Hacker News new | past | comments | ask | show | jobs | submit login
How to Track Six Key Metrics for Your Web App (thinkvitamin.com)
107 points by ryancarson on Sept 1, 2010 | hide | past | favorite | 9 comments



Your spreadsheet looks pretty cool. I think though that you could really do with a little more explanation about why it's doing some of the things it's doing.

Take, for example, Customer Acquisition Cost Ratio (CACR) for the 2nd quarter (cell [H33])

Going by the formula that's:

  [H33] = CACR for 2nd Quarter = 
  ( 
    ([CMRR for June] - [CMRR for prev March]) 
    * 4 
    * ( [Gross profit for Q2] / [Total CMRR for Q2] ) 
  ) 
  / [Marketing costs for Q1]

Why do we only use the numbers for March and June? What is that magic 4? Why are we dividing by the previous quarter's marketing?

Your post is your chance to explain why you are right and why we should trust your spreadsheet. Don't just wave your hands and say "it's complicated".


Appreciate the feedback. I learned how to calculate those numbers from Bessemer Ventures PDF, which I linked to in the article. Essentially you're taking the difference in CMRR for the quarter and annualizing it, then dividing by the marketing cost for the previous quarter.


Just updated the post, BTW.


This is going to be extremely useful for me. I know the basics (churn, CMMR, etc.) but I'm missing some of the others.

I'm not quite sure I understand why CPA is so simple. Right now I'm not spending a dime on advertising so I'm not spending anything to acquire these users (it's WOM and SEO). Once I start advertising shouldn't that calculation go against customers customers acquired through my advertising efforts instead of all new users?


Marketing costs are only directly related to the new customers you acquire the same month (or sometimes the following month - as long as you're consistent). This isn't always strictly correct, but it should give you a good average CPA value.


I'm in the tricky situation of not having enough users within BitBuffet.com to really gauge these numbers accurately. I'm currently paying about 8 months of customer revenue to pull in a new paying customer, but honestly, that number isn't reliable.


How much "throughput" do you need before stats like this start to mean anything? For example, if you only acquire 5 new users per month, can you really calculate a meaningful churn rate?


If you only acquire five new users a month, then something is seriously wrong.


No, but you can use past months aggregate total to estimate churn rate.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: