Introduction
Some of the most difficult questions about a startup have to do with making predictions about the future. Estimates are just part of the game if you’re trying to manage the precious few resources you have at various points in the beginnings of your company. The problem is we’re all pretty terrible about extrapolations because if you’re an entrepreneur, by definition, all your estimates will be wildly optimistic.
Making mistakes about your skills and speed can lead you to hold off on a billing system until the last minute or estimate that you have 1 month to launch when you really have 4. And if you’re not careful, hubris might kick in and have you predicting a $1 million in revenue within 8 months of launch.
In this feature, we’d like to help you minimize those mistakes by dissecting 4 very different web application companies that are at varying stages of their life cycle. In addition to age and size, these companies vary in their business models, programming languages and views on approaching the web as a platform. Specifically, we’ll be sharing code line counts, business processes, conversion rates, support requests revenues per customer. It’s good stuff and we’re hoping it’ll help you identify some trends that’ll make your predictions a bit more accurate.
The Participants
Here is a quick overview of the companies featured in this article:
Wufoo is our web application. It’s an online HTML form builder that helps anybody collect information over the Internet. When you design a form with Wufoo, it automatically builds the database, backend and scripts needed to make collecting and understanding your data easy, fast and fun.
Blinksale is used to invoice clients for services or products sold. Blinksale is used by attorneys, accountants, designers, IT professionals, software developers, journalists, contractors, engineers, architects, videographers, and more. Basically, if you need to send invoices, Blinksale can work for you.
FeedBurner is the leading provider of media distribution and audience engagement services for blogs and RSS feeds. Their web-based tools help bloggers, podcasters and commercial publishers promote, deliver and profit from their content on the Web.
RegOnline is one of the largest online registration systems on the web having processed over 3,537,094 registrations for more than 49,133 events worldwide. Over the past eight years they’ve invested over 20,000 development hours into the system and is considered one of the most mature and robust event registration systems available today.
The Basics
Looking over the basics, everything is fairly standard. The longer a startup has been around, the more employees it has, and programming language alone will not make or break a company.
| Language | Time to Launch | Launch Date | Current Employees | |
| Wufoo | PHP | 6 months | July 2006 | 3 |
| Blinksale | Ruby on Rails | 3 months | July 2005 | 6 |
| FeedBurner | Java | 5 months | February 2004 | ~30 |
| RegOnline | .NET | 3 years | 1996 | 50 |
The largest discrepancy is with time to launch. All companies were started with 3 or less people working on the code, but RegOnline in particular took the tough route: bootstrapping. The original founder worked a full time job, and wrote the code in his off time. This approach is definitely more demanding and requires much more dedication than receiving funding or backing by your freelance firm, but it can be done. All participants agreed that language had little to do with time to launch.
Code Line Counts
Not surprisingly, the longer a company has been around the more code it has. Experience also shows that once you are past your base set of code, the percentage dedicated to server side code starts to increase. This is primarily due to added complexity of features and other enhancements (spam protection, security, error handling, etc.) that do not necessarily have an interface element attached to them.
As far as the original interface is concerned, it’s interesting to see how Wufoo, the youngest company, has 29% of it’s code invested in JavaScript with the second highest line count for that language among all 4 companies. Arguably, this trend will probably continue with other young “Web 2.0” apps as they adopt more Ajax intensive programming techniques.
As far as choice of language goes, all 4 companies are satisfied with their choices. There are a few minor complaints — FeedBurner wishes to some extent that the lighter, simple web interface tasks were in a different language (but they’re just fine with the heavy processing in Java), while RegOnline is aware of the fees that need to be paid in order to use a Microsoft based development platform — but overall, everyone is happy. In the end, it seems that choosing a language that you’re comfortable with can result in less lines of code and a quicker development time.
Line Counts per Business Process
The goal of this section was to uncover any task that might be overlooked. When writing a software product, the tendency is to focus 100% on the application. Items like support, marketing, and especially billing never cross your mind — in addition to the rest of the underbelly.
While these business processes seem to only contribute towards 10% of the code base for the two most established companies, it is important to recognize that the core for the business code has to be in place during launch, if not shortly after.
Support Requests
Before even looking at the numbers, it is worth noting that support is more demanding than most would anticipate. While multiple things can lead to support requests (poor code, bad interface, complexity, lack of documentation, etc.) what seems to matter the most to customers is how quickly, friendly, and accurately you handle it.
| Support Requests | |
| Wufoo | 12/day |
| Blinksale | 8/day |
| FeedBurner | 45/day |
| RegOnline | 234/day |
A few successful approaches to reducing the number of requests that have worked for the above companies are the introduction of the following:
- Forums
- Knowledge Base
- Documentation (with videos & screenshots)
- Useful, custom tools for support staff
The amount of time dedicated towards improving support can speak volumes. A majority of the panel agreed that support requests tend to lean towards “How do I …” and “It would be cool if …”, which is why when RegOnline established their knowledge base they saw a huge decline in tickets of that nature. Personally, I am in agreement that this type of support reigns supreme, but be aware that heavy use of JavaScript (with JSON) and foreign characters will also keep you up for quite a few nights.
Conversion Rates
Getting users is one of the more stressful aspects of a startup. Without users, you have nothing. And contrary to grand expectations pre-launch, those users are tougher to convert than anticipated. In fact, 1% is a fairly standard number to expect. So of 100 people visit your homepage, 1 person will become a paying customer.
| Free | Paid | |
| Conversion per 100 Visits | ||
| Wufoo | 7% | 1% |
| Blinksale | 11% | 1% |
| FeedBurner | 8% | N/A |
| RegOnline | 1.52% | 1.14% |
Free accounts can also generate revenue through ads, awareness, and link backs, but the amount of free accounts you have does not necessarily increase the conversion rate that leads to getting paid. Blinksale has the highest conversion to free ratio (even though their signup form has 20 fields!), but they still come in at 1% paid. On the other hand, RegOnline has a very limited free plan, which shows a significant drop in free signups along with a much higher free to paid conversion.
To place more emphasis on the difficulty to convert, we can look at the conversion of beta testers and free advertisements for Wufoo. After a few months of limited beta testing, everyone who provided active feedback was given a free account. Of 7000 other testers, less than 1% became paying customers even though they were given a 50% off lifetime discount. Similar to that, ads that we place on both Particletree and on free forms result in approximately a 1% conversion to paid.
Average Revenue per Customer
The three companies listed all rely on monthly and/or usage fees in order to make money. Obviously, RegOnline is in an enviable position here. A few things that may result in their revenue is the maturity of their product (and the number of features they have available) along with their 10 year reputation. Not to mention, their customers are all making money (event registration), so they are willing to spend money in order to make money.
| Avg Revenue/Customer | |
| Wufoo | $13.03/month |
| Blinksale | $12.25/month |
| RegOnline | $131/month |
Wufoo and Blinksale have a similar approach in that they are trying to get a swarm of users paying less money. The goal is to reach some of the smaller businesses that are more reluctant to jump into a big purchase, and also to increase word of mouth through the volume of customers. Finding the target price to charge for your product or service is no easy feat. When Blinksale first launched, they realized they underpriced their product, and had to make the appropriate adjustments later down the road. The grandfather adjustments worked out for them, but time and programing effort had to be provided to solve the problem.
Equally challenging is how to raise the average price per customer. There are a couple of ways to attempt this:
- Keep adding features, and create truly worthwhile features that are only available to higher paid plans.
- Improve the marketing/site direction of new users, so that they are more compelled to sign up for a higher paying plan.
- Make it obvious in the application that certain features are missing from their lower paying account, without being blunt and obnoxious about it.
- Make upgrading, downgrading, and canceling as painless as possible.
- Varying levels of support.
Speaking for Wufoo, we’ve been working on these ideas for a year now, and noticed varying levels of success. The worst payoff has been premium support, but we haven’t tried throwing phone support into that package. One noted benefit of phone support is the ability to resolve an issue immediately. Adding features is definitely a plus, but nothing drastic. Actually, slow and steady seems to be the way things are, so there may be no home run solution that increases this number by 20% in one go.
Seasonal Trends
Perhaps the most demoralizing thing that can happen to a first year startup is the change in sales from month to month. The only way to get through it is to go through a year and see what happens, but try to prepare yourself for the fluctuations, and to predict what months will and will not work for you.
| Best | Worst | |
| Wufoo | Feb | Dec |
| Blinksale | Jan | Dec |
| FeedBurner | Jan/Feb | Dec |
| RegOnline | Sept | Dec |
For most web services it will be a given that December is a horrible month comparatively. To put it into perspective, RegOnline has noticed an 18% decrease of registrations compared to the annual average, and Wufoo saw a 50% decrease. And even though the 50% figure is distorted by growth, December was still 25% lower than the prior October. A best guess prediction for the drop in sales is the trifecta of budgets running low towards the end of the year, the holiday season, and money being spent elsewhere (bonuses, travel, consumer goods).
Thank You Participants!
This article is based on the panel we presented at SXSW 2007. Many thanks goes out to hours of data collection put in by all the members: John Zeratsky (of FeedBurner), Josh Williams (of Blinksale) and William Flagg (of RegOnline). We asked a lot from them and they were generous enough to help us out and share their company’s information. You can download a copy of the graphs from the presentation’s slides here.
I really liked this post Ryan. Its well written and you give some good information. I was surprised by the conversion rates for Wufoo.
Hey! I know it’s being picky but the title is slightly misleading. At the beginning I thought you were going to analyse bankrupt companies, and then I saw feedburner in the list - couldn’t believe it was bankrupt.
See first meaning of http://dictionary.reference.com/browse/autopsy - that’s what the title looks like: “inspection and dissection of a body after death, as for determination of the cause of death; postmortem examination.”
I know it should be understood as (2) an analysis of something after it has been done or made, but you know what I mean… :)
It was an interesting article anyway. Companies do not disclose those data everyday!
Excellent article- some really interesting information going on behind the scenes of these popular apps. I really appreciate the insight, as an app developer its nice to see how much code goes into the apps, how they are built and managed. Thanks Ryan.
Nice writeup. Its good to see some of this “comparative” data with real production products. Makes it easier to judge the projects I’m working on both professionally and personally.
Great article Ryan. Perfect for anyone beginning a web-app / startup. It was heartening to see that a very small idea, like a form generator, can be successful (depending on your definition of successful ofcourse).
Thanks for the info :)
Wow ! Everything I really wanted to know… but never asked!
Thanks :)
I’m just having a chuckle that my (profitable!) Web app has less lines in its entirety than any of these four apps have in just CSS ;-) Not sure if that’s a bad thing or a good thing, but funny nonetheless!
Peter, I’d have to say that’s a good thing.
One thing I don’t understand is what’s the point of comparing an app written in 1996 to 3 others written after 2004? I mean come on, then we show the code line count and make .NET look like crap. After 1996 three versions of .NET have come out with numerous libraries, same project would’ve finished in about 6 months with 1/6 of the line count today… We’re basically comparing apples to oranges…Was Rails even around in 1996?
I wouldn’t look at it quite like that. Just looking at features and configuration alone, RegOnline has quite a lot going on. In fact, nothing substantial can be gathered from the code line counts in relation to the language chosen. As far as time to develop goes, I’d wager that a one man team working nights and weekends only would still take around 2 years to finish even with modern languages and libraries.
The better ideas to consider with that graph are the breakdowns between sever, client, designers, etc. Also, it gives people an idea just how much code they will have to manage, which leads to thoughts on source control, flexibility, and so on. And most importantly, having a company that is 10 years old lends a good amount of consistency to the statistics — especially the business related ones.
Great article, but the ratio I care about is how many people register vs how many end up paying, as the people who create accounts are the ones who are at least somewhat interested in the product. Who cares about how many people see your homepage - your page gets on digg, you’ll get 20,000 visitors and none of them will be worth anything.
I think the “Code Line Counts” section could use some clarification. Is this lines of code written by the company only or does this include line counts of any languages they are using? And, where is the cutoff? 26,000 lines of handwritten backend code for even a large Rails project seems absolutely massive…. so, does this number separate code written and code used? And where does that statistic end? (classes like Array and String have code in them as well!)
Ah, how could they be using .net in 1996? It was still in the MS Labs(probably still with the company MS bought it from) in 96, not even a public beta existed at that time. You should probably change the wording to language used now for that particular entry.
Biopsy, not autopsy.
None of these sites are dead.
I don’t get these comparisons.
I mean, does it really surprise you that it took 2.5 years longer to build “the most robust online event registration system on the internet” than it did to build a form generator? Does it really surprise you that they have 50 employess (85% of which probably do support)?
You need to pick applications that are in the same problem domain. Especially if you’re going to compare conversion rates, marketing expenditures, etc.
Nice pie charts though.
One of the best articles I’ve read in a long time! Great insight into Web development and very useful for startups. :) Cheers.
Having worked on several large web projects over several years using asp php and rails. The recent projects i worked on would still not be complete if we had used php java or asp. our team members have worked with all the above technologies and and in our collective experience we rails as being so far in front as far as time to launch, ease of bug fixing, and maintenance. Java app (how many XML config files did you write) I agree that language alone makes little difference, however, the framework is very important. since i now boycott all MS technology i cannot comment on .NET (its been years) but why pay M$ when Rails is free and makes web app development a joy. so .NET is a framework, java has a framework. Whats does PHP have? in my experince it has way too many different way to skin the same cat. There is no true centralized PHP community rallying around one framework, plus that language is flawed in many ways. No one would set out to write PHP today, its been a series of hacks from day one, when rasmus put up some perl scripts to count views of his resume.
Ruby is a most eloquent language and Rails is a framework thats so smart its turning heads worldwide and brought into the international spotlight. One the best thing about Rails is Ruby. Yes its true I am now a rails zealot. Don’t get me wrong I loved PHP for years, but ruby on rails has helped me move to a new level. My intention is not to start technology dribbling contest but assert my view that Rails is the new future of web development. We watch the rest of you on the sidelines as we whiz by to launch our next app in 3rd 4th 5th maybe even 8th of the time.
Great article but i really believe, framework of choice makes a difference to you time to launch. (Mufoo doesn’t look like a framework to me sousing it in this comparison is somewhat misleading)
Great article, and a fascinating insight into how these companies work. What I’d really love too see is how a startup would approach designing a large application, from a software engineering perspective.
I’m so glad you knew I needed a hug. As someone starting a web2.0 business it is really fascinating to know what makes all those other websites tick. i’m working with drupal with is php but I would much prefer ruby on rails. http://www.buythatlocally.com
This would be even more useful with ballparks for the number of total users (or even total visits, which would match the way you currently present the data) of each app. Revenue from a 1% paid conversion rate of 1000 users is much different than for 10,000 users. It also gives a good view of the market size for each app, which I think is pretty valuable.
But good work nonetheless.
Very nice article, Ryan. I’m involved in a few different “web 2.0” projects so it’s good to see time-to-launch numbers that low. (At least, I consider 3-6 months low for launching a new product from scratch.)
What I would be very interested in is marketing tactics for all companies. As a programmer, this is the most difficult piece of the puzzle for me.
Thanks again for the write up.
This is an interesting article. these kinds of stats are always interesting to read so you can build an overall picture of whats going on.
A new perspective. I like that. What I feel is missing is which technology scaled better and hardware infrastructure requirements. Also a project with python could have been included in the list.
Very cool article. Makes me feel a lot better about the ups and downs in the new site I am co-running/starting. We took 14 months to launch (like Reg, I am still working during the day, and the only coder atm), but have been getting very good results with users, just nothing in the way of real revenue yet, but its nice to be reassured that this will come as well.
Great article. But I use Blinksale — depend on it, actually — and the word “autopsy” gave me a scare, as it did other commenters. Don’t DO that to me, man!!
Also … your layout & color choices make the comments really difficult to read. A little less creativity is sometimes a good thing.
I’m surprised no one has yet commented on the fact that this was a post was a direct result of the SXSWi panel bearing the same name. I, as more of a developer than designer, enjoyed this panel - and panels like this, such as Barenaked App - at SXSW and would like to see more of them.
And commenter Steven Holloway, I just wanted to point out that you sound like a complete cock when you speak about Ruby being the far most superior language over all others. You have to be a complete moron to believe that. Every programming language has it’s pros and cons, you’d do well to realize this fact.
One interesting aspect here is the support request counts. I wish you had a little more breakdown there. At first glance I suspect that RegNow is higher not only because they are larger than the others, but also because they deal with a less technical clientele.
Yeah, the SXSW panel was outstanding, thanks for that, guys!
Nice article. Especially the conversion rates and time to launch a business are interesting. As others have said, the article could have a little more details on some topics. I guess the companies may not be willing to release that information.
Good article—but the autopsy thing threw me. I’ll tell you a story about Wufoo: I signed up for a paid account, didn’t use it for four months, and was all set to cancel when they sent me a personally signed Christmas card. That kept me going for four more months, again without using the service at all. I finally started to use the app, had a problem, and they got back to me super-quickly and fixed the problem. Now I’m set for another six months or so…I’m not sure if the moral is that great service counts for tons, or that I should keep better track of my finances, but you get the idea.
how do you know the lines for server side code? I’m currently developing an application which I hope will be as innovative as these..
I also noticed the lovely form styling of the comments is from WuFoo ;)
I couldn’t understand some parts of this article o.us poetry, but I guess I just need to check some more resources regarding this, because it sounds interesting.
Everyone needs a hug.
I am trying to figure out how this weird comment list is rendered.
The aim of propagandism is to influence people’s opinions or behaviors actively, rather than merely to communicate the facts about something. For example, propaganda might be used to gather either support or disapproval of a certain position, rather than to simply present the position, or to try to convince people to buy something, rather than to simply let them know there is some thing on the market.
What separates propagandism from “normal” communication is in ways by which the message attempts to shape opinion or behavior, which are often subtle and insidious among other characteristics. For example, propagandism is often presented in a way that attempts to deliberately evoke a strong emotion, especially by suggesting illogical (or non-intuitive) relationships between concepts or objects (for instance between a “good†car and an attractive woman or a sex symbol). The “super-villain” Captain Nazi on the cover of Master Comics #21 (1941), flanked by his American patriotic enemies Captain Marvel (left) and Bulletman. Art by Mac Rayboy. The “super-villain” Captain Nazi on the cover of Master Comics #21 (1941), flanked by his American patriotic enemies Captain Marvel (left) and Bulletman. Art by Mac Rayboy.
An appeal to one’s emotions is, perhaps, a more obvious, and more common propagandism method than those utilized by some other more subtle and insidious forms. For instance, propagandism may be transmitted indirectly or implicitly, through an ostensibly fair and balanced debate or argument. This can be done to great effect in conjunction with a broadly targeted, broadcast news format. In such a setting, techniques like, “red herring”, and other ploys (such as Ignoratio elenchi), are often used to divert the audience from a critical issue, while the intended message is suggested through indirect means.
This sophisticated type of diversion utilizes the appearance of lively debate within what is actually a carefully focused spectrum, to generate and justify deliberately conceived assumptions. This technique avoids the distinctively biased appearance of one sided rhetoric, and works by presenting a contrived premise for an argument as if it were a universally accepted and obvious truth, so that the audience naturally assumes it to be correct.
By maintaining the range of debate in such a way that it appears inclusive of differing points of view, so as to suggest fairness and balance, the suppositions suggested become accepted as fact. Here is such an example of a hypothetical situation in which the opposing viewpoints are supposedly represented: the hawk (see: hawkish) says, “we must stay the course”, and the dove says, “The war is a disaster and a failure”, to which the hawk responds, “In war things seldom go smoothly and we must not let setbacks affect our determination”, the dove retorts, “setbacks are setbacks, but failures are failures.”
In this example, the actual validity of the war is not discussed and is never in contention. One may naturally assume that the war was not fundamentally wrong, but just the result of miscalculation, and therefore, an error, instead of a crime. Thus, by maintaining the appearance of equitable discourse in such debates, and through continuous inculcation, such focused arguments succeed in compelling the audience to logically deduce that the presupposions of debate are unequivocal truisms of the given subject.
The method of propaganda is essential to the word’s meaning as well. A message does not have to be untrue to qualify as propaganda. The message in modern propaganda is often not blatantly untrue. But even if the message conveys only “true” information, it will generally contain partisan bias and fail to present a complete and balanced consideration of the issue. Another common characteristic of propaganda is volume (in the sense of a large amount). For example, a propagandist may seek to influence opinion by attempting to get a message heard in as many places as possible, and as often as possible. The intention of this approach is to a) reinforce an idea through repetition, and b) exclude or “drown out” any alternative ideas.
In English, the word “propaganda” now carries strong negative (as well as political, mainly) connotations, although it has not always done so. It was formerly common for political organizations, as it had started to be for the advertising and public relations industry, to refer to their own material as propaganda. Because of the negative connotations the word has gained, today it is uncommon to admit producing propaganda, although opposing parties sometimes accuse each other of producing propaganda. Other languages, however, do not necessarily regard the term as derogatory and hence usage may lead to misunderstanding in communications with non-native English speakers. U.S. propaganda poster, which warns against civilians sharing information on troop movements (National Archives) U.S. propaganda poster, which warns against civilians sharing information on troop movements (National Archives)
Public relations pioneer Edward L. Bernays in his classic studies eloquently describes propaganda as the purpose of communications. In Crystallizing Public Opinion, for example, he dismisses the semantic differentiations (“Education is valuable, commendable, enlightening, instructive. Propaganda is insidious, dishonest, underhanded, misleading.â€) and instead concentrates on purposes. He writes (p. 212), “Each of these nouns carries with it social and moral implications… The only difference between ‘propaganda’ and ‘education,’ really, is in the point of view. The advocacy of what we believe in is education. The advocacy of what we don’t believe in is propaganda.â€
The reason propaganda exists and is so widespread is because it serves various social purposes, necessary ones, often popular yet potentially corrupting. Many institutions such as media, private corporations and government itself are literally propaganda-addicts, co-dependent on each other and the fueling influence of the propaganda system that they help create and maintain. Propagandists have an advantage through knowing what they want to promote and to whom, and although they often resort to various two-way forms of communication this is done to make sure their one-sided purposes are achieved.
[edit] Types of propaganda A 1947 comic book published by the Catechetical Guild Educational Society warning of “the dangers of a Communist takeover”. A 1947 comic book published by the Catechetical Guild Educational Society warning of “the dangers of a Communist takeover”.
Propaganda shares techniques with advertising and public relations. Advertising and public relations can be thought of as propaganda that promotes a commercial product or shapes the perception of an organization, person or brand, though in post-World War II usage the word “propaganda” more typically refers to political or nationalist uses of these techniques or to the promotion of a set of ideas, since the term had gained a pejorative meaning, which commercial and government entities couldn’t accept. The refusal phenomenon was eventually to be seen in politics itself by the substitution of ‘political marketing’ and other designations for ‘political propaganda’.
Propaganda was often used to influence opinions and beliefs on religious issues, particularly during the split between the Catholic Church and the Protestants. Propaganda has become more common in political contexts, in particular to refer to certain efforts sponsored by governments, political groups, but also often covert interests. In the early 20th century the term propaganda was also used by the founders of the nascent public relations industry to describe their activities. This usage died out around the time of World War II, as the industry started to avoid the word, given the pejorative connotation it had acquired.
Literally translated from the Latin gerundive as “things which must be disseminated”, in some cultures the term is neutral or even positive, while in others the term has acquired a strong negative connotation. The connotations of the term “propaganda” can also vary over time. For example, in Portuguese and some Spanish language speaking countries, particularly in the Southern Cone, the word “propaganda” usually refers to the most common manipulative media — “advertising”.
In English, “propaganda” was originally a neutral term used to describe the dissemination of information in favor of any given cause. During the 20th century, however, the term acquired a thoroughly negative meaning in western countries, representing the intentional dissemination of often false, but certainly “compelling” claims to support or justify political actions or ideologies. This redefinition arose because both the Soviet Union and Germany’s government under Hitler admitted explicitly to using propaganda favoring, respectively, communism and fascism, in all forms of public expression. As these ideologies were antipathetic to liberal western societies, the negative feelings toward them came to be projected into the word “propaganda” itself.
Everyone needs a hug.
Everyone needs a bum.
Here is another comment
Thanks for this article - and thanks for pointing me towards Wufoo, it looks super userful and simple to use.