Assists: Frequently Asked Questions (FAQ)
This article provides answers to common questions about Singular's Assists feature, including its core goals, technical logic, and how to leverage user-level data for deeper insights
Table of Contents
General Questions
A: The primary goal is to provide multi-touch attribution (MTA) visibility to help marketers understand the full user journey and the real impact of channels, especially for mid-to-upper funnel activity like CTV and video networks. It also helps quantify the overlap between channels and provides validation for budget allocation decisions.
A: Singular tracks and reports the Last Attributing Touchpoint plus up to four additional contributing touchpoints (T-1 to T-4) for user-level data. This is a default of 4, but additional touchpoints can be configured upon request.
A: No, currently Assists only supports installs and not re-engagements.
A: Assits and MTA is included for all unified accounts, across all tiers.
A: No, currently Assists only supports mobile installs (not web).
A: The Assist/MTA feature was enabled for all unified orgs starting Dec 1, 2025. Backfills are not available and advertisers can expect to see data doing forward from Dec. 1, 2025.
A: Yes, data from SANs (like Facebook and Google) is included in the user-level reporting for Assists. However, due to how SANs work (only providing the last touchpoint in the claim response), Singular only has the last touchpoint for them. Deduping on the source level helps cover related issues.
Technical Details and Logic
A: The core of the MTA calculation involves a basic modeling approach. It uses configurable weights (set per organization) for Single Attributed Installs (SAI), Co-attributed Installs (CAI), and Assisted Installs (AAI), ensuring that the total modeled installs equal the total last-touch attributed installs.
The default formula uses the penultimate touchpoint (T-1) for partial credit calculation. The formula is:
MTA Installs = (SAI × 1.0) + (CAI × 0.3) + (AAIT-1 × 0.7)
For a source to be considered an "Assist," its touchpoint must be within the T1-T4 window and not have won the Last Touch Attribution.
A: Singular dedupes per source per install. This means that for the T1-T4 window, we only count one touchpoint type (click or view) per source. If a source has both a click and a view in that window, the click is prioritized for the assist count.
A: No, the attribution windows defined in your Partner Config are not changed by this feature. The touchpoints collected for Assists operate within a separate framework for tracking the full user journey, but this is not changing the main attribution window logic.
User-Level Insights
A: Yes, user-level transparency is a key advantage of Singular's offering. You can access granular details, including how many touchpoints each install (Assist Touchpoint Count) had and what was the sequence of the touchpoints (Assist Level).
A: User-level data can be accessed by export logs or shared to your data warehouse via Data Destinations (ETL). Assist user-level data is not currently supported via BI postbacks.
A: User-level data provides granular transparency by detailing every contributing touchpoint for a given install, including the standard breakdowns like source, campaign, touchpoint type, timestamp, and attribution method. Additionally, it includes:
- Assist Level: The sequence of touchpoints in the user journey.
- Assist Touchpoints Count: The total number of unique sources for a given install. This is reported only on the row where the Assisted Level is 0.
- Install ID: A unique identifier that allows you to group all relevant touchpoints for a single install.
A: You can use user level data to further analyze what is the specific partner cross over. And you can use user level data to understand the full funnel (which touchpoint came last, and then penultimate, etc).
A: To be able to generate the report to review the partner cross over you need to first set up the data structurally so that you can have all the 0 touchpoints in one colunm and touchpoing 1,2,3 in seperate colunms. Which means you need to flatten the data assist data. You can do so manuallyvia google sheets for example. You can run an export logs report with the correct app, dates, etc import it into GoogleSheets and then run a script.