This is actually the formula for compound monthly growth rate, a modification of the commonly used compound annual growth rate (CAGR), with an extra -1 thrown in after num_periods. The correct formula for compound growth is:
1. Examine your top (most engaged) x users each month
2. Quantify turnover in that group
3. Qualify which features/content they are using more/less than quantify more/less
4. Evaluate (3) against your business model - are your priorities focused where your top users are focused and does your business model work if that is where the focus remains?
--
Can spin your own version of above but believe the key is to focus on your most active, engaged cohort...the cohort that will ultimately sink or swim your service. Other analysis can be valuable too but not if it distracts from understanding your key user/customer.
This, I agree with. In general, very few companies make up a large percentage of the profits, illustrated perfectly by the whale curve. It is quite shocking to see that thing pop up when you run it across real data, but it shows you instantly who to keep happy.
It's a terrible strategy to play Pokemon (gotta get 'em all) with your business. No matter how hard you try, you simply cannot increase profits over certain customers and you risk collapsing your profits if you place your effort there.
Keen IO is an analytics backend as a service. We built it so that developers wouldn't have to build in-house analytics infrastructure ever again. You can use our SDKs to track any event in your mobile app, website, or backend, and we expose powerful analytics features by API so you can do whatever you need to do with your analytics data.
There's no easy drop-in solution that will give you all the answers for your specific business, but we tried to make our platform extensible so that it can be used answer any type of event-based analytics questions (like those mentioned here).
Plugging my own thing again, I built http://www.instahero.com to solve this exact problem. It has many pre-made metrics where you just select your inputs and it gives you the metrics on a dashboard/report.
What we've found is that different analytics services help us answer different kinds of questions.
For example, here are the tools we use to find each metric he mentions:
Distribution
1. New users by channel: Google Analytics Source breakdown satisfies what I need to know here. If you want to look at conversion to a Goal broken down by source, there's the "Goals" view in GA.
2. Total users: easiest for us here is to just query our database.
3. Cost of customer acquisition, lifetime value, pay back period: again Google Analytics if you're doing adwords advertising.
Engagement
1. Active users: here a tool like KISSmetrics, Keen IO or Mixpanel will let you see how many people are really succeeding in using your service.
2. % of users using top 3 key features in a given month: same tools as active
Revenue
1. Revenue growth: Stripe or w/e your payments solution is.
2. Conversion to paid: KISSmetrics and Mixpanel are the best tools to help you measure conversion rates.
3. Avg spend per paying customer: Stripe or w/e your payments solution is.
4. Churn rate: Mixpanel probably does this best, but querying your database might be a better option depending on how you measure churn.
5. Burn rate: your bank account!
What we've found is that no single tool answers all these questions well. Most companies we talk to use a mix of Google Analytics for basic stats and distribution measurement. KISSmetrics, Mixpanel or Keen IO for engagement metrics and tracking. Customer.io, Vero and Klaviyo for email marketing analytics, etc. etc.
<shameless plug>
This is why we built https://Segment.io, so that we could easily send our data wherever we wanted, and get the best analytics answers from many different, complementary services.
This is actually the formula for compound monthly growth rate, a modification of the commonly used compound annual growth rate (CAGR), with an extra -1 thrown in after num_periods. The correct formula for compound growth is:
The T6m average is the arithmetic average of, surprise, the last six months: where x is a vector of the last 6 months' data (sales, users, etc.). The T6m average will be in the same unit as the monthly figures, not percentages.