Attribution Method Exposure

Prev Next

Overview

When analyzing campaign performance, knowing which touchpoint receives credit for a conversion is only part of the picture. Just as important is understanding how and why that touchpoint was attributed.

Attribution method exposure gives you visibility into the reasoning behind each attribution decision. Rather than simply showing that a click from a particular ad network led to an install, attribution method fields reveal the underlying methodology. This tells you whether the match was made through a direct device identifier, a partner-reported signal, or a privacy framework.

Branch surfaces this information through two fields available in your Last Attributed Touch Data (LATD):

  • LATD attribution method

  • LATD attribution method type

Together, these tell you both the specific signal that produced the attribution match and the broader category of methodology it belongs to.

Benefits

The transparency from attribution method exposure is valuable for several reasons. It helps you:

  • Assess the reliability of your attribution data.

  • Understand how privacy changes are affecting your attribution in practice.

  • Have a starting point for diagnosing changes if counts look different than expected.

How attribution is decided

Attribution matches fall into a few broad categories based on how a conversion is linked to a touchpoint. Understanding these categories helps you interpret the values you’ll see in the LATD attribution method and LATD attribution method type fields.

Deterministic attribution

A deterministic match occurs when a known identifier directly links a conversion to a touchpoint. This is the highest-confidence form of attribution. There is a concrete, one-to-one connection between the ad interaction and the conversion event. Examples include matching on a device advertising ID (such as IDFA or GAID) or a developer-provided user identifier.

Deterministic matches can be performed by Branch or reported by a self-attributing network (SAN) like Google or Meta.

Probabilistic attribution

When a direct identifier isn’t available (due to privacy restrictions, user opt-outs, or platform limitations), Branch can use probabilistic modeling to infer a match. Branch’s probabilistic matching is powered by Predictive Modeling (PREM), which analyzes patterns in device-level metadata to estimate the most likely connection between a click and a conversion.

Probabilistic matches vary in confidence. A strong probabilistic match indicates a high degree of certainty based on the available signals, while a weak probabilistic match has a lower degree.

External SAN attribution

In some cases, a SAN asserts the attribution win using its own measurement infrastructure. These partners use a range of technologies perform this attribution, including partner-specific click identifiers, privacy-preserving measurement frameworks, and modeled conversions. Depending on the technology used, the underlying match may be deterministic or modeled by the partner.

The LATD attribution method and LATD attribution method type fields give you visibility into which specific technology or framework the SAN used, so you can understand the nature of the match event when it originates from a partner.

Branch fields

Branch exposes attribution method information through two LATD fields. These fields are available in your export data and help you understand both the specific signal and the broader methodology behind each attribution decision.

LATD attribution method

LATD attribution method identifies the specific signal or mechanism used to link a conversion event to a marketing touchpoint. It represents the actual identifier, framework, or identity system the produced the match.

Possible values for LATD attribution method

Description

ADVERTISING_ID

Deterministic match using device advertising identifiers (IDFA, GAID).

GCLID

Google click identifier used to attribute conversions to Google Ads interactions.

GBRAID

Privacy-safe Google identifier used in place of device identifiers under privacy restrictions.

AEM

Meta's Aggregated Event Measurement (AEM) framework used for privacy-preserving attribution.

ICM_ODM

Google’s Integrated Conversion Measurement / On-Device Measurement (ICM/ODM) modeled attribution system.

PERSONA

Branch identity graph matching across sessions, devices, or channels.

CROSS_DEVICE

Attribution derived from probabilistic cross-device modeling.

DEVELOPER_IDENTITY

Deterministic attribution using developer-provided user identifiers.

PUB_SUITE_SAN

Attribution reported by a publisher suite SAN partner.

MIR

Partner-specific identifier used to associate a click with a downstream conversion.

LATD attribution method type

LATD attribution method type classifies how the attribution signal was generated or validated. It describes the broader methodology behind the match, grouping methods into categories based on reliability and mechanism.

Possible values for LATD attribution method type

Description

PARTNER_DETERMINISTIC

Deterministic attribution confirmed by an advertising partner using known identifiers.

PARTNER_CLICK_IDENTIFIERS

Attribution based on partner-provided click identifiers embedded in ad interactions.

PARTNER_MODELLED

Attribution performed by a partner using probabilistic models.

PARTNER_PRIVACY_FRAMEWORK

Attribution derived from a partner's privacy-preserving framework.

BRANCH_DETERMINISTIC

Deterministic attribution performed by Branch using its identity graph or known identifiers.

BRANCH_MODELLED

Attribution performed by Branch using probabilistic models (PREM).

Access attribution method data

You can access LATD attribution method and LATD attribution method type through the following channels.

Ads Analytics

In legacy Branch, navigate to Ads Analytics > Activity.

The LATD attribution method and LATD attribution method type fields are available as columns in the activity view, allowing you to see attribution method information alongside your conversion data.

To learn more about Ads Analytics, visit our guide.

Caution

Please note that these fields are not yet available in new Branch.

API exports

When using Branch APIs, use the following fields for LATD attribution method and LATD attribution method type:

  • last_attributed_touch_data_plus_attribution_method

  • last_attributed_touch_data_plus_attribution_method_type

Both fields are available in your log-level export data through the Custom Exports API and the Scheduled Log Exports API.

FAQ

What's the difference between LATD attribution method and LATD attribution method type?

LATD attribution method tells you the specific signal or identifier that produced the attribution match (e.g., ADVERTISING_ID, GCLID, AEM). LATD method type tells you the broader category of methodology that signal belongs to (e.g., PARTNER_DETERMINISTIC, BRANCH_MODELLED). Think of LATD attribution method type as the "category" and LATD method as the specific "instance" within that category.

Why is the LATD attribution method or LATD attribution method type field empty or unpopulated?

These fields may be unpopulated for older conversions that occurred before attribution method exposure was available.

Can a single LATD attribution method appear under more than one LATD attribution method type?

Yes. For example, a SAN partner may use the same underlying technology in both a deterministic and a modeled capacity, which would result in the same LATD attribution method value appearing with different LATD attribution method type values.