Feature Usage Across Trials

Data Collection

We’ve already collected data on trialists in previous analyses, so we’ll reuse those datasets.

# get existing trials data
pro <- readRDS("pro_trial_actions.rds") %>% mutate(trial_type = "Pro")
premium <- readRDS("premium_trial_actions.rds") %>% mutate(trial_type = "Premium") 
business <- readRDS("business_trial_actions.rds") %>% mutate(trial_type = "Business")

# merge data frames
trials <- pro %>% 
  rbind(premium) %>% 
  rbind(business)

# remove unneeded data
rm(pro); rm(premium); rm(business)

Data Tidying

We’ll create a couple new fields that tell us some more information about the trialists. We will also take an important step and remove users that did not connect any profiles. This is an attempt to reduce noise. We will also add boolean variables that let us know if the trialist has created at least one post and another that tells us if the trialist has at least three posts.

# add boolean columns
tidy_trials <- trials %>% 
  filter(profiles >= 1) %>% 
  mutate(started_at_signup = trial_start == signup_date,
         trial_type = as.factor(trial_type),
         used_first_comment = updates_with_first_comment > 0,
         used_calendar = calendar_actions > 0,
         viewed_post_analytics = viewed_posts > 0,
         viewed_overview_analytics = viewed_overview > 0,
         exported_csv = exported_analytics > 0,
         created_draft = draft_created > 0,
         used_shop_grid = shop_grid > 0,
         shared_update = updates > 0,
         added_profile = profiles > 0,
         used_extension = extension_updates > 0,
         used_mobile = mobile_updates > 0,
         has_update = updates > 0,
         has_three_updates = updates >= 3)

Exploratory Analysis

Let’s start this analysis by visualizing certain distributions for individual actions, and comparing them across trial types. We’ll start with profiles.

Profiles

The plot below shows us the proportion of trialists that added a certain number of profiles before their trials finished. It’s interesting to note that the three distributions are similar in shape. Users that are on Business trials are more likely to add more profiles, but we still see that the majority have five or less.

What if we looked at users that had three or more updates? The plot below includes only those users. We can see that the distributions have a more distinct shape with a “center” of sorts at 3 profiles. There is skew towards the long tail, but generally the distribution of profiles is centered around 3 profiles.

Posts Created

Let’s see how many times these users created posts. The plot below only includes trialists that created at least one posts. The distributions are similar in shape, with most trialists creating between one and fifty posts. Users on Business and Premium trials tend to create more posts than Pro trialists.

Extension Usage

Below we can see that most trialists did not use the extension. I suppose it’s most useful for Twitter anyway.

Mobile Usage

Users aren’t taking much advantage of the mobile apps, even if they have created three posts in total. The distributions are similar for all three trial types.

Post Analytics

Let’s see how many times these users viewed post analytics. The plot below only includes trialists that created at least three posts. We can see that Premium trialists are more likely than Pro or Business trialists to view post analytics. It’s arguable that it’s easier to start Pro and Business trials, so trialists can be somewhat less qualified. The distributions are similar in shape.

Overview Analytics

We can see that Premium trialists are more likely to have viewed overview analytics during the trials. These percentages only include trialists that created at least one post.

Export Analytics

Almost no one exported an analytics report.

Drafts

I only included users with at least three updates below. Premium trialists were more likely to try drafts.

Shop Grid and IG First Comment are relatively new features, so I’ll look at them last.

Instagram First Comment

Again, Premium trialists are more likely to have used the first comment feature. The following trialists started their trials on or after July 1, 2019.

Shop Grid

Premium trialists are more likely to have used the shop grid feature. The following trialists started their trials on or after July 1, 2019.

It’s hard to see any other differences because so many people haven’t used the feature. Let’s only look at people that have used shop grid at least once.

The distributions below show some interesting characteristics. Of those that used Shop Grid at least once, Business trialists were significantly more likely to use it again. A much higher proportion of Premium trialists only used the feature once.