This is the last post in this series on modelling product revenues. We started with building a static profit&loss statement in step 1. In step 2 we introduced variability and a way to summarise the results of a lot of experiments. In this last part in the series the focus will shift to actually looking at how to use this model to understand tradeoffs. Before that though we’ll slightly need to expand on the use of datatables, so by all means check back to step 2 to get a refresher on how they work.
The real key to these models is that they allow comparing different factors in a common denominator
As usual there’s an excel file to go with this post and just like the previous one it’s set to not automatically recalculate. If you want to see it update, on windows you need to go to the options menu (click the office button in newer versions) and then go to the “formulas” category. There you’ll find calculation options. Under workbook calculations you should select “automatic”, which updates everything and is the default in excel. Mac users, this is under preferences (Cmd+,), calculation and there under “calculate sheets” you’ll find the same options.
Expanding on datatables
There are 2 new simulations in the spreadsheet, one dealing with churn and one with the number of new contacts you reach every year. They’re both constructed in the same way so I’ll only explain how to build the one called SimChurn as it contains a couple of new excel techniques building on the use of datatables.
First, the things that are the same as in the previous setup: cell A2 contains the result we want store for every experiment, cell A3 to A102 contain 1 through 100 for the number of experiments we want to run.
However, rather than having a single column for results we now have column headings that represent various levels of churn we want to try out.
What we will automate here is typing a level of churn in our assumptions and then running 100 experiments for each of the levels we want to try out.
If you remember from last time, the datatable allows 2 inputs: a row input cell and a column input cell. What that means is that we can substitute the values in our header row (B2 to F2) in a certain cell in the spreadsheet and then run the 100 simulations with that value fixed. Similarly you could run this with the column (A3:A102) substituting that as you go along. There is one limitation however, you can only substitute the values to a cell on the same tab (I.e. SimChurn). In order to handle that we use cell A1 to hold our churn value (20%) and on our customer sheet we simply put the churn value equal to “SimChurn!A1” … The “real” value for churn is now stored on the SimChurn tab and the value on the Customers tab is simply a link to it.
So in order to construct the actual datatable and vary the churn rate we select cells A2 to F102 and go to the data menu —> datatable. The row input cell is cell A1 (I.e. We’ll substitute our desired churn values into that cell, and because it’s linked to the churn value on customers our other formulas will use it). For the column input select any empty cell as we don’t want to use this value in calculations.
The result is that you now have 5 times 100 results, each column being calculated with the churn rate of the header. For each of those 5 columns you can perform the same analysis we did previously by building the summary, standard deviation table and histogram.
The picture below has the 5 histograms in one picture. As you might expect, the lower the churn level the higher the lifetime value will be.
As already mentioned, the second simulation is built in exactly the same way and varies the number of customers you initially meet. No surprise there, more customers initially means higher lifetime value.
So far we looked at several individual elements but haven’t really got to actually making any tradeoff. We simply looked at the effect of varying certain variables in our model. It’s time to turn to the summary tab now, which is where everything comes together.
The summary
The base structure is very simple. The summary table of “SimBase” is labelled baseline. And the same calculations (mean, standard deviation, min, max, 95% interval) have been calculated for every level of churn and for the various levels of new contacts / year we put in the respective simulations. So on to understanding tradeoffs.
If we do nothing at all, our 95% range for the lifetime value is about $ 488.000 to $ 535.000. This is not the amount you could invest as you already accounted for costs of building/running in your P/L. This is anticipated profit over 5 years (ignoring inflation).
You now have an estimate for how much a certain product improvement is worth over the lifetime of a project.
Now here’s killer insight number 1: reducing your churn by 50% (I.e. From 20% to 10%) moves your lifetime value up by about $ 200.000. So in theory it’s worth spending $199.999,99 to achieve this.
Killer insight number 2: Reaching out to 115000 people every year also moves your expected lifetime value up by about $200.000. Unlike the churn assumption, the cost for reaching these extra 15000 people is already captured in the cost of the P&L so in this case it’s a matter feasibility in finding 15000 extra contacts at the assumed quality.
Killer insight number 3: Combine 1 & 2 and ask yourself the question if you were to reach 15000 people more every year for 5 years would that be more or less feasible than trying to cut your churn rate in half.
