DISQUS

Andrew Chen (@andrew_chen): Matt Humphrey of Bumba Labs on User Retention Curves

  • Tara Kelly · 5 months ago
    Spot on analysis. Retention is definitely key, especially for the premium model. If, say, your free users take 3 months to garner up the need to purchase a premium upgrade, then it's pretty simple math that you have to keep them around that long in the first place.

    What's harder to measure is the less-linear effects of retention, for example, sharing. Folks invite others to join up so they can share with them. The more people that join, the easier the sharer's life gets. If those new users aren't sticking around, your service becomes less useful to your original sharer. A low retention could easily kick off a downward spiral here.
  • Bhanu Sharma · 5 months ago
    Andrew, great post.

    Went through your example above, and can't figure out your calculation behind the 1 month churn.

    "In the short run, the numbers are close to the same:

    * 80% monthly retention, after 1 month = 600
    * 90% monthly retention, after 1 month = 700"

    Shouldn't that be 800 and 900 instead?

    80% monthly retention, after 1month= 1000*(0.8)^1=800
    80% monthly retention, after 2months= 1000*(0.8)^2=640

    I am sure I missing something obvious.
  • Andrew Chen · 5 months ago
    Yes, typo ;-) Let me go fix it.
  • Brian · 5 months ago
    Great post. This is something we pay close attention to at HomeStars. Coming from a telecoms background I look at in terms of COA (cost of acquisition), churn, and average monthly (or annual) revenue ARPU.
    So the general formula to remember is ARPU * (1/churn) - COA = lifetime value of the customer.
    And, as you rightly point out, churn plays a big effect here. Increasing the revenue from your customer base can be quickly wiped out by losing them quickly.
    It also notes another interesting fact - it's okay to spend money to get customers - JUST MAKE SURE YOU KEEP THEM!.
  • Jeremy Nusser · 5 months ago
    Nice post! - great overview of why you need to back up customer acquisition (and viral marketing) with solid customer retention. Only sticking point is that "retention rates over 90% are unrealistic". We have several customers with retention rates over 90% - it depends on the industry and offering.

    Also, we (@Vindicia) have a best practices guide for those who are looking for ways to improve customer retention - http://bit.ly/cust_retention
  • Ilya Grigorik · 5 months ago
    This model assumes a constant attrition rate.

    Without doing much digging I imagine that's a reasonable simplification, but I wonder how it plays out in real life? Have there been any studies on how user attrition rates change as a function of time since they signed up?