Pick the choices you'd make at launch. The estimate comes from about 800,000 Google Play apps released in 2018 and 2019, and how many of them had reached 100K installs by mid 2021. It starts filled in with an example app.
Estimated chance of reaching 100K installs
-
It's a logistic regression trained on every app released in 2018 and 2019 that had a known size and content rating. Before fitting the final version I trained it only on 2018 launches and tested it on 2019 launches, and it scored 0.84 AUC on those unseen apps.
It shows what went along with success, not what causes it. Apps that already have users tend to add in-app purchases and websites later, and bigger apps usually had more work put into them. So treat the levers as patterns worth knowing, not guarantees.
Data is a June 2021 scrape of 2.3M Google Play apps. Full analysis and code are on GitHub.