Using Amazon's S3 is a good idea while your company is still acquiring a lot of users, but at some point in future if you are a storage company and you are in it for a long haul, you have to build your own storage solution.

I think Dropbox is approaching to the point when they have to build their own storage. You can't be a storage company without having your own storage solution in backend. It would be similar to Yahoo not having their own search engine while being a search company.

Makes me think of why Backblaze built their own "storage pod". http://blog.backblaze.com/2009/09/01/petabytes-on-a-budget-h...

According to them, their storage costs are ~20 times less than those of S3. A bit different use-case for sure. But as wave says, once you become a storage company you will inevitably need to do it in-house to be cost-effective.

First comment, forgive me.

"You can't be a storage company without having your own storage solution in backend."

I am not sure that is necessarily true. There are plenty of web hosting resellers who are successful without ever buying a single server or a single line. They've repackaged and re-sold a service. Dropbox can operate in a similar fashion, re-packaging and reselling Amazon's services.

> Dropbox can operate in a similar fashion, re-packaging and reselling Amazon's services.

Yes, they can, and apparently have, but they're getting to the size now that they can achieve some significant economies of scale by building out their own storage infrastructure. Their profit margin has to be pretty low with S3.

Like Engine Yard and Heroku?

When SmugMug got large, didn't they move from in-house storage to S3? Granted, one of the reasons was cash flow, which isn't as much of a concern for a VC-backed startup.

I'd wager that "large" SmugMug is vastly smaller than Dropbox (probably comparable to Dropbox of a year or two ago). I'm certain Dropbox is storing orders of magnitude more data than SmugMug at this point in time.

I thought yahoo's search results are now coming from bing...

I am referring to in past when Yahoo used Google and other search engine solution instead of buying or building their own before it was too late.

Exactly.

I think they're actually quite different circumstances (though your point is valid, that a storage company should own their own storage solution). Yahoo never thought of themselves as a search company. I think they considered search to be a commodity... as long as it works thats all that matters, and thought the value of the advertising should be their focus.

Very interesting calculations. I think patio11 also wrote about Dropbox's numbers a little while ago, but I don't remember where.

One big omission from the article - Dropbox probably gets discounts from Amazon. Obviously we can't know how much, but I assume they're paying even less on storage than is calculated there.

I made a fairly offhanded comment about them here: http://news.ycombinator.com/item?id=2353418

I generally wouldn't suggest using back of the envelope calculations for much of anything, by the way. Dropbox has stupidly attractive unit economics and a proven customer acquisition channel which explodes like whoa. There. That is probably accurate and possibly useful. Dropbox has N% margins, as demonstrated by back-of-envelope calculations? That is perhaps not quite so accurate or useful.

Thanks, that's what I was talking about.

Back of the envelope calculations aren't necessarily useful, but they're fun. They let you analyze problems and try and think of all the different moving parts that go into building something - that's why they're so popular in hacker circles. The conclusions are probably useless [1], but the debate and analysis is what you take away.

[1] I can't remember what it's called, but I remember reading a theory (on Wikipedia) that says that people tend to estimate pretty well when they estimate the multiplication of lots of numbers. Mostly because, on average, they make mistakes in both directions (i.e. half the numbers are too high, half are too low).

I think back-of-the-envelope will at least get you within an order of magnitude of the real numbers, which gives you a reasonable range to make assumptions.

> I assumed that the average user will make 10 put/copy/list requests and 10 get requests daily

Did those assumptions seem ridiculously low to anyone else too? My Dropbox is constantly syncing. Every time I save a file, move a file, delete a file, on any of my connected devices, I am making a request. I would estimate that I make nearly 500-1000 requests a day.

That also suggests that the total bytes transferred is a lot more than the author estimates.

I am pretty certain that Dropbox pays more for the requests/transfers than it does for space nominally.

Another thing: Did the author neglect the "30 day versioning" (or forever-backup) feature that Dropbox has? I think each user probably consumes a lot more than the suggested few hundred megs (average), in part because of all the changes.

Since the changes/uploads of the files are just diff's of the existing files rather than full copies for each version, I can't imagine that the versioning would add too much to the author's rough estimations.

But that doesn't change the number of requests. Perhaps it lowers the total bandwidth consumed, but what about the number of requests?

I think the conversion rates used in this analysis are an ineffective way to examine services like Dropbox.

Given that most users probably convert once they have filled up their Dropbox (which takes time), a cohort analysis done by segmenting users based on how long they have used the service would probably paint a much more accurate (and attractive) picture of Dropbox's business model.

For example, while the global conversion rate might 2%, the average conversion rate for customers who have used the service for 1 month might be .1%, 6 months: .5%, 12 months: 2%, 18 months: 5% etc etc.

If you are only using the global conversion rate to model the business accelerating growth will drag down the average conversion rate, incorrectly making the unit economics appear less attractive. On the other hand, if your model uses cohort analysis you will see accelerating growth accurately translated into future cash flows.

I forget where I read it but I am pretty sure a cohort analysis of this sort was used by Evernote when they raised their round from Sequoia.

Edit: Clarity

$100k per employee in the valley? That seems low.

Fully loaded (healthcare, office space, payroll taxes, perks, IT, etc) are probably closer to $200k per.

Probably under 150.

Not sure why the downvotes but average all-in costs of startup employees in the Bay Area startups is less than $150k.

He made a minor mistake. In his high estimate, he used the high end of the range for de-duplication savings (20%) when he should have used the low end (5%). Not a big error (something like $4.5m/mo instead of $4m).

I don't think that's a mistake on his part. If you put a single large video or iso file on there that you downloaded, it's practically guaranteed that someone else has already uploaded it.

And one 700 MB file will really skew that percentage of deduplication savings.

The point is that 5% de-duping (ie less deduping) would lead to the high number.

Dropbox is one service for which I definitely have no problem paying $10 a month.

Why does he add S3 standard-storage-class cost and S3 reduced-redundancy-storage-class cost in the second paragraph? Each object in S3 is either one or the other -- not both.

Since Drew's hitting up $35k a plate fundraisers, I can only assume they figured out the economics at scale.

http://techcrunch.com/2011/04/20/after-a-full-afternoon-at-f...

I think your analysis is flawed. While I think your costs are probably on target (except for compensation costs, which I think you are short by about 50-100%) your revenue guesses I think are grossly off the mark.

Dropbox has 25,000,000 users. If 5% are converting at $100/year then they are making $125,000,000 a year in revenue.

You're numbers are nowhere near that mark. And I think Dropbox probably has a much higher conversion rate than 5%. I wouldn't be surprised if they weren't on $200,000,000/year run rate.

I love Dropbox as much as anyone, but unless you're using it for audio, video or (lots of) photography, the free offering is plenty. I'd be surprised if they were much north of 5%.

That said, they've got a great "long game". Everyone will fill it up EVENTUALLY-- and then you have to choose between nuking files and converting to premium.

I don't use Dropbox for audio, video or photos, and I'm already using more than 3.3G. A lot of my folders are probably things I put in there to quickly share it with someone or access it from another computer. I could move them now and free up some space. But, I can't be bothered. That's one reason why people will convert to premium. The hassle of sorting through folders is probably not worth $100 to many.

I've created over 1GB .psd files just in couple of weeks and my total usage is now around 40GB. So I guess lot of designers need the pro version if they want to use Dropbox.

"And I think Dropbox probably has a much higher conversion rate than 5%."

Why do you think so?

"Dropbox Could Generate $100 Million In Revenue This Year" http://www.businessinsider.com/dropbox-revenue-2011-3

There is no way conversion is anywhere near 5%. The vast majority of users hardly use it.

That (brilliantly) is not directly related to conversion rate.

I think at some point the freemium model needs to be abandoned. Dropbox is getting there - 25 million users (or accounts) and a service that works pretty well. At this point why not charge a nominal fee for the low end of the service - even if it's 12 dollars a year (only $1 a month!). Or a few months free followed by a monthly fee. I'd be willing to pay after having used the service free for a while - although I personally will have to cause I need more storage.

Show some math.

25,000,000 @ 5% conversion == $125,000,000/year in revenue.

That sounds like a pretty amazing business model to me.

You are assuming the high end of conversion at 5%. And you are also assuming those 25,000,000 accounts are all held by different people, when in reality there are people who hold multiple accounts under different email addresses for different purposes or to avoid upgrading. It's not so easy to switch from one acct to another, probably a good thing on the part of DropBox, but people still do it.

Regardless, it's a great service and if it isn't already a great business I'm sure it will be.

The interesting question is then: how can Dropbox make more profit? By wiggling the conversion rate only?

Preferably, they are working at tangential services that leverage the data that is already stored, capitalizing on their assets more than once (i.e. normal Dropbox usage).

Ideas? I for one love the idea of integrating one's repository of files wherever relevant. Say, accessing your Dropbox/Documents in Gmail, your Dropbox/Photos in Flickr etc. If Facebook is the internet of social, Dropbox may very well be the internet of files, removing the barrier between online/offline even further. I'd pay for that.

Dropbox is now targeting businesses: https://www.dropbox.com/teams

They could invest in their own servers/storage which would be high up front costs, but lower once amortized than using a middleman for storage...

I'm all for napkin calculations but there are so many fudge factors and guesses in this that to draw conclusions would be meaningless. It's a shame, because there's some good information in the article, and some intelligent and reasonable estimation work, but sentences like "I've estimated the average stored file at 1.6MB based on my own account as well as those of a few people I've asked" are unignorable.

Very good analysis and impressive that they can survive with a burn rate of $3.1M - $5.8M per month without additional external funding.

If they are using S3, then their storage cost is close to 100% variable. Assuming their rate of paying customers stays consistent with their free customers, regardless of how much they grow, isn't it safe to assume that their model is tailored to at least cover their cost of storage with their revenues?

If that wasn't the case, it might be hard to justify giving them more money, unless they were very close to break-even on that cost. So I'm assuming that they're bringing in enough revenue to cover their storage cost, regardless if their storage cost is 100k / month, 5MM a month, or even higher.

If more privacy-conscious users start using TrueCrypt after the revelations made (Dropbox being able to decrypt your data) in their latest blog post, http://news.ycombinator.com/item?id=2470074 , that conversion rate will need to be higher.

True, though I bet 90% of users don't even realise there's been an "uproar" over Dropbox's security.

It's not even an uproar really. The only people who care are privacy freaks who didn't take the time to think through the implications of Dropbox's feature set when they first signed up.

i still wonder why dropbox and zumo drive are still not acquired ... eventually google, microsoft (may be even apple and amazon) will be offering the same service... so why invent the whole wheel again? these are the amazing YC companies ... whoever buys them won't be disappointed for sure...

ZumoDrive was acquired by Motorola. http://techcrunch.com/2010/12/22/confirmed-motorola-mobility...