Churn is one of the most important SaaS metrics. Let's dive right into it, exploring what churn really is and how to calculate it properly.
Definition
Churn (aka churn rate or cancellation rate) can be defined as ratio between the number of customers that have left (cancelled) the service in a given time period versus the total number of customers who could have done so.
Calculating Customer Churn
First let’s make sure we approximate churn correctly. There are many definitions, and you can easily be calculating it wrong.
Let’s take an example of a service that has 100 customers at the beginning of a month, then during that month gains 25 new customers, while 5 cancel (churn), leaving you with 120 customers at the end of the month.
I’ve seen different (typically more wrong) calculations for churn:
- Take what you had at the beginning of period (5/100 = 5%)
- Take what you had at the end (5/120 = 4.16%)
- Take the average (proposed in the Lean Analytics book) resulting in 5/110 = 4.54%
Following the definition given above, the correct calculation for churn would be 5 (number of customers that churned) / 125 (total number of customers that could have churned) giving 4% churn.
Calculating churn like this also means you can look at churn as counter-growth.
If one looks at growth as what they had at the end of the period divided by what they begun with, it makes sense to look at churn as a ratio of what one ended up having (because of churn), divided with what they could have had if there was no churn. In this example 120/125= 96% retained, 4% lost (or churned, same as above). This concept does not require thinking about cohorts if you want to see the big picture – similar to how when we look at growth we usually do not care whether the customer signed up on the very last day of the interval being looked at.
it’s worth noting that there are more complex ways to calculate churn. See here how Shopify or WP Engine do it.
Monthly vs Annual churn
While 4% may not sound as much this formula will help convert it to annual churn.
annual churn = 1 – (1 – monthly churn)^12
So for 4% monthly churn we get
1 – (1 -0.04)^12 = 38% !
This means that this company will have to replace 38% of customers each year just to stay with the same number of customers.
Typical churn rates compared
As said in the beginning, churn rates vary from market to market. Hosting business may have 2% to 3% monthly churn, while a dating service like Match may have a whooping 30% monthly churn. The investors may have their own view, for some only 5% to 7% annually is acceptable.
The table below shows known annual churn rates (source).
Annual Churn | Company | Industry | Data Year | Country |
1.0% | Cox (triple-play customers) | Cable TV | 2002 | US |
4.0% | C I Host | Web Hosting | 2003 | US |
4.0% | Earthlink | Internet Service | 1999 | US |
4.5% | Greenberg Traurig (lawyers) | Legal | 2003 | US |
4.6% | Reed Smith (lawyers) | Legal | 2003 | US |
4.6% | Sonnenschein (lawyers) | Legal | 2003 | US |
4.9% | Local Telecom | 2000 | US | |
5.0% | Piper Rudnick (lawyers) | Legal | 2006 | US |
5.3% | Baker & McKenzie (lawyers) | Legal | 2005 | US |
5.7% | Local Telecom | 2002 | US | |
6.0% | Sirius | Satellite Radio | 2005 | US |
6.0% | Wireless | 2002 | US | |
6.1% | Local Telecom | 2001 | US | |
6.3% | Holland & Knight (lawyers) | Legal | 2004 | US |
6.4% | White & Case (lawyers) | Legal | 2003 | US |
6.5% | McGuire Woods (lawyers) | Legal | 2003 | US |
6.6% | Sirius | Satellite Radio | 2004 | US |
7.0% | Morgan, Lewis (lawyers) | Legal | 2003 | US |
7.2% | Sirius | Satellite Radio | 2006 | US |
7.5% | Nextel | Wireless | 2001 | US |
7.9% | Howrey (lawyers) | Legal | 2003 | US |
8.0% | Triton PCS | Wireless | 2001 | US |
8.0% | U.S. Cellular | Wireless | 2001 | US |
9.0% | Wireless | 2001 | US | |
9.6% | Duane Morris (lawyers) | Legal | 2003 | US |
9.6% | Wireless | 2000 | US | |
10.0% | Web Hosting | 2003 | US | |
10.0% | Western Wireless | Wireless | 2001 | US |
10.3% | Akin Group (lawyers) | Legal | 2003 | US |
11.0% | Alamosa PCS | Wireless | 2001 | US |
14.0% | Virgin Mobile | Wireless | 2005 | GB |
15.0% | Nascar.com (premium subscribers) | Sports Media | 2004 | US |
16.0% | Nextel | Wireless | 2005 | US |
17.0% | Colorado teachers in ‘excellent’ schools | Education | 2004 | US |
17.0% | Schnader Harrison (lawyers) | Legal | 2003 | US |
17.0% | DBS TV | 2002 | US | |
17.2% | Vodaphone | Wireless | 2005 | IT |
18.0% | DirecTV | DBS TV | 2003 | US |
18.3% | Vodaphone | Wireless | 2005 | DE |
19.0% | Alltel | Wireless | 2005 | US |
20.0% | Hutchison Telecommunications | Wireless | 2005 | IN |
20.0% | Wireless | 2001 | AU | |
21.9% | Vodaphone | Wireless | 2005 | ES |
22.0% | Analog cable subscribers | Cable TV | 2002 | US |
23.0% | Cingular | Wireless | 2005 | US |
23.0% | Sprint | Wireless | 2005 | US |
23.0% | Colorado teachers in ‘unsatisfactory’ schools | Education | 2004 | US |
25.0% | Wireless | 2005 | GB | |
26.0% | Sprint | Wireless | 2005 | US |
26.0% | Subscribers | Cable TV | 2002 | US |
29.7% | Vodaphone | Wireless | 2005 | GB |
30.0% | LD Telecom | 2002 | US | |
31.0% | Globe | Prepaid Wireless | 2003 | PH |
31.0% | Pagers | 1998 | US | |
34.8% | T-Mobile | Wireless | 2005 | GB |
35.0% | Maricopa County (anglers) | Recreation | 2002 | US |
36.0% | Las Americas – Cable California | Cable TV | 2002 | MX |
37.0% | E-mail addresses | 2003 | US | |
45.0% | E-mail addresses | 2004 | US | |
46.0% | Prepaid Calling Cards | 2004 | US | |
46.0% | Digital cable subscribers | Cable TV | 2002 | US |
51.0% | Globe | Prepaid Wireless | 2004 | PH |
52.0% | Florence (AL) Times Daily (readers) | Newspapers | 2005 | US |
58.0% | Snowball.com | E-mail newsletter | 2000 | US |
78.0% | Touch Mobile | Prepaid Wireless | 2004 | PH |
93.0% | VOOM | HD TV | 2004 | US |
93.0% | Runoff at time of sale | Home Mortgage | 2002 | US |
If you can control churn within reasonable limits for your market, you are doing OK and should focus on acquisition. However, let churn go wild and you are in deep problem; all your acquisition efforts will be spent chasing churned customers which will cause you to hit the SaaS growth ceiling, a topic we will cover in a separate article.