All posts filed under “Geeking out

comments 22

37signals is one hell of a profitable business

[UPDATE]: Based on a number of comments below, via e-mail, and through additional research I’ve updated my original estimates.  I think 37signals had revenue of ~$4million in 2007 and ~$8million in 2008.  Please read my new post for details and to download the updated spreadsheet to play with the model on your own.

Inspired by a recent post on 37signals’ Signal vs. Noise blog, I decided to play around and see if I could try and figure out exactly how much money they’re making. Though my analysis is crude (as they obviously won’t release key figures), I think I’ve got a pretty decent handle on their turnover. Based on my calculations, they made around $3.5 million in 2007 alone, and look to make over $5 million in 2008.

So how did I come to this? Let’s dig a little deeper…


37signals clearly doesn’t talk about specific numbers much at all. They’re private and close-held, so there’s no reason to. That said, they have provided some numbers in three key posts that I’ve seen:

What’s making money and how much had some initial discussion and a datapoint for number of Basecamp projects.
Ask 37signals: Numbers had key data for all of the different products.
The New Backpack’s first 24 hours provided a few new numbers on Backpack.

To complete the analysis I took pricing data from the various pricing pages and other data from the overall 37signals website(s).

General Model

There were a few different ways that I could have modeled 37signals’ revenue. I decided to look at the data the same way that 37signals charges for their products: month-to-month. Each month, a certain number of people sign up for accounts. A large majority of these are free accounts, so the use of these accounts are limited. (And some people will just register and try them out, never to return.) A slice of these accounts will be (or will turn into) paying customers.

Each subsequent month, a new group of accounts are registered. However, this time a certain number of accounts are carried over (except for those that cancel the service), and these are added to the new paying accounts for that month.

What 37signals provided in their “Ask 37signals: Numbers” post was not the number of accounts, but instead they posted:

  • Basecamp: # of projects
  • Backpack: # of pages
  • Campfire: # of rooms
  • Highrise: # of contacts

They posted more information, but I decided that I would try to create a model that would match these datapoints.

Essentially I use this model with assumptions on how each pay and free account generate new projects/pages/etc., play with the numbers within general assumptions of growth and ratio of free/pay accounts, match the data provided, then calculate monthly revenue based off an assumed revenue per pay account times the number of current pay accounts. Simple!

General assumptions

I had to make some assumptions to use the model above:

  • Only 1-3% of new accounts are pay accounts
  • Each free account creates a certain # of projects, rooms, pages, etc. over the lifetime of the account
  • Each pay account creates a certain # of projects, rooms, pages, etc. per month while a pay account
  • Between 1-10% of pay customers cancel their service each month
  • The growth of people registering for new accounts has been growing steadily since each service launched

More on each application


Basecamp is clearly the main revenue generator for 37signals. It’s been going the longest, and so has built up a good number of pay accounts and people who really value the service. I estimate that at the beginning of 2008, 37signals is making nearly $300k per month from about 5k pay accounts.


Backpack is the second oldest application for 37signals, and with the recent changes to make it multi-user, I think this is really set for a lot of growth in 2008. Up until now, I estimate that 37signals is making about $35k per month from about 2300 pay accounts.


Campfire was released in early 2006. This is a hard product to estimate, but based on my (rather wild-ass) guesses, I believe they are making about $9k per month from 600 pay accounts.


Highrise is the new kid on the block, and competes with more sophisticated CRM software. (I’m sure Jason and Co. would describe their competition as needlessly complicated CRM software.) I estimate that over the past year Highrise has grown to contribute $30k per month to 37signals from 1100 pay accounts.

Other streams

37signals doesn’t just make money from their web applications. There are three other streams that should be taken into account:

  • Job Board (and Gig Board)
  • The Deck
  • Getting Real book sales

Since each Job Board ad runs for 30 days and costs $300, I just counted the number of current advertisements to guess their revenue. From this alone it looks like 37signals is making nearly $40k per month! The Gig Board adds to that, but it looks to be negligible. Please note the sampling error… to make a more accurate measurement I would need to count the number of ads every “x” many days and average them all out. My version is quick and should at least be in the right ballpark.

The Deck is an advertising network that runs on the 37signals blog. I calculated the total revenue for the Deck based on the cost of an ad and the number of advertisers, divided by the number of blogs on which Deck ads appear, and added in a factor that represents a somewhat larger share of the revenue for 37signals since it’s one of the most prominent blogs in the network. Based on all of this, I estimate that they earn about $6k per month from The Deck.

Finally, they have certainly made some money from selling copies of Getting Real. 30k copies of the PDF version sold for about $19 each, and the book version sells for $25. I estimate that they earn about $10k per month from these sales (PDF and hard copies).

Grand Total

All together, I believe 37signals currently generates revenue of approximately $400k per month from a diverse stream of applications and other revenue sources. If this continues, they are on track for yearly revenues of almost $5 million. This makes them incredibly profitable. They have 9 employees, plus some other help from time to time (it appears.) Jason Fried and DHH (and Jeff Bezos) must be doing quite well.

Think I’m full of s**t?

Well, I don’t blame you. To quote a great Radiohead song, I might be wrong. This analysis is a bit crude because of a lack of data. You may have very different assumptions that you would use in this analysis would could dramatically change the figures here. My growth model might be quite a way off.

Here’s the Excel file I used to calculate everything; try it out for yourself.

Right-click this link to download the 37signalsRevenueModel.xls

Just plug the assumptions into the relevant boxes and see how the figures line up for you.

Summary and why?

Let me make it clear that I’m both a big fan of 37signals and a user of their applications. I did this partly for my own edification and partly to better understand the industry of SaaS. (I’m working on an app of my own, but since I’m not a professional developer it’s taking quite me a while.)

Clearly 37signals has be incredibly successful, and I’m very happy for them. They clearly give back, through Ruby on Rails, thoughtful blogging, etc. And if they’re making $5 million a year, it’s no wonder why Jeff Bezos wanted to invest!

That said, 37signals clearly laid out in Getting Real that “Outside money is plan B.” With revenues of $3-5million and a small team, that fits completely with Furqan’s thoughts on early-stage venture capital funding. They could probably scale their apps a lot larger, but by maintaining a clear and consistent focus, they’re able to do what they do best and keep their quality and craftsmanship at the highest levels.

comment 1

Weekly round-up

Here is a weekend round-up of stuff I’ve been meaning to blog about for a while. Hope you find it interesting.

  • Ever wanted to buy or at least ride a Segway? If you don’t want to ride one across the United States like these guys, maybe you should consider building one yourself. Some UMich engineering students built one for $1750 in two months. Check out the video here (complete with cheesy music):
  • Another cool YouTube video here. Someone takes apart a common “Intelligent Design” argument by showing that in fact, it’s quite reasonable to expect a theoretical box of clock parts to assemble into a clock. It’s fairly straightforward math-wise, using genetic algorithms. I learned and used the same technique back as an undergraduate to calculate optimal satellite orbits in order to conduct “fly-bys” of particular planets within particular windows. Good stuff.
  • Great quote here.

    “I suddenly understood with great clarity that nothing in life—except death itself—was ever going to kill me. No meeting could ever go that badly. No client would ever be that angry. No business error would ever bring me as close to the brink as I had already been.”

  • If you have to use Outlook, first of all, I’m sorry for you. But to help ease your pain, try out Xobni. (Inbox spelled backward.) It will revolutionize the way you work in Outlook.
  • These are great rules to live by: (from Scott Berkun)
comment 1

Random notes and weekly round-up

Here are a few notes and links from this week…

  • I think I’m an 80%’er. From Simple Bits:

    […] I mentioned a fantastic book: Let My People Go Surfing by Yvon Chouinard. Yvon founded the expensive-but-awesome clothing company, Patagonia.
    […] The book covers the history of the company, Yvon’s philosophy on design, and being a reluctant business owner. It’s a great read, with a lot of insightful head-nodding.

    One part stood out in particular, when Chouinard talks about how he sees himself as an “80 percenter”:

    I’ve always thought of myself as an 80 percenter. I like to throw myself passionately into a sport or activity until I reach 80 percent proficiency level. To go beyond that requires an obsession and degree of specialization that doesn’t appeal to me.

  • Jimmy Kimmel and Sarah Silverman are dating, but perhaps not much longer! 🙂 (From the Elapsed Time blog.)

    And Matt Damon’s done this kind of thing before… here’s a clip from the film Eurotrip:

  • This is not something I would want to run into, even if it has been extinct for ~390 million years.
comment 0

Breaking news via Twitter

Today has been a big news day. The international financial markets went crazy, and the US Federal Reserve cut a key benchmark rate by 0.75% about an hour before US markets were going to open after a long holiday weekend. Last night was a major Democratic party debate, where the candidates started really doing battle with each other. Heath Ledger was found dead in his NYC flat. There were other big stories, but it’s late and these were the big surprises of the day.

What’s odd about all this? I found out about both of the “Breaking” news stories from Twitter.

I know that the tech-saavy people have already experienced this phenomenon, but as I’ve described earlier, I’m much more of a Twitter voyeur than a true Twitter-community participant, so it’s never happened to me. To be on the receiving end of breaking news twice via Twitter on the same day was fascinating.

While there’s really no implication of this in anything, it’s still interesting to note. Perhaps I’ll pay more attention to Twitter in the coming weeks/months.