MessageGears Segment is a visual segmentation product that is built to directly utilize a brand’s data warehouse and activate recipients through third-party destinations. Read more to understand common usage of the platform, as well as technical documentation.
What is it?
MessageGears Segment is a visual segmentation product that is built to directly utilize a brand’s data warehouse and activate recipients through third-party destinations.
What does it do?
Segment is essentially a visual representation of a rules engine that allows marketers to categorize and enrich their recipients post data extraction. A query is set up to create the Starting Population for the Blueprint and then the user is able to visually create rules or evaluation steps for their given needs. As the data is extracted from the database from the supplied query it is streamed through the rules set up by the user. Each of these rules is mutually exclusive and allows for labels to be added to each recipient record based on their journey through the Blueprint. These labels can then optionally be recorded back to the database through the MessageGears Audience Recording feature
What is mutual exclusivity?
The Blueprint is going to be evaluated with each node or segment being mutually exclusive. Mutual exclusivity is placed into the product to ensure that a recipient qualifies for at most 1 node at any given level, and is not duplicated across several nodes. The priority of exclusivity is left to right, meaning that if a recipient qualifies for the leftmost node, that recipient will not be evaluated against any other nodes on that level. If the recipient does not qualify for the leftmost node, then the recipient will be evaluated against the node to the right.
What happens during a count?
During a count initiated by the user through the application the Blueprint will go through it’s full process. This means the data will be extracted from the database and then streamed through the rules set up by the user. An upcoming initiative slated for late Q4 2020 or early Q1 2021 is currently in development to allow for the Blueprint to count much more rapidly. The intent is to allow the users to optionally use a cached representation of their data and parallel processing to allow for quick counts as they are building their Blueprint. This combined with other improvements done to the actual rules processing will allow for quick feedback on an in development Blueprint.
What happens on Launch?
During a Campaign Launch a Blueprint will always go through the entire extraction and evaluation process to ensure that the campaign is using the most current data from the consumer’s database. For a given campaign launch, users are able to filter the Blueprint in order to decide what segments to include or exclude for the given campaign launch.
How do extraction and evaluation perform?
The Blueprint performance is highly dependent on the customer’s environment and the Blueprint itself. If you are using a cloud based data warehouse such as Snowflake or Google BigQuery, approximately 60% of the processing time will be the evaluation of the Blueprint. This number is going to fluctuate depending on the complexity of the Blueprint, the amount of customization in the Blueprint through using Freemarker, and the environment that the MessageGears application is hosted in.
Target Groups utilize a simple query builder UI to allow users to define segments by creating targeting criteria that a user must meet. If a user meets all of the targeting criteria, they will qualify for that segment and then, due to mutual exclusivity, move onto the next level of evaluations. Target Groups will automatically generate an “Everything Else” node. This node is a catch all for any recipient that did not match any of the criteria sets for segments evaluated on that level. The amount and complexity of Target Groups will impact the Blueprint performance.
Random Split allows users to create pseudo-random percentage based buckets of users. Random split creates logically weighted “buckets” and uses a psuedo-random algorithm to determine which bucket each recipient should be placed in based on the weight assigned by the user within the UI. The end result is a predictable size of each bucket and non-predictable assignment of each recipient. The performance of split is based on the size of the starting population as well as the number of buckets being created.
Merge allows users to funnel any leaf node on the tree into a single node. This allows users to group together any combination of segments.
Filter allows users to dissect Target and Merge nodes. For example if I have a target group that segments users by loyalty level (Diamond, Platinum, Gold, Silver). I can then Filter out Diamond and Platinum into their own branch and Gold and Silver into another. This is useful when there is a step that creates numerous segments but a user wants to only continue operating on a subset of them. Filter is purely a representation tool and does not consume any additional processing power.
Activation Through 3rd Parties
The Segment tool offers connectors to multiple 3rd parties such as Airship, Facebook, and Google. MessageGears follows our partners best practices for data upload and can meet or exceed our partners’ throughput demands. For more information, contact email@example.com.