In this post, guest author Ryan Carson goes through some of the lessons learned from building a Web app in four days. Carson is the co-founder of Carsonified, a web shop in Bath, UK. They’ve built four web apps, created ThinkVitamin.com and run events like Future of Web Apps. If you’re bored you can follow Ryan on Twitter.
The time it takes to design, build and deploy web applications has been steadily shrinking, especially with frameworks like Django, Rails and Symfony. With that in mind, we decided to push ourselves and attempt to launch a web app in 32 hours. Four crazy days later, Matt was born.
The app we built is a simple tool that allows you to post to multiple Twitter accounts. We learned a ton during the experience so I’d like to share some of those lessons with you.
How we did it
We have a team of nine people which were divided as follows:
- Two developers
- One designer / front-end developer
- Two bloggers
- One copywriter
- Three PR folks
I would say you only need three people if you want to strip it back to the bare minimum, which would look like this:
- One developer
- One designer / front-end developer
- One blogger / PR person
Our app was built in Python using Django and is hosted at WebFaction. It uses the Twitter API, Git and Codebase for version control.
How much did it cost?
On a basic level it cost us a week of salaries (around $10,000). There are some other small costs which I’m not including like rent, electricity, coffee and taxes. We got hosting for free because of a connection we have with the company but if you paid for that you might expect to pay not more than $400 for the first month (for a simple app).
Team building
Building a web app quickly is not only a great idea if you need to get your idea to market fast but it’s also a great way to build team morale.
You don’t need to build a brand new app in order to benefit from this idea. You can actually take time off to work on a new feature or direction for your current app.
There are some serious benefits to stepping away from your normal work and producing something totally new and creative:
- The best boost you can give you or your team is to provide the time to be creative. Turning off your phones and email and just focusing on something new and exciting will do wonders for your energy level.
- It could generate some amazing buzz around you and your company or products.
- You’ll come back to your current projects with a new perspective and renewed energy.
- It will push your team to learn new skills. For example, Will, our head of sponsor relationships, spent the whole week doing PR – something new for him.
Tips on working wisely
Here are a few tips that you should keep in mind if you’re focusing on building apps quickly:
- Limit meetings to one 10 minute chat in the morning and one 10 minute wrap-up at the end of each day. Meetings are the best way to kill productivity and crush creativity so keep ‘em short.
- Get people away from their machines at lunch. Go for lunch together and maybe throw the frisbee or play Wii. The excitement and creativity will quickly deteriorate if you don’t have a break during the day.
- Simplify the site and app as much as possible. Try launching with just ‘Home’, ‘Help’ and ‘About’.
- Make sure to build on a great framework like Rails, Symfony, Django or Objective-J. Part of our experiment was playing with Django and comparing it to Rails and Symfony (a PHP framework). We’ve found that Django lacks the rigor of Rails or Symfony, thus might not be an ideal choice for future projects.
- Go with the first logo idea and color scheme from your designer. You shouldn’t over-analyze the look and feel of everything as this process can go on indefinitely. Design the logo and move on. This is why you need to hire good designers and trust them to be good at what they do.
- Be technologically agnostic. If your developers are saying it should be built in a certain language and framework and they have solid reasons, trust them and move on. Again, this is about hiring smart people and getting out of their way.
- Coordinate how your designers and developers are going to work together. Our designer creates static HTML and then passes it to the developers who use the HTML as a basis for creating templates. These templates are then committed to a Git repository and from then on, the whole team works from that one repository.
- It’s not enough to just have a designer and a developer. You need a dedicated person who’s focus is solely spreading the word about your application and working to get media coverage. There’s no way we could get the kind of coverage for Matt that we hope to achieve without several of us working full time on it. However, do not hire a PR agency for this – there needs to be an authentic passion for the app that can only come from your team. (For instance, I asked TechCrunch to cover it, and Erick came back with the suggestion to write this post).
- Get your ‘Creation Environment’ setup correctly.
Building your Creation Environment
If you want to build quickly and creatively, you need to set up an environment that encourages and facilitates that process. If you don’t have the following basics down, your team will be constantly battling annoying issues instead of getting on with building. You’ll need:
- Good version control. I suggest Git.
- An easy-to-use source and changeset browser. We use Codebase.
- Solid server infrastructure. Why not build on Flexiscale, Grid-Service, Mosso or EC2 and let the big boys worry about uptime and server load?
- A ‘one-click’ deployment system. This means that deploying the code from your repository should take just one click. If it’s any more complex than that, there is potential for complications and downtime. Capistrano is brilliant if you’re using Rails.
- Printers, chalk boards and meeting space. People need the physical space to throw around ideas. We’ve painted an entire wall with blackboard paint so the team has room to sketch ideas.
- Coffee, water, music and healthy snacks.
If you really get these right, it makes building and creating so much more enjoyable and fast.
So that’s it …
Thanks for listening to the Matt story. Please share your advice and experience by commenting below. If you want to see a whole day of development squeezed down into four minutes, watch the video below. Enjoy.
Matt Week – Day Two Time Lapse – Music by MGMT
Related reading, ideas and tools
- Contest to build an app in 48 hours: Rails Rumble
- Post about the danger of too much functionality: Features are a one-way street
- Powerful web frameworks: Django, Rails, Symfony, Objective-J
- Version Control: Git and Subversion
- Source and changeset browsers: Codebase and Beanstalk
- Cloud hosting: EC2, Flexiscale, Google App Engine
- Easy deployment: Capistrano
$10k for a multiuser tweeter? Wow, I am certainly undercharging my clients.
nice writeup. i’d be psyched to work at a place that did things like this for team-building and morale. i participated in a startup weekend last year and had a blast – it wasn’t the most successful business venture ever, but i learned some things and made some new friends and it was totally worth it just for that. one of the biggest problems we had that you didn’t really cover was just deciding what to build and how it should work – there were like 60 chefs in our kitchen and they were all trying to make a different soup in the same pot.
I’m sure that most of us, from countries outside of US or UK, can easily make a much more complex website with that kind of money. For some, $10K can mean the base for a startup. Anyway, what I’m saying is that those bucks are not accessible for many developers.
Kudos for Carson, because I know he is a very good entrepreneur and for sharing with us this idea.
It’s definitely inspiring to see what can be accomplished today for so little investment in time and money. I would have liked to see the author mention CakePHP in the list of frameworks he recommends, as v1.2 is easily better than symfony and mirrors Rails much more closely in PHP.
Give nextdb.net a look, and checkout some of our demos. If AJAX apps are your thing, nextdb’s javascript API makes it extremely simple to connect to your hosted database.
But how do you build a useful web app in four days for $10,000?
I was just wondering about that. The description he provided sounded like a “Web Site” not a very funcational “web app”. I am sure you can create a small web app that does a few small things in four days.
but to write a web front end for an ERP system that contains 100 modules would take maybe 5 days (lol)..
“$10k for a multiuser tweeter? Wow, I am certainly undercharging my clients.”
I’m going to trump them. The other day I saw posterous. I rebuilt the functionality in a day with attachments, ect…
I am going to attempmt clone and superset Matt as a proof of concept today by myself, for free. I will GPL the script and make it available on this thread in a few hours.
I have never used Twitter API in my life and I am not going to use any frameworks except PHP itself. I have no design skills what so ever. If you need proof check out my social networking website.
I would like to hear from Matt team why they never considered Google’a App engine as a infrastucture. You still get to work within Django and Python without worrying about App Server/DB and their scalability.
I understand the bit (which could be huge for some) about relinquishing control of server platform to Google, still it deserves a mention.
@Don Wilson: I thought the same, but they did have 9 people working on the project.
My rough calculations mean the salaries for each of the guys involved, working 5 days a week (no holidays) would be around $40,000/year.
But in my opinion, 9 people on that is way too many, unless its a hobby project with friends, but even then I’m sure a smaller team could do things more efficiently…
You have to remember that 5 people were just there to get the word out, and one was just writing text.
So only 3 people actually developing/designing.
One reason for not using google app engine is if you need a relational database. A second reason might be you are not a python programmer, and third reason might be that you want an AJAX architecture.
I wouldn’t consider AJAX as an “architecture”…
Does this mean I can hire this group to make me a web app in one web for $10k? If so, I’m interested.
I’d love to hear something about what the 2 bloggers and 3 pr people were doing. How much can you possibly blog about 1 project in 4 days?
I might have to start using Twitter to give this little app a try. If anything, it proves a point that you don’t need a lot of money to bring a product to market in a short period of time.
R
It would be great if you could also follow up with a post when you make the first 10K in revenue using your newly created web app.
thats really inspire me, thanks…sometimes here in asia..for 1k usd..u can do wonders
@15, that’s exactly what I thought. To clone this I am going to use CURL to post the tweets to twitter. I am creating a user class, twitter connections class for that user, then I am mirroring the class in MySQL.
The connections are iterated in HTML when the user posts a tweet with checkboxes.
When the user posts each checked connection is iterated and and the tweet is posted. The tweet is then stored locally in a history class and SQL table.
All the SQL tables are created and so are the classes. I am now working on a very easy twitter PHP class with CURL whos implementation I got from the API page on the site.
I won’t be done by lunch, but probably by dinner.
@Ashley Williams: Also, I believe Carsonified works a 4-day work week.
Kinda overkill. Nine people for a django twitter mashup? And not only that, the concept’s been done by at least 4 other sites AND its on a dying platform that fails constantly.
Its a $10k PR stunt, not a web app.
why host on http://www.webfaction.com/ and then mention below you should use a scaling host like grid service from media templet? Seems inconsistent.
um…this is a “product”? where is the ROI?
You said “Make sure to build on a great framework like Rails, Symfony, Django or Objective-J.”
You probably need to add Grails to that list. You not only get rapid application development, but also the robustness required for an enterprise application.
Exelent – to PR team. As this made it to techcrunch.
And poor to the development team… I would give a max of 2 hours to the developer for creating the same functionality. The design is really creative and is worth 2-3 days of work.
If you could be satisfied with a simplier design you could have the same for 300$ and 4 hours of time on getafreelancer.com
@ 19 “Its a $10k PR stunt, not a web app…”
True you are…and not even an original PR stunt…
google “$12,107.09″
People always seem to give Ryan crap about his numbers, but I appreciate the way he always includes everything. I assume that the $10,000 figure is USD. If he’s paying 9 salaries (in the UK, no less) it’s not hard to see where most of that money goes.
http://www.orji...LinesOfCode.txt
This is what Matt boils down to. This one function and 3-4 MySQL tables to store the user info and the extra accounts.
Since there are only a few pages, I am going to use a makeshift skinning system with html comment tags replacing the data elements instead of PHP Smarty. It shouldn’t take but a couple more hours. I just need to pretty it up. I only have GIMP right now. I ordered Adobe Creative Suite Premium for Dreamweaver but it hasn’t arrived yet.
orjii.com/mattInaFewLinesOfCode.txt
This is what Matt boils down to. This one function and 3-4 MySQL tables to store the user info and the extra accounts.
Since there are only a few pages, I am going to use a makeshift skinning system with html comment tags replacing the data elements instead of PHP Smarty. It shouldn’t take but a couple more hours. I just need to pretty it up. I only have GIMP right now. I ordered Adobe Creative Suite Premium for Dreamweaver but it hasn’t arrived yet.
This is what Matt boils down to.
http://www.site...LinesOfCode.txt
There are 3-4 SQL tables as well. The cartoon design is the longest part.
No doubt interesting post…
However, what would be more interesting is:
1) What was the architecture of the site. Was it a single server or a multi tiered infrastructure? What did it look like?
2) What did the bloggers do? Your numbers don’t seem to add up. Are you saying that the 2 two bloggers worked full time (est 64 hours) on this project. If not how much time did they really put in and what exactly did they do. Also were you able to quantify their results.
3) Same question for your 3 PR people. What exactly did the do and what was the direct impact of thier work.
4) Why don’t you release the code?
johnmwillis.com
It’s a PR stunt – Carsonified are good at that! Er, 4 days? How come one of their employees was posting about ‘Matt Alpha’ on June 11th??? Tch…
http://216.239....lient=firefox-a
You guys are (as usual, for TC readers) completely missing the point – this was a team-building, morale-building project designed to bring the team together and make something cool just for the hell of it. Who cares if you could do it in a day for $20 plus coffee.
Carson and his team did something for a week, taking the days off from their usual job, to push out something fun and pointless (to their core business, anyway), and no doubt learned something in the process.
Woah, I didn’t know they actually built apps anymore, I thought they just held conferences… Have they done any significant (revenue generating) web apps since Drop Send?
“Who cares if you could do it in a day for $20 plus coffee. ”
The whole point is that you can build a new startup cheaply.
The point of building it in a day for $20 and coffee is to say that there is limited value in something that is too easily emulated with 10 lines of code and some HTML plus photoshop.
This is awesome, great work guys! I’ve actually been working on a video series called “A Startup A Week” where we pull together small teams of exceptionally talented devs and designers, and see what they can put together in just one week. It’s definitely a fun excercise, and it’s amazing what you can actually get done when you cut out all the reviews, meetings, etc..
You can check out what we’re doing over at http://www.astartupaweek.com
#28 Dude this looks like it was a marketing exercise that turned into a team building exercise. I think the point is that if they can’t quantify the results of a project the exercise are useless (marketing wise and team wise).
I just tested the site and #26 is correct there ain’t a lot there put your hat on.
very cool.
I designed and built crowdstatus and justaddme in less than a day.
I guess team working can add time to a development :p
also .net is very quick to develop with, membership system out of the box saves a ton of time and it scales pretty well too.
It seems like the post emphasizes implementation and PR as two main drivers for a successful application, which I think is just plain inexperience. The post also seems to divert from the subject of building an application to building a startup company.
Any kid can build an application these days for free, and chances are good it will be as good in terms of quality, performance and robustness as the one this team built for 10K. So to be truly minimalist, one can achieve the same result with 1 person and no money. I don’t find a PR/blogger person as a must have separate entity. If the author finds it possible to merge three developer roles into one, I don’t see why the ability to write few coherent sentences could not be part of the same person skill set?
If the post is really about building a product, than one big piece that is missing here is marketing. I’m not even sure by reading the post what the application does or what audience it targets. There is also the question of success metrics. Are they after the users? Money? Fame?
On the other hand, if the post is about building a company, it does not adequately address what it takes to build one that will sustain past the winter. There are plenty of good books written on this and a post like this can never be a substitute for a good book on building a company.
Among many things I find lacking, the work environment described in this post may be more acceptable as a temporary workshop or a computer science summer camp for college kids. I really doubt smart professionals would get excited about a company if all it has to offer is a wall size white board, chairs, desks, coffee and water. Eh.
Nicely laid out ideas and a decent read. we all love nifty “How-to”s and Top-10 lists and this is a perfect example why it scored on popurls.com
must say though, $10k is steep for such a simple app. checkout what it can do and still buy four of us meals for a quearter – http://WWW.FANBOOM.COM
great video … way cool
and hmm, 5k in india? but not so cool i think
I heard that group tweeting is illegal in some states.
I decided to only implement the functionality in my clone and style it Google clean style for the sake of time.
The templates are all html files in /templates and you will be able to edit them in Dreamweaver. 100% of the functionality will be there in the clone though.
betahouse just had a developer meet up last weekend. While there, some really cool ideas came out of it.
The team I worked on continues to fine tune the application we built and as soon as its been scrubbed, we’ll post to the world to use.
A developer meetup is no different than going out for 4 hours to smack a ball into a tiny hole in the ground. We understand there won’t be 15B valuations at the end of the event. It is just a way for various people, with various skils, to get together to put something together in a day or in 4 like above.
How many of you get to do this at work? Probably none because lifecycle management doesn’t allow for it. This environment is just being creative. That’s all you need in order to understand why people do it.
Happy 4th…
@6, @19, @21, @34 Thanks for bringing some sanity to the discussion. I’m all for people taking a week and launching something for the heck of it, that looks like how Twitter started, after all, but why is this TechCrunch material, unless the implicit message is that most companies profiles on TechCrunch have products with a similar (lack of) depth? It is no easier to do a startup now than it ever has been, this is because the challenge of doing a startup has never been how quickly you could write the code. If you’re trying to quantify things, I suppose you could say that the more useful thing to measure would be how great an idea, how large a market, and how much of a barrier to entry can you and your team build in six months of undivided attention? That’s how most successful startups were built 10 years ago, 5 years ago, and are still built today. You might be able to build more in your six months than you could 5 years ago, but you’re going to still need to have 2 man-years of lead over your competition, not the 3 man-weeks that this project has, for this to stand a chance as a business.
Well put.
Almost as good as Hashrocket! They’re a Rails shop, too. I guess that must say something.
http://www.hash...et.com/products
“The point of building it in a day for $20 and coffee is to say that there is limited value in something that is too easily emulated with 10 lines of code and some HTML plus photoshop.”
Wow. I never comment on TC, but I simply have to here.
I’d ask Chris to spec out what it’d take to build the original version of Digg, YouTube, Ebay, Craigslist, Reddit, Twitter, etc.
Building a web business oftentimes has very little to do with how complex the code is, and the defensibility of businesses has nothing to do with how easy they are to replicate… Witness all of the clone apps out there (founded by people who probably think the same way you do).
Hi!
Thanks for sharing your experience. It gave me many things to think about. I really got the feeling that you wanted to _give_ something to the reader, I appreciate that.
The most telling statistic about this project is the 1:1 ratio between PR people and coders.
He would have gotten more traffic (and made more in revenue) by burning $10,000 and posting the video on YouTube.
The time lapse video is sweet.
We recently switched to Git here at TechCrunch and we are already seeing the benefits.
I like Matt’s design. Good job!
I developed this web site: http://simplese...hapartments.com for about $6K. One developer. The customer supplied the header and footer graphics. What you can’t see from the web site is the extensive administrative interface used to enter the property data and photos.
It’s now deployed for two cities and architecture to roll out new cities easily and quickly.
I charged them $100/hour but gave them a break on some of my time. You know how it is. If they’re successful then they have a whole list of enhancements they want. It did take me about a calendar month but that was because of needing to service other customers at the same time and to give them time to evaluate each step and come up with feature changes.
Need a web app for under $10K. I can help you.
Peace,
Rob:-]
we’ve done a similar concept which is http://blitzweekend.com
we had 11 teams challenged to make a working product in 48 hours. Lots of Rails developers, but also python, javascript, php and other frameworks. People come up on Friday late afternoon and must present their work Sunday.
Most of what Ryan said here is true. Focus is king. Also having trust and having talented members.
It was a great opportunity to push your limits, and also find potential co-founders, or even start a new company
We’re planning a 2nd version in upcoming months if people are interested.