In Localytics, an Event is a timestamped record representing a particular behavior which has occurred within the app. For example, an e-commerce app might trigger an Event called "Added To Cart" each time an item is added to the app's shopping cart. This traditional, granular approach to Event-tagging has the benefit of representing each behavior as a unique, individually-timestamped record, which means that if a user adds five items to a cart during a session Localytics will capture this with five unique timestamped Events (with one data point for each). This approach is advantageous when specific details about each behavior are required for reporting or are required to power downstream engagement campaigns. For example, if you want to be able to see which categories of items are being added to the cart.
However, it can also be advantageous to capture behaviors using a so-called Session Summary Event, particularly in cases where tracked behaviors occur with great frequency. For example, consider a social photo-sharing app that wants to report on the number of photos that are liked during a session. The traditional, granular approach would call for every individual "Photo Liked" event to fire independently, which produces a high volume of data points and is likely to be noisy for reporting purposes. An alternative approach would be to use a Session Summary Event.
In this case, rather than tagging an individual Event every time a photo is liked, the app can be configured to keep track of the number of photos that have been liked during a session and then record a "Session Summary" Event at the close of the session with an Attribute called "Number of Photos Liked". In a gaming app, the same concept could be used to trigger an event called "Level Summary" with Attributes such as "Number of Attempts to Reach Completion", "Number of Points Earned," etc. In a news app, this might be used to report on the number of ads seen during a session.
To summarize using an analogy: Consider a basketball game where there are 100 shots that are made over the course of the game. One way to represent this would be to capture every individual shot as a unique, timestamped record with its own individual Attributes (which would produce 100 data points). Alternatively, the summary approach would involve capturing game-level summary statistics in a single record that is triggered when the game is completed, with Attributes that record what happened during the game -- e.g. number of shots attempted, number of shots made, etc. (which produces 1 data point). This summary approach can be advantageous both for reporting reasons and to ensure that data points are used strategically and sensibly.
In order to use Summary Events, your developers should configure your app to keep track of statistics that will be written as Attributes (e.g. in the screenshot above, the app would keep track of "Ads Viewed" and then pass this to Localytics as an Attribute). Developer documentation for tagging events is available on our Docs site (e.g. iOS, Android, Web).