Rolling Quick Ratios for Product Decision-Making
We have previously written about the user growth quick ratio as a a good way to understand the health of a product. We are hearing feedback from different product teams we work with that they quite like it as a metric but want to understand better how they can use it for actionable decision making. So in this post we discuss how to…
- Look at quick ratio on a rolling basis on a KPI dashboard so you don’t have to wait until the end of the month;
- Use different window sizes to see long-term trends and look past seasonality effects; and
- Compare growth efficiency in the important segments of your business, be they marketing channels, geographic areas, or platforms.
Using Rolling Quick Ratio for Responsiveness
You will recall that the the user growth quick ratio is a comparison of two different time periods of equivalent length. For example, calculating a monthly quick ratio calculation involves summing the new and resurrected users in each month and dividing by the number of customers who failed to continue active usage from the previous month (churn). The result is a monthly chart that looks like the following (source: TVC internal data):
You don’t need to wait until the end of a month to make a monthly quick ratio calculation. To make it a daily metric, compare the most recent 28 days with the previous 28-day period and make the same (new+resurrected)/churned calculation. (Note: we prefer to use 28 days versus 30 to approximate a month because it removes day-of-the-week effects.) A plot of the 28-day rolling quick ratio for the same product and time period as the above chart looks like the following:
Note that the 28-day quick ratio drops from 2.5 on May 13 to 1.3 on May 31. By continuously tracking the quick ratio, this product team would notice the steep decline in growth efficiency — mainly due to increased churn — in near-real-time and not have to wait until the end of May to see the drop. Plotting the 28-day quick ratio on a daily basis helps us monitor the pulse of growth efficiency and quickly address any downturns in one of a startup’s most important metrics.
Varying Window Size for a Complete Picture
Another way to understand the growth dynamics of a business is to vary the length of the quick ratio window. The example chart below compares 84-, 168-, and 364-day windows for a low-frequency product with highly seasonal sales that peak in summer months. (Again, we use windows that are multiples of seven to eliminate any day-of-the-week effects).
The product’s inherent seasonality is manifested in the peaks and valleys of the 84- and 168-day window quick ratios (the blue and orange lines, respectively). Larger window sizes trail smaller ones and are, thus, not as helpful for responsive decision-making. In this example, the 84-day window leads the 168-day window by about four months.
Because understanding this seasonal business involves making year-over-year comparisons, one way to do so with growth efficiency is to look at the relative heights of the peaks of the seasonal 84- and 168-day windows. In this case, growth was more efficient in 2017 than in 2016, and 2018 is somewhere between the previous two years.
Another way to compare year-over-year growth efficiency is to look at the 364-day window (the red line), which helps us ignore seasonal effects. While it lags the other two windows, it helps us see the overall trend, which dipped at first (A, above), then increased from late 2016 to April 2018 (B), but has since started to decline again (C). Combined with the recent weakness in the other seasonal windows, this product is sending a strong signal that growth efficiency is in decline and needs to be addressed.
Segmenting Quick Ratios to Prioritize Resource Allocation
The rolling quick ratio can help visualize which customer segments are contributing positively to the overall quick ratio, and which are dragging it down.
The black line labeled “All” that cuts through the middle of the chart represents the rolling 364-day window for all customers that was shown in the previous section. The other lines show the quick ratio of the four largest geographical markets over time. The overall quick ratio tracks the shape of Location A’s red line, but is dragged down by lesser performing segments. Of the four segments, Location D (the red line at the bottom) with a 364-day quick ratio of just 2.1 is having the strongest negative impact on overall quick ratio.
The chart below breaks out each location’s growth accounting to identify what is driving their quick ratios.
Location D is showing the most orange Churn on a consistent basis, perhaps indicating a systemic factor. In addition, increased churn is also evident in Locations A, B, and C in the last four months or so. The product stakeholders should dive deeper into the systemic churn in Location D and the recent churn in A, B, and C in order to mitigate the drag on the whole company’s growth.
In addition to geographical locations, segmented quick ratio analysis can apply to any segmentation method, whether it is the marketing channel, product type, platform, or user persona.
All of our growth accounting functions, including the rolling quick ratio, are contained in growth_accounting.py. A sample for how to consume those functions to calculate the rolling quick ratio on a sample transactional data set is in rolling_qr_example.py. If you would like help using these functions, please let us know in the comments.