternative platforms such as Twitter, Instagram, Facebook, etc. In other words, Quora needs to completely change their incentives structure. I think at some point, Quora’s platform value was it allowed “non-validated” users to build a reputation that could translate into interesting opportunities, but when the quantity of content got so big, community votes weren’t the best indicators of quality content as they previously were, rendering titles pretty meaningless. There needs to be a reason for users to come back, perhaps with skill-specific roles (i.e. a cryptocurrency hedge fund) pledging to hire based on what they see is a sustained period of excellent answers or something along those lines.
Multi-Channel Distribution: There’s so many possibilities. Voice (the Alexa App is out), Live Q&A for Podcasts, Audio-Driven Responses, National Polling, SaaS for BigCo’s. Some are mission-driven, some economic, but there’s a million directions the platform can still go!