Saturday 2 January 2016

App ratings and reviews


This blog post is on the app ratings and review systems on Apple iTunes, Google Play and Amazon App Store. All three of these stores provide a 5 star rating scale and a text review (title + body text).

The problems I see with the current review/rating systems are as follows:

  • Single score doesn’t communicate in much depth what the reviewer is trying to say
  • Single scores don’t benefit content discovery
  • Review text is unstructured

To improve upon the existing rating/review system, it's worth considering:

  • what the reviewers are trying to communicate
  • who the reviewers are communicating to
  • of that which is being communicated, which is the most important

What reviews/ratings are trying to communicate:

  • Quality
    • Look and feel / graphics
    • Features
    • Bugs
  • Extent of recommendation
  • Value for money
  • Ideas for the app / ideas for derivative apps
  • That there is / isn’t a market for future apps of this type

Who reviews/ratings are trying to communicate to:

  • Other potential customers
  • Developer
  • Other developers
  • The App Stores (typically to complain about an app)
What is most important:
  • For me, the most important thing that reviewers are communicating is the extend of recommendation

Suggestions for a better system

Each review would have the following sections:

  • Message to the developer (private)
  • Over-all rating (5 star scale)
  • Over-all review
  • Set of recommendations (see below)

Three-variable recommendation system:

  • Who the recommendation is for (tag-like system, when the user starts to type the name of a group, they're shown options to pick from) (e.g. fans of X)
  • Why you're recommending it / What you think they’ll get out of it
  • How much you're recommending it
Users would be able to follow recommendations streams to help them find new content.

Structuring notes with time

Consider the following two scenarios:

  • you write a diary, in which every now and then you write something about your objectives
  • you have a list of objectives that you add to every now and again

The former scenario is less structured. Whilst it might be possible to search the diary for entries containing text about career objectives (or use tags), there may be duplication in the results, or incompleteness. And the data is unlikely to be in exactly the same format (unless some kind of standard format is specified in advance).

The latter scenario doesn't (generally) have time information, which inhibits the ability to track how one's objectives change over time. Even if the objective list was actually a table with two fields, objective and date added, that still wouldn't capture whether the objective made the list at each review point over time.

Is there a general name for this type of problem?