Originally published October 2019. Some platform details have changed since; the spreadsheet method still measures association, but attribution changes can alter which installs are labelled paid or organic.
Short answer: the 2019 video plots paid against organic installs and fits a trendline. Its example slope is about 0.26 organic installs associated with each additional paid install. That is a descriptive relationship, not proof that buying one more install causes 0.26 extra organic installs. Watch it on YouTube.
Page Contents
Why do paid installs bring organic installs?
The video gives several reasons. Paid users invite others or tell people by word of mouth, and some users counted as organic were actually driven by an ad but are not attributed to it. These are possible explanations for an association; shared seasonality or changes in attribution can also move the two series together.
How do you measure it?
- Chart daily paid and organic installs for the period you want to analyze, and take a seven-day rolling average of each, which smooths out day-of-week spikes.
- Check the association. The video uses the correlation coefficient. A value closer to 1 means a stronger positive linear relationship. A low value does not rule out delayed or nonlinear effects; inspect the data and relevant country or operating-system differences.
- Make a scatter chart with paid installs on the horizontal axis and organic installs on the vertical axis, and add a trendline.
- Read the equation. The example is about 0.257x + 204, with paid installs on the horizontal axis. The intercept is the fitted value at zero paid installs; it is not proof of a guaranteed 204-install organic baseline.
- Treat the slope as a hypothesis. The video proposes using it as an organic multiplier in financial models. Validate that assumption with a controlled spend or geographic comparison before counting it as incremental growth.
What should you keep in mind?
The numbers are one app’s example from 2019, not benchmarks. Overlapping seven-day averages smooth noise but can make correlations look stronger; common trends and lagged effects still need checking. For a worked version with trials instead of installs, see how to analyze the relationship between paid and organic trials. The multiplier is used in how to model out DAU from a retention curve, and for how paid and organic installs get attributed today, see how MMPs work in a post-ATT world.
Frequently asked questions
Do paid installs drive organic installs? They may, through invitations or word of mouth, but a correlation between paid and organic installs cannot establish how much growth the advertising caused.
How do you quantify organic installs per paid install? A fitted slope summarizes the association in the observed data. To estimate installs caused by paid activity, compare outcomes against a credible no-extra-spend control.
What if the correlation is weak? Check data quality, time lags and relevant segments. Do not keep subdividing until you happen to find a strong relationship.
