I’ve built a mobile SDK with new user targeting architecture that improves user privacy, while also giving developers more precision and criteria to target users. I know that sounds super unintuitive, but by inverting standard user targeting architecture becomes possible. The SDK is live on Github, and we have full docs to try it out (links below).
Here’s the high level idea of how it works (more detail in in blog post linked below):
Zero data collection: the data flow is unidirectional from server to client. The client never needs to send information to the server for targeting
Powerful on-device logic engine: you can write targeting logic with conditional strings using powerful but familiar syntax. It supports logical operators, functions, arithmetic, set operations, dates, random number generation, database queries, and more! This runs completely locally on each user’s device.
Rich build-in targeting properties: 100 properties you can query, covering device information, user context, sensors, location, permissions, connectivity, peripherals, locale, app info, and much more.
Local event database: each client builds a rich database of user engagement history (app launches, session times, terminations, and user actions, custom events, etc). You can query this and target users, without streaming interaction data to any server.
Local database for property history: allows you to see if the current state is exceptional or the norm for this user.
Logic isn’t hardcoded: you can still update your logic over the air anytime, without app updates. You simply push new logic to clients instead of updating server-side logic.
Since everything is local and data never leaves device, we can offer more precise targeting criteria, without the additional scaling complexity, privacy concerns, costs, or legal concerns that come with server-side data collection of contextual data. We can do all this without IDFA or device fingerprinting.
I’ve built a mobile SDK with new user targeting architecture that improves user privacy, while also giving developers more precision and criteria to target users. I know that sounds super unintuitive, but by inverting standard user targeting architecture becomes possible. The SDK is live on Github, and we have full docs to try it out (links below).
Here’s the high level idea of how it works (more detail in in blog post linked below):
Zero data collection: the data flow is unidirectional from server to client. The client never needs to send information to the server for targeting
Powerful on-device logic engine: you can write targeting logic with conditional strings using powerful but familiar syntax. It supports logical operators, functions, arithmetic, set operations, dates, random number generation, database queries, and more! This runs completely locally on each user’s device.
Rich build-in targeting properties: 100 properties you can query, covering device information, user context, sensors, location, permissions, connectivity, peripherals, locale, app info, and much more.
Local event database: each client builds a rich database of user engagement history (app launches, session times, terminations, and user actions, custom events, etc). You can query this and target users, without streaming interaction data to any server.
Local database for property history: allows you to see if the current state is exceptional or the norm for this user.
Logic isn’t hardcoded: you can still update your logic over the air anytime, without app updates. You simply push new logic to clients instead of updating server-side logic.
Since everything is local and data never leaves device, we can offer more precise targeting criteria, without the additional scaling complexity, privacy concerns, costs, or legal concerns that come with server-side data collection of contextual data. We can do all this without IDFA or device fingerprinting.
SDK docs: https://docs.criticalmoments.io/quick-start
More details on the architecture: https://criticalmoments.io/blog/how_to_target_users_without_...
Github: https://github.com/CriticalMoments/CriticalMoments