= 27d567bc2d48d9f40e9ccc20ea370b15)
jkaufman comments on Why don't more rationalists start startups? - Less Wrong
You are viewing a comment permalink. View the original post to see all comments and the full post content.
= 783df68a0f980790206b9ea87794c5b6)
Loading…= b715d02e85f7b3b4ae65ca3d3e450868)
Subscribe to RSS Feed
Powered by Reddit
= f037147d6e6c911a85753b9abdedda8d)
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (99)
Not formally, but I've talked to lots of people about my startup, and have obtained consensus responses.
My idea is to make college reviews better, by asking specific questions instead of general ones. See http://collegeanswerz.com for all the questions. For example, how does the workload impact the social life? Are the school sponsored events popular, or do they suck? Do kids go home on weekends? Are your professors generally fair? How social was this dorm? How difficult was this major? Is there a convenient place to get groceries?
A few people don't think there's much of a demand for that, but most people (>95%) think that there is a demand for that. The overwhelming criticism is that I won't be able to get students to answer the questions. I've been experimenting a lot with methods of getting students to answer the questions, and I think I've arrived at something that works: $10 to answer 25 questions, or $25 to answer 50. At a $300 budget per school, this gives me about 10-12.5 answers per question (~60 questions per school), which seems pretty good. And $30 per school of targeted Facebook advertising has been successful at spread the word of the offer to students. This has worked for the 3 pilot schools I have.
From there, I'll have to
1) Raise ~$100k from an investor to do this at the other 297 schools (cost is ~$330 per school to pay and advertise to answerers)
2) Spread the word of the website to current students. I think following the advice here will do that sufficiently. The main methods are internet ads, talking to guidance counselors (I know a couple who are interested), high schools, blogs and news outlets who talk about how to choose a college.
3) Bonus: raise VC money, and expand into things like video tours and live chat. (I call this 'Bonus' because I think that if I failed here, I'd still have a successful company, just not a wildly successful one).
What do you guys think?
Ignoring the general question of "should more people start companies", some feedback from poking around at your site:
First impression: not a real site. I'm not sure what about it sets that off for me, unfortunately. Maybe the amount of white, the size of the main picture (too small), general non-glossyness, font of the title text, homemade appearance of the logo, and the broken aspect-ratio on the main picture? Hiring a designer can help here.
The focus isn't that clear. As a visitor I can't tell what I'm supposed to be doing. Am I supposed to search for a college? Click on "colleges"? I understand it's about picking a college, but I don't see what I'm supposed to do next.
The "get paid to answer questions" looks a bit scummy and isn't generally relevant. Can you limit it by IP so it only shows up to people browsing to you from those universities? And then simplify it to something like "Answer Questions, Get Paid"? It's also a weird width.
Most of the front page appears to be selling the site. Which you need to do if you're charging for a service, but you're not. Instead you want to get people into the flow of looking at colleges as soon as possible. Save the "what makes us better than everyone else" for an about page or a pitch deck.
Minor: having the questions in a monospace font is weird. As a programmer I love monospaced fonts, but most people don't.
You have an audience issue with questions. You're going for a site where people see answers to real questions, the kind the would ask someone in person if they were socially close and wanted the real deal on a college, but those questions aren't the same for everyone. In particular, there are some questions which probably do matter a lot to some people but are irrelevant or even offensive to others. "How hot are the girls?" was the main one that jumped out to me here.
The comments take too long to load. They're not way slow, but just clicking around some it was distracting having to wait for the spinning C each time. This limits the magic of clicking around and seeing people's responses to questions.
Thanks for the quality critiques!
Regarding the design things, I agree with you. I basically just started learning programming and design this summer and I'm definitely flawed in each of them. I would like one of my first hires to be a designer, but I do think that the current design satisfices (won't wow, but won't draw people away). Do you?
Also, there are some ways in which I think my design is much better than that of my competitors. For example, on http://www.unigo.com/, it takes 2 clicks to pull up a college page (College Guide => type in search) when it should just be a search bar on the top right. Also, my site lets you browse by question, rather than by reviewer. And another one, when you're on a school's page, you could toggle the information section rather than having to click it and leave the page you were on.
I've heard this before, and am not sure what I should do. I figured that even if it isn't immediately apparent, the user would end up proceeding to click on Colleges, and then figuring it out within seconds. But brainstorming this and redoing the home page is definitely on my to-do list.
I'm not sure how to limit by IP, but that would be cool if I were able to do that.
Good point.
I know they aren't the same for everyone, and so I try to be comprehensive. But being comprehensive will inevitably lead to the presence of questions that some applicants aren't interested in. I feel that being comprehensive is more important than annoying some people with some questions.
That's something I'll need to address. In the short term, I'm already using Disqus, and I don't think it's worth the time to go back and rewrite the code to have my own users and comments and stuff so I could make it faster, but I think that I will eventually. Meanwhile, Rails 4 just came out with turbolinks, which should speed it up a bit (I still need to upgrade).
Before we get into more details, here are some higher-level thoughts on the business overall: how will people get to your site and how will you make money from them?
One strategy is to pay for ads. People doing this already are easy to find: search in an incognito window with questions like "college information", "choose college", or "college for me" (be imaginative). Click on the ads that don't look like they're for particular colleges. This drops me on a few landing pages:
You don't know very much about these businesses, but you know they're all making enough money on an ad click for it to be worth it for them to advertise here. (Looking in the AdWords keyword planner suggests they're paying $5-$20 per visitor (CPC).) If one of these sites is doing something that seems weird to you, don't write it off as dumb: they're making money or they wouldn't be able to afford to show up when you did the search. Maybe they have some low-hanging fruit in improving their layout, and they certainly don't all look the same, but if in doubt they probably know something you don't.
It's also possible for people to come to you via searches. Do some searches yourself; see who shows up. This business model is harder to replicate because it's much less clear how they got onto the front page for a search. SEO advice tends to be terrible, and for good reason: it's an adversarial game where success is very lucrative and people are constantly pushing limits to see how scummy/profitable they can get without stepping over the line and getting banned (recent example: rapgenius).
Word of mouth and viral/social are also possible, but even less predictable. But then my bias here is for paying for traffic because I used to work in advertising and it all feels very natural to me. But however people get to your site, you want a good sized flow so you can experiment and test things and see what works. Iterating with no one watching is hard to get to go anywhere. (One nice thing about the non-ads flow is that focusing on making your users happy and have a good experience lines up with your incentives. If you go the ads flow every visitor is to a first approximation someone you'll never see again and so you don't care about their experience just how much money you can get from them. It's like businesses that cater to regulars vs tourists.)
Once people get on your site, how do you get money from them? The simple strategy is to put up ads, like adsense. These pay very little, like $0.01 to $1 per thousand visits. Maybe somewhat higher because you're in a vertical (education) where there are advertisers with a bunch of money to spend (colleges). There's no way you can pay for ads at $5 per click and make your money back at $0.001/visit. So what are those sites like educationmatch doing? They're collecting information from people and selling it to colleges. This is called "lead-gen" advertising. The most obvious form is when you land on a page and it immediately asks you for information about yourself. If you're going to place search ads, you basically have to do lead-gen; nothing else is profitable enough per-visit. (When you start off you'll sell leads to a lead-aggregator who sells them to the colleges. When you get big you negotiate deals directly with the colleges and get a bit more per lead.) If instead you want to make money by putting ads on your pages you need to go with cheaper but higher-volume sources of traffic, like being awesome, getting people to link to you and talk about you, and getting high up on search results pages.
Back to specific design notes:
To actually answer this question you need data from your site and competitors to compare bounce rates. You're not going to get that, so the best you can do is calibrate retrospectively: when you do get a designer and redo your site you should A/B test the redesign so you can see it's impact. This only works if you have enough traffic by then to get meaningful results.
My guess is that the appearance will make some people leave with "this looks unprofessional/fake" but this isn't something I'm good at estimating. You should at least fix the aspect ratio on the main image; I found the vertical squishing very distracting. Cropping it instead of scaling it in the browser would work, as would using only one of height= and width=.
unigo looks more professional to me, though still not as polished as something could be. But polish isn't everything; many of the sites I linked to above aren't going to win any design awards. Many of them look kind of tacky and overly slick. They're all making money, though.
Sit down with someone and watch what they do with your site. Then sit down with someone else. Look at your competitors and take the best ideas from many sources, then mock up a new front page design and sit down with another person. Repeat until you run out of fresh victims/friends. You're too small for fully-automated a/b testing, so substitute quality for quantity and watch real users in person.
Many colleges have a contiguous IP range. For example, everyone browsing from on-campus at Swarthmore will be 130.58.something.something. Make guesses from looking at the IP of the college webserver, and if various departments that do their own webhosting (cs, math) all are in the same IP range you're probably good. You can also look through your server logs to see the visits from people you know are at certain schools. It's also possible that this can be looked up somewhere.
It's a balance. Trying to channel "teenage girl" I have the impression that "how hot are the girls" would be something I would find off-putting. But my mental model is pretty bad here, having never been one.
A final note: this all sounds kind of mercenary. My perspective is that's what you have to do if you want to make money in such a tightly competitive market. That's part of why I'm glad I don't work in this industry anymore.
That's the plan.
Ok; keep in mind that this means bringing in a lot of traffic. You're talking about wanting to build a company you could sell for $5-10M, which means either something like $500k in profit a year or the expectation of even more profit in a few more years. Your costs might be $2M/year, so you'd need to bring in $2.5M/year in ad revenue. At a $1 CPM that's 2.5B pageviews a year or 7M pageviews daily. That's 1/60th the traffic Wikipedia gets, and Wikipedia is massive, so this is a very challenging traffic goal.
via http://tech.fortune.cnn.com/2013/10/10/a-website-for-rating-big-life-decisions/
It's funny, now I'm arguing for the outside view instead of the inside view :)
If they "run largely off of ad-based revenue, especially targeted advertising", and "have 21 employees", I could too. My guess is that the numbers work because they have higher CPMs because they have such a targeted audience.
My overarching point though, is that it could be monetized, not that advertising is the best way to do it. My evidence is that all my major competitors are multi-million dollar companies, and that I think I could out-do them (more users, more engaged users, better information, more brand recognition...).
But you can't put yourself in the same reference class as these companies, because it's most likely that they started out with more resources, bigger teams, more experienced founders, and/or more connections than you.
You could make that argument against a lot of startups.
Yes, it's probably true that there are successful startups who began with less resources than the incumbents in their industries. Yet these companies may start out with more resources than you seem to have available.
There are many business models other than launching a multi-million dollar mass consumer website that monetizes from advertising. In my view, such a business is relatively risky and resource-intensive. It seems to me that you have chosen an especially challenging business model for your first startup.
Bringing a product to a mass market is a challenge; see the examples edanm and I raised of restaurants and movies flopping because they fail to resonate with the market. The wider the market it is that you are aiming for, the harder it is to make a product that resonates with them.
So, I did not intend to sound like I was making a fully general counter-argument. I want to specifically caution founders where there is a large gap between their current resources/experience, and the demands of their business model, especially when their business model requires targeting a mass market, and they have no experience marketing to even a small or niche market.
Sorry for the delay, but here's my pitch: http://www.collegeanswerz.com/pitch . It should clear things up and enable a more productive discussion if you're still interested.
Also, check out http://lesswrong.com/r/discussion/lw/jmn/salary_or_startup_how_dogooders_can_gain_more/ .