Beemray enables to build audience segments using algorithms based on location, in-app behavior, event attribute, profile, time, engagement metrics, app usage & other behavioral signals across all channels.
The cool thing about Beemray segmentation capabilities is that it enables utilisation and combination of any, also self-created, elements. To make it even more versatile, it also allows nested structures, and selecting any operator within a segment. Everything in real-time.
In practical terms, here are a few examples:
- Broad example: “Users who have not visited my app in the past 5 days”
- Granular: “Women in London who read a specific article and have not purchased a subscription”.
- Nested: “BMW drivers that have parked in Stockholm city center at least four times in a week and have taken a test drive of X1 or X5 at a local dealer.”
- Nested with many operators: “BMW drivers that have parked in Stockholm city center at least four times in a week but have not taken a test drive of a new X1 or X5 at a local dealer but did parked his car on southern Stockholm between 2016-01-01 and 2016-03-01”
The audience segments are the backbone of personalised push, in-app, and browser targeting, but they also drive connections outside the Beemray ecosystem as well. Integrate segments to your DMP, CRM, Facebook and other ad networks to reach your users and drive better targeting.
What is a segment?
Segment is a subset of devices which fulfill all the specified attributes within the segment. A segment contains one-to-many attributes based on user behaviour. Each of these attribute can contain conditional operators if the event fulfills this single attribute or not.
Segment can contain three types of attributes:
- Profile attribute describes demographical information that user has defined in the application about himself/herself like gender, age, marital status and hobbies.
- Event attribute describes a triggered event based on user behaviour. Single event can contain simple key-value-type structured data elements together with a event's title. As an example we can take the event "ADD_TO_CART" as an event title containing values of "PRODUCT_TITLE", "PRODUCT_PRICE"
- Location attribute defines the current location (lat/lon) of the user/device.
The flow of filling a user into a segment is dynamic and real-time: Example: A user executes an event in application (clicks a product page). This event is compared with all the segments of an account. If all the attributes of any segment matches, the client will be added as a new member of the segment. A beem content response will be delivered to the device if the segment and optionally other conditions of the device belong to a beem campaign.
Each attribute type has a different kind of condition option to check if the corresponding value from an event passes the condition rules of the attribute. Supported comparison operators for numeric attributes are < (less than), > (greater than), <= (equals or less than), >= (equals or greater than) and = (equals).
Supported comparison operators for textual attributes are "ilike" (case insensitive partial match), "like" (case sensitive partial match) and "=" (equals to). Also, well-know reqular expression matching is supported.
Supported operator for location attribute is always full comparison if given location is contained by the spot which has been set as an attribute. Each spot is geographical area of the map.
Devices that have fulfilled conditions and attributes of a segment can be used as a target audience for a beem. The updated segment data is available in real-time. A segment can also be analysed by compiling smart queries against the segment data to get more info of cumulative totals down to one hour resolution.
Segment location heat map is also available for more detailed analysis.
Integrations are either for feeding 3rd party data to Beemray or providing behavioral segments to external platforms. Ask for our Segment API documentation to learn and integrate external systems such as DMP, CRM etc with the Beemray platform.
General Best Practices
- Being more generic will help you target more users and draw more useful divisions between user segments.
- For example, rather than capturing a separate event for reading each of 50 different articles, it would be more effective to capture simply reading an article as an event.
- If you over segment your user data, your findings will lose statistical significance and will not guide the development of your app and marketing initiatives as effectively.
- You will “miss the forest for the trees” when evaluating user-trend data.
- Events should be tied directly to your marketing and conversion goals.