Ask HN: What's your churn rate?

[mattjung]

A question to web-application providers: what percentage of your clients quit your service every month? is the rate different for free accounts compared to paying accounts?

p.s. Any links to some stats?

[jasonkester]

Here are my numbers for S3stat. We don't have free accounts, though we do have a 30 day trial.

  Visit to trial conversion:   3%
  Trial to paid conversion:    26%
  Average subscription length: 11.5 months
  Average cancels per month:   2%

Over time, those last two numbers keep getting better, since there's more time passed for our first loyal customers to inflate the average subscription length, and there's more people on board to lessen the impact of a few people cancelling.

It's worth noting that you only need to calculate one of those last two numbers to know the expected lifetime value of a visit to your site. Until you asked the question, I had never actually calculated my churn rate as a percentage of active users.

[paraschopra]

Interesting stats. How long have you been running S3stat?

[jasonkester]

The beta went live at the end of 2007, and I started charging for it in February 2008.

S3stat goes squarely against one of the biggest pieces of advice I give to fellow startup kids: Never base an entire product off a missing feature in somebody else's thing.

Thus far Amazon have been good in keeping their AWS reports in the same state of disarray I found them in 3 years ago. I don't expect it will stay that way forever.

[paraschopra]

I'm sure by then you will find other successful niches :)

[tnorthcutt]

Do you have any sort of automated solution in place to calculate those stats for you on a running basis, or do you just manually re-calculate at set intervals?

[jasonkester]

I do a little obsession-fest at the start of each month where I pull down income reports for the services I run, update some spreadsheets, and roll around naked in piles of virtual money.

If I could have the bank deliver it all in the form of one dollar bills spilling from canvas bags with dollar signs stenciled on the side, that would help the process immensely.

[pplante]

No public stats I can personally provide. But I have seen it vary wildly depending on the type of service. Businesses are generally slower to adopt a new service/tool, and once they are locked in can be apprehensive for change. Consumers can change on a whim if something better comes along.

Here are some interesting stats I found for Evernote: http://techcrunch.com/2010/05/28/video-evernote-ceo-phil-lib...

Evernote has been pretty open in their stats over the course of their existence.

[dan_manges]

It can also vary a lot within a type of service depending on the target market. At Braintree our churn rate is higher for the start-ups that we work with (because they sometimes go out of business) than it is for companies that have been around longer and are profitable.

[petervandijck]

Agreed with businesses: once the charge on the CC is approved, they will be slow to remove it (because re-approving it can be though).

[paraschopra]

Constant Contact is a public company, there for they share stats in their annual report http://investor.constantcontact.com/common/download/download...

Some key stats. Their churn is ~2.5% and their cost of acquiring customer is (whopping!) $370

PS: planning to write a post analyzing all such open stats. Do you know any other public Internet companies?

[petervandijck]

This might help too: for really crap services (say, weightloss or the like), where people sign up and then pretty much never use the service, a typical average churn rate is 3 months. ie. it takes people 3 months on average to unsubscribe from a service they don't use.

[paraschopra]

Are you speaking out of experience or do you have some stats? It seems unlikely that people take at least 3 months to unsubscribe a service they never use.

[petervandijck]

Out of experience with 2 large services. Can't share stats, sorry. It may seem unlikely, but it's true :) I'm talking average, many take up to 6 months. Now you know why there are so many ads for crappy services online :) ps: I didn't say "at least", I said "on average".

[petervandijck]

The math of subscription services generally goes like this: acquiring a paying customer is going to cost you about 15-60$ each (for typical lifestyle services). So lifetime value needs to be more than that. Once you're live, you tweak those numbers, make acquisition more efficient and increase lifetime value.

[thebigjc]

Is there an equivalent or similar metric for non-subscription commerce sites? I operate one, and I'm struggling to find good references on the best way to determine I've 'lost' a customer.

[pospischil]

Im working on a service, currently in private beta, that I think you (and lots of others here on HN) would be interested in.

Shoot me an email and I'd love to tell you more about it: Pospischil at gmail