Essentially none of those 30kloc were part of the actual product, just set dressing. You could easily launch the same basic product without writing a single line of code, just handling everything through e-mail attachments and a paypal button. I don't believe that the user experience would be significantly worse for it.
In the middle of the article, the OP lists various mistakes he made, but I think he's basically wrong on all counts. His essential mistake IMO was launching too late. In six months of work, he gained no insight whatsoever into the market. He could have learned just as much with a handwritten flier on the college notice board - "Your mock exam reviewed by a postgrad, £10. Email foo@bar.com".
For the technically-inclined, coding is the perfect form of procrastination. It can absorb a near-infinite amount of time and feels quite productive, but it's usually a distraction. Steve Blank's most important message is that in an early stage startup, your job is to learn about the market. Anything which doesn't connect you with your customers is wasted effort.
Not understanding your (potential) customers is often mistake #1. We spent 2 months talking to potential customers before writing a line of code and now have 1,000+ in a fairly small niche.
Maybe one day, too busy now and I don't really enjoy writing. Have some good stories though. I'd probably start with "How I turned a $60 bottle of scotch into $2mm".
The market wasn't there but I disagree that that means he should have started with little or no product.
I forget who said it, but the best advice I've heard from a V.C. is that one of the biggest mistakes they see is founders going overboard trying to avoid making the mistakes their last startup made.
That said, I had an experience at a startup that did what the author now considers "the right way." We went out and sold before we had a product built. We figured out the pitch that the customers wanted to hear and had all of our best leads ready to go. But then it took more like a 12 to 18 months to build the thing, not the 3 to 6 we were promising. Unless you've built something exactly like the current product before, there are always technical problems you don't anticipate, and they always get solved the same way: add more engineering time.
Your earliest/best sales leads are a precious resource. Do not look like a dumbass in front of them. The second batch is not as easy to find as the first or they would have been in the first batch.
Great article. I don't think we can repeat this enough:
Hard work != value. Clever code != value. Writing something hackers think is cool != value.
Sometimes I think the hardest part of startups is re-aligning our value system from what we've learned in school and society into something that's actually useful for startups.
While it can of course get annoying, hipness should not be discarded so easily.
I'd argue that the hipness of a language/tech stack is actually an important business consideration today, if only due to the short supply of good startup dev talent. Good developers have their pick of jobs right now. For instance, most of the PHP badasses I know are regularly turning down PHP work in favor of working with something newer.
After all, the smartest people like to learn more than anything else.
Creating new languages doesn't create any value. It's not solving a problem. It's just solving the same problem that has already been solved countless times before.
Very true. I've been thinking that although I started coding to make a business (tho I used to write QBasic games when I was 10 too), I developed a coder's value system.
I can't seem to keep in mind that code is a means to an end!
Someone called Humourisok posted this, but it is dead for some reason. I thought it was a good idea, so repost:
----
Humourisok 20 minutes ago | link [dead]
from the tech point of view you have an open platform to share and annotate scanned docs. Sort of "web 2.0 Sharepoint". The niche that I see for such a service is legal. Lawyers can receive, annotate and remark on clients drafts for proposals, contracts etc. So it's something like Scribd+ or Docstoc+, for closed network of service provider & his/her clients, where a lot of paperwork is required. Legal services seems like the case.
----
"A hellbanned user is invisible to all other users, but crucially, not himself. From their perspective, they are participating normally in the community but nobody ever responds to them. They can no longer disrupt the community because they are effectively a ghost. It's a clever way of enforcing the "don't feed the troll" rule in the community. When nothing they post ever gets a response, a hellbanned user is likely to get bored or frustrated and leave."
If you visit this user's page at http://news.ycombinator.com/threads?id=Humourisok you can see that their first 3 comments had negative karma, after which the hellban was applied.
I think it is automatically triggered if you reach a certain negative cumulative karma.
I was hellbanned in the past and i was not trolling. I had over 1,000 karma earned in 6 months. My average karma per post was well over 6.
I was not trolling. My posts were not getting voted down, despite often taking unpopular positions.
I was, however, taking a position that one of the moderators of this site disagreed with, and (without realizing it) wrote a response making an argument in response to a post from one of the Y combinator people.
I was hellbanned for disagreeing with the politburo. It was pure censorship because pg, et. al, cannot tolerate people who disagree with them on a political topic. Ironically, I was taking the pro-startup, pro-capitalist position.
Since people are aware of hellbanning, they should be aware that it is used to silence people for disagreement, not just against trolls.
Could you share your prior username? I'm hesitant to believe you because every [dead] comment I've seen so far has been by a user with a net negative karma.
I probably thought about 3 ways he could use the tech and apply it to another problem. He should not quit, he should pivot his startup. I think there is something here. I for one, will download, install and play with the app.
Word processors already support annotation and change tracking quite well. In the legal community are there still frequent dealings with scanned images (i.e. not the actual doc files)? If so would something like this be more practical than using the word processor functions?
On a moment's thought I'm guessing that perhaps yes, there might be a need for this, as responses to discovery might be in the form of boxes of paper rather than digital files, for the very purpose of making shared review and analysis more difficult.
"you have an open platform to share and annotate scanned docs."
I'd run with that!
Is anyone else running with that?
I mean sure you can annotate in acrobat reader (since Acrobat Reader 9), and sure you can use Flash paper + FLEX like Adobe connect and cozimo.com. But if just received a PDF as an email and I want to send back some comments/ suggest corrections what would I use?
Is there a web app for that?
The problem is, these all have his original problem - clients who take years to decide anything.
B2B is difficult for startups, unless they already have good industry connections. On the other hand, consumers and small businesses are a little stingy.
Well, but it wasn't B2B. He had asked his own school for that relationship so they could market it for him locally - but the service itself is decentralized and in no way depended on that relationship. (It was probably a mistake to wait for them, but hindsight's always 20/20, particularly from the sidelines.)
Completely missing from your postmortem (from which I learned a lot and for which I am very grateful) is the fact that the service you offer, to judge from the demo alone, is frankly not very good. Perhaps your analytics show that few visitors bothered to look at the demo, so maybe the point is moot, but I think the demo is awful, and if I wanted to take over this business, I would completely redo it, or remove it altogether (sometimes it’s better not to show the product!).
I’m not talking about the implementation, which is great. I’m talking about the actual advice given in the demo, which you are offering as an example of the kind of thing students should be willing to pay for. The first comment is, essentially, use “and” instead of “&.” The second is, in effect, “make this bit a separate paragraph, and tie it in better.” And so on. The advice is generic or, when concrete (as in the case of “use and”), banal. There is very little of it. And there is very little of value in it. And since it’s a bit of a struggle to read the handwritten text of the sample essay, it’s even harder to tell if the generic advice is relevant--except in the case of the advice given for the conclusion, which is so general as to be universally applicable, which is not to say it is good advice.
I got a sense from your postmortem that you in part want to blame your lack of success on students not being interested enough in their own studies to pay for your service. I’ve taught thousands of students, so I know well how few are willing to write out practice exams, and how few of those are willing to seek out feedback. But a very small percentage of a very large number can surely translate to a modestly profitable business (especially, in this case, if you had plans to expand into the huge American market and into the college essay review business, which is what my crummy site is trying to do). God knows it’s hard to get students engaged, but the ones who visited your site were looking for something, and I don’t think they found something worth paying for.
Interesting. Looking at the demo now you're right: the first bit of feedback on screen is trite, and the rest is a bit wishy-washy.
I had a look at the analytics, and most people went for more info rather than the demo. However, those pages are really, really bad and sparse. Should have worked on those a lot - I remember I was planning to add tutor profiles.
Software myopia again. BTW if anyone wants to look at the analytics let me know and I'll give them access.
I would be interested to hear more about the web analytic.
Also, I am wondering if anyone has tried the "fliers on campus" or "ads on message board" advertisement campaign
for a e-learning or tutoring-related website. I would like to know how to know what kind of visitors one can expect from the idea mentioned earlier by jdietrich.
Another problem with the demo is that it's hard to find. This sounds dumb, because there's a big star-flash thing saying "Look at the demo here" - but I tend to ignore star-flash graphics as marketing fluff, and it appears to be the only link to the demo.
Then, when I got to the demo, it loaded slowly because the images on the page were large. It was difficult to understand immediately just what I was seeing. After a while, I managed to get a mindset that could interpret what I was seeing and I ended up loving the technical flair - but as a customer, not a Hacker News post-mortem fan, I think I would long since have been lost.
Probably, instead of a demo, a cartoon of some sort would have presented the idea better, like (1) snap a picture of your essay with your camera, (2) upload it with our easy interface, (3) see comments by our staff of starving PhDs, standing by. Something fun and non-threatening. That could sell.
There's a great deal of techie myopia in this site, and I speak from a similarly myopic standpoint. All credit to you, Tim: the actual site is really quite nice from a technical standpoint, and it really can't be seen as wasted effort, because this is definitely going to be a point in your favor with any venture you do later.
Another troubling fact is that the advice given in the demo also suffers from numerous grammatical errors. This is unlikely to give a potential customer much confidence in the quality of the advice.
I'd argue the main reason it failed was it was a bad idea. Their was no market for your product no matter how well you made it.
Try not to take away execution lessons when the idea was poor. The execution was fine and if you'd have used a good idea you'd be doing reasonably well now.
As an aside, does anyone have links to other startup postmortems?
There's a more fundamental flaw, something that you wouldn't have uncovered by interviews....
1. For the money to get someone to review your paper, you can get someone to write it for you. http://www.bestwritingservice.com/ has essays for ten dollars a page. If you're willing to pay someone 20 bucks to review your essay, you're willing to pay 50 to write it.
2. Most universities have writing centers where people will review your papers for free. These places are oddly underused. The people who would go don't need them, and the people who don't know don't realize they do.
Re your point one: This service wasn't for help with what an American would call term papers. It was for help with exams, which you have to write yourself (unless you hire someone to impersonate you, which I'm sure is not unheard of in big classes but is certainly riskier).
You're right, I misread that. At that point, why not just go to the professor or TA during office hours? Maybe other Universities work differently, but each class had office hours where I went to school, and most students didn't take advantage of that. And who writes out example exam questions anyway?
So yes, I stand corrected, Customer development would have uncovered those problems.
>And who writes out example exam questions anyway?
As a maths/physics student (though I did various other courses too) I pored over past papers, in part as an indicator of the papers to come (though they always caught one out) and in part for practice at recapitulating proofs and writing answers under time pressure. I certainly didn't spend the course working toward the exam but it would be silly not to attempt past papers on a course that has run unchanged for previous years.
Most of our papers had example answers with them though.
Personally I had my best exam results when I did exactly that, I wrote as much as I could for each historical exam question. I would research from book, lecture notes etc.
However, I wouldn't pay for this service when my lecturers and teaching assistants would do this for free and as they were the ones setting the exams I would prefer their advice than that of an unknown person.
There are a lot of times that it doesn't make sense to write any code before selling the product, but there are also times it does. The whole decision comes down to "Can I convince somebody this will work?"
The less likely your product is to be implementable, the more likely you will need code up front: 140 character mini blogs? Sure, fake it with photoshop. Real, Turing complete AI? You better have a prototype because nobody will believe you can do it.
Most startups tend to lean towards the former, but don't get sucked into that mentality if you lean towards the later. You'll be wasting your time.
I think it's not all or nothing. Write just enough code to be able to make a credible pitch. In some cases this might be little more than HTML/Photoshop. Or you might need a bit of a backend but it can be mostly mocked up or support only a few contrived scenarios that you'll be using in your pitch.
It's sad no one ever paid for the service, but if not don't talk to customers or potential customers (people who pay, not "users" i.e. in the sense of people that do not pay), you'll never know what they want. Generally, it's only when people part with cash that you know you are providing value.
However, major kudos for releasing the code and design docs so others can try some tangent, along with the post-mortem so others can avoid similar pitfalls. That's seriously cool and not something I often see people do for projects they invested so much in. Thanks.
FWIW: I had a student cheat by using a paid answer service during a final exam (in class but done online, easy to abuse a website when I'mnott looking). I figure it cost him over $300 to buy the answers (obviously not his writing style and proficiency); still failed the class.
Make sure assisting on those "practice" exams aren't real ones.
Nice writeup! I agree with you - I love reading post-mortems. Success stories are worthless (too much survivor bias) but post-mortems really tell me something.
I would argue that the majority of your target demographic doesn't have the inclination to put disposable income into this.
Most of the time, when I needed help, I'd find someone who could help me for free.
Maybe it's different in the UK and other non-US places, but here, IME, there is no large-scale culture of hiring tutors.
People typically hire tutors when they are out of their depth.
Just my thoughts and experiences. I wish you the best next time!
Here's an idea: in the USA, university student-athletes are typically supported by an army of tutors, perhaps the athletic dept. would be interested in something like this to streamline review of student writing.
Though as you note, I'm not seeing a lot here that wouldn't work just as well over email, perhaps backed with a simple issue-tracking system of some sort.
Edit: "You" in the prior paragraph is addressing Mr. Ruffles.
I used to work in the higher ed vertical. They have about the longest sales cycles imaginable because so much of every decision is committee led (speaking at the campus level; department sales are a different story). You also can never expect them to just make a decision. If you are selling to the entire campus, you need salespeople who understand the process to work you through it.
My advice to startups in that vertical: make it a department SaaS sale. You avoid central IT and it's committee process hell. If you really want the whole campus, start with a few departments, make them happy, then work your way up.
Personally I think the main problem was lack of commitment in customer discovery and how much were you paying the PHD students? Why weren't they motivated enough to help you advertise? In the US, most graduate students assist professors in teaching classes, so they would have direct contact with the people who would want your services the most.
What was your motivation for your startup? Did you just want to make a buck, or did you really care about helping students do better in their exams? Do you think the decisions you made would have been different if you had different motivations?
Interesting thought about phd students being a potentially untapped resource for startups. Anyone want to launch phdturk? The domain name is available (for now anyway)
I think he just misjudged how most students study. In law school almost all of my exams were essays. The fact is that most people did not study by writing practice essays.
To answer an essay question well you need to be able to do two main things: know the information the question asks for, and present that information in writing. For me and everyone I studied with the first part was crucial and the second not that important. Everyone trusts their own writing skills to write a proper answer once they have all the info.
So when people study, they concentrate on memorizing the information, not on writing the answer. People just assume that come exam time you will be able to write the answer.
Of course it would be very nice to be able to write a bunch of practice answers, but usually there is not enough time to study, you always feel like you do not remember all the info that you are supposed to remember, so you spend most time trying to remember more things.
It's a good point. What's interesting is that the academics I surveyed think students do need help with writing and expression. However, if students don't think so, as you point out, no perceived need to meet and therefore no sales...
imagine if it had taken you 2 weeks to write the code instead of 6 months and you spent the rest of the 2 years on customer discovery instead, imagine if you were able to make significant feature changes to your product in a few days based on what you found out about what your customers wanted .. do you think you would have been more successful?
The fact that you used lines of code as a metric for how much work you put in, and the fact that you think 30k is a lot of code .. and you used php/flex .. I don't want to rain on your parade but there are MUCH MUCH better coders out there, who can do stuff much much faster. I don't necessarily recommend the following languages overall but you might want to check out ruby or python and see what is possible in just a few days.
Dang great story. I've been working on a few projects here and there. The first one was too big a project and I called it quits after 2 months because 1) I did it for learning purposes and 2) it was too big to complete and launch in regards to other competitors. But 3 years is a long time to spend on a project. There are definitely lessons to be learned here and as long as you don't lose hope, you will eventually find something that works. Your mention of testing if people would actually buy your service/product w/o writing the code behind it is a great idea, much like what I read in the 4 Hour Work Week. Definitely something entrepreneurs should apply if possible in order to reduce the amount of time potentially spent on doomed projects or projects that need to pivot.
First: a time / value mismatch; students don't want to spend more time doing something, then waiting before they get feedback, and your PhD students are taking longer to crit, annotate, explain their reasoning etc. through an interface than they would take vocally. In fact, a paid telephonic [ VOIPic? - Ed. ] mentorship service would probably do better.
Second: you didn't understand the market, not just your customers. The market is _everyone_ you have to deal with, not just those that buy the bread.
Third: I'd place PG's #5 (Obstinacy) covers both your 'knowing how to code' problem, and the fixated / obliviousness.
Knowing how to do something is just knowledge. Knowing why, when, and when not to is wisdom. In the end, it sounds like you're glad that you could afford the lessons.
Maybe if you made the app free to use and have introduced a karma system , such as the one on StackOverflow, it would maybe make your (ad)venture more successful.
Personally the Flex UI puts me off -- the first impression after clicking the demo button. (The site frontpage looks ok)
I agree that a simple text annotation version would be much better and appealing. (its easier to upload a .doc or .pdf of your work, than scanning it, people who prefer to handwrite are less likely to use any online services than those who prefer doing stuff on their computer when possible
I wouldn't say it was a total failure. You learned something, and you can always reuse code (such as the 700 lines for credit card payments) for other projects.
>Please take the code (30k lines, PHP/Flex), design & docs and make a go of the business.
Could the OP please give a more formal license statement, even if it's just on this page?
Also, as others have said/hinted the execution looks (after a brief glance) to be pretty hot. The design is certainly crisp and clean and the screencaps look well laid out. Pivoting seems a good option.
Sure - in the repo I've specified the MIT license for the code, and CC license for the docs etc. That said, I'm very happy to give you whatever license you'd like - just let me know.
Another idea is to market to teachers. Scan in homework, grade/annotate on the computer, and send results to student and parents. Then, teachers wouldn't have to lug all that paper around, parents are involved in the feedback loop, 9 month sales window, doesn't require institutional commitments (any teacher could use it).
I would pay for that! The amount of paper that my kids bring home from school every day is staggering. If I could go online and see scans of all their work (along with all the school newsletters, PTO flyers, lunch menus, etc.) that would be wonderful.
Of course you'd need to figure out how to make it easy for the teacher to scan/upload of stacks of different sized, often folded/wrinkled paper.
Actually figuring out how to scan papers would be the hardest part. I think it would require an affiliate deal with a document scanner company. Or perhaps they're cheap enough that you could give away document scanners for a year's subscription (say $120 a year plus a free $60 doc scanner).
It's certainly something I wish the Open University (http://open.ac.uk) would use - I'm studying maths there at the mo and have to post my coursework in.
I'd guess parents would be keen on seeing how their children are doing too.
Thanks for providing so much detail into the inner-workings of your venture. Most people wouldn't have the guts to put themselves out there like that. If you think about wealth in terms of 'prevention of loss,' you've just made everyone on this forum quite a bit wealthier.
NP! It was a little scary, but it's been a very positive experience. I've got a lot from reading people's comments here, and it's crystallised a lot of the thinking in the article.
I hope you're right re prevention of loss, that'd be fantastic!
The only thing about this that irks me is not writing code until you have interested users. Wouldn't it be odd to say "oh, my product? It's coming, and it's awesome, but I haven't coded it yet." I'm all for the fake it till you make it but perhaps a bit too much here.
People answer polls on FB (which publishes your answer with your name) with whatever the person they want their friends to think they are would say. No-one wants to appear a cheapskate. That's why everyone said they would pay, regardless of what they really thought. An anonymous poll might have gotten more accurate results.
I'm not much of a fan of downvoting, but I suspect it was downvoted because trying to make this in to a "Rails rocks!" thread is pointless. That may not have been your intent, but it would have gone that way.
"I believe that using it you would have developed the app faster."
"Faster" than what? Using a fairly well-established PHP MVC framework with a large community and presumable decent PHP knowledge under his belt? Just because he knew Rails doesn't mean he doesn't know how to be productive in PHP. I'd also further add that Rails in 2008 is a far cry from where it is today. Hosting for PHP was a far better story than Rails in 2008, just to point to one facet.
There was nothing in the OP story that pointed to the tech choice having been a factor in the demise of the project. In fact, the whole point is that focusing on tech at the expense of finding paying customers is a surefire recipe for failure.
Bringing up the tech is not useful, which is probably the reason for the downvote.
You coded it in six months and the whole project start to when you gave it up was only a year? Most startups I know of take several years to get any revenue at all. You need to have brass balls and a rich friend / family / side consulting to keep eating and paying rent / expenses.
You received 3000 GBP in startup money competitions, no strings attached, the first year. That's pretty damn good. I know startups that have raised a million dollars and don't have that much revenue.
Don't forget advertising. Sure you need lots of hits to make good revenue, but I know guys living off only ads for their apps. They also offer paid versions, but these tend to be a minority of the revenue. Don't underestimate what adsense/admob plus some social media exposure can get you.
You probably don't realize how successful you are already. That leads me to believe that if you stick to it, you're going to be quite successful. It may take 20 years, but eventually you're going to get some traction.
It's not the idea. It's not really even execution. It's persistence. If you keep trying your startup long enough, you'll eventually make enough to live. Maybe not be rich, maybe not make as much as you could have in a big corp (TM), but you'll survive, and you'll be your own boss.
My family has been doing startups for literally generations. I'm not rich, and neither are they, but we've survived. We did it by being persistent.
Stick with it, and I know you'll succeed. Don't give up.
No, I think he was right. After that year he still has zero revenue - not a single person has actually signed on for his service. I think he has made a wise decision by realising that the market that he expected is simply not there.
I wouldn't have expected that it takes several years for most startups to see revenue. Profitability, sure, that's a harder thing, but for something like this you would surely expect to have a few customers after a year if it was going to work?
Persistence is a great thing, but surely there has to be a point where you realise that you're selling fridges to penguins and therefore it's time to try something else?
sticking with a project that has no momentum for a long time can be a huge waste of time. better to focus on finding momentum as quick as possible either in the current project or abandon and find new one. there's a lot of interesting things to do in the world
Agreed. Better to go out and find the zeitgeist than to wait for it to come to you. For a first failure without mentorship I would actually consider this a pretty fast one. :)
WHen I first visit your landing page, and stared at it for 5 minutes.. I still didn't understand what the site is about. I kept asking myself HUH? Huh? I still dunno what it's about.
In the middle of the article, the OP lists various mistakes he made, but I think he's basically wrong on all counts. His essential mistake IMO was launching too late. In six months of work, he gained no insight whatsoever into the market. He could have learned just as much with a handwritten flier on the college notice board - "Your mock exam reviewed by a postgrad, £10. Email foo@bar.com".
For the technically-inclined, coding is the perfect form of procrastination. It can absorb a near-infinite amount of time and feels quite productive, but it's usually a distraction. Steve Blank's most important message is that in an early stage startup, your job is to learn about the market. Anything which doesn't connect you with your customers is wasted effort.