Caliper Get access

Early access preview

Data and privacy

Caliper measures attribution by asking the person, not by tracking the device. That single design choice determines its entire privacy posture.

First-party by construction

The answer to “how did you hear about us” comes directly from the person using your app. Because of that, Caliper does not read the IDFA on iOS or the Google Advertising ID on Android, does not fingerprint the device, and does not require an AppTrackingTransparency prompt to function. There is no device graph, no probabilistic match, and no cross-app identifier in the loop. The data is first-party by construction — it is your customer telling you something, recorded in your own app.

This is a different category of data from network-claimed attribution. A SKAN postback or a fingerprint is an inference about where an install came from; a Caliper answer is a statement from the person who installed it.

What is collected

Caliper collects two things:

It does not collect advertising identifiers, does not build a profile across apps, and does not capture free text outside the constrained auto-suggest field.

Data residency

Caliper data is stored in the United States.

Retention, export, and deletion

Your responses are yours. They are exportable on demand — through warehouse export on Growth+ plans, or by request — in a structured form you can load and join. They are deletable on request, including deletion tied to a specific install for downstream-erasure obligations. Caliper does not sell your data and does not pool it across customers; one customer’s responses are never mixed into another’s reporting or any shared model.

Relationship to SKAN and AdAttributionKit

A Caliper answer is deterministic at the install level: it is reported by the person and attached to that one install, independent of how many other installs a campaign produced. That is structurally different from SKAdNetwork and AdAttributionKit, where attribution depends on crowd-anonymity tiers — conversion values can be coarsened or withheld when a campaign does not clear a privacy threshold, and timers and aggregation blur which install did what.

Caliper does not compete with or replace those frameworks; it sits beside them. Where SKAN tells you, at an aggregate and sometimes coarsened level, which campaigns drove installs, Caliper tells you per install what the person says drove them — including the hard-to-measure channels (a podcast, a friend, a creator) that ad-network attribution cannot see at all. Because it is self-report attached to a single install, it does not depend on a crowd-anonymity threshold being met, so the signal does not degrade when volume is low.

Get access