Dropbox


A few years ago, Dropbox added the bar to the user's profile indicating how much space the user had used up.

Screen Shot 2015-12-04 at 2.31.55 PM.png


This visually tells the user what percentage of space they have used and what is remaining, helping them predict when they need to get more space. It is said that this bar, which was an intern's project, paid for Dropbox's entire internship program up to that point and then some.

This could be taken one step further though. What if the user could see their history of usage and a prediction of how soon they would run out of space based on current rates of usage? Users would be able to very clearly see what space they are using and not using and could tell far in advance when they need to get more space through either buying Dropbox Pro or engaging in other space-gaining activities such as referring friends.

Problem:

It can be hard to predict when or if you will use up your Dropbox space and the moment when you finally do use up your space, it is not likely to be a good time to deal with buying more or engaging in other space-gaining activities. The things like referring friends in particular take time that you do not have when in the middle of trying (and failing) to upload a file to your full Dropbox.

Hypothesis:

If Dropbox were to give users more insight into how they are utilizing their accounts and give them an estimate of when they will run out of space, users would be more likely to use their accounts more when they buy more space or engage in other space-gaining activities like referring friends.

Action:

A/B test users by showing a small percentage (say 0.5%, or 1.5M people) a graph of how much space they have used over time along with an estimation of how long until they have used up all their space. To take it even further, it could be an A/B/C/D test with two other groups as well - those who see just the graph and those who see just the estimate.

FullSizeRender.jpg
A: Graph and Estimate


FullSizeRender-1.jpg
B: Just Graph


FullSizeRender-2.jpg
C: Just Estimate

D: Control Group with just original bar

Metrics:

There are several key things to measure about the results of this experiment:
  • Users actions before and after seeing the graph and/or estimate.
  • Users actions who never see the graph and/or estimate
  • Time users begin engaging in space-gaining activities before their account is full (with and without the messages).
  • Average level of engagement of users with and without messages. (Number of files uploaded/shared and other activity)
  • Whether predictions correctly predict usage patterns for both those who are told the estimate and those who are not told (because simply seeing the estimate can cause users to change their usage pattern). If we show users these messages, then they should be roughly accurate.

Evaluation:

See whether users change their usage patterns after seeing the messages - do they engage in space-gaining activities or delete things or both? Are they more active and use more of the space they own or do they become more frugal, decreasing usage or even deleting files.

Which messaging had the greatest positive affect on users: graph, estimation, or both? (Positive affect being started earlier getting more space when needed and more fully utilizing their Dropbox space.)