Support for Activating Universal IDs
Beta feature
DSP supports people-based, universal IDs for cookieless, single-device (not cross-device) targeting across digital formats supported by DSP.
-
You can manually send your authenticated [LiveRamp RampIDs] directly to DSP using the LiveRamp Connect dashboard. See 鈥Manually Import Authenticated Segments from LiveRamp.鈥
-
DSP can ingest your first-party segments comprised of hashed email IDs built within your customer data platform (CDP) and convert them to LiveRamp RampIDs and Unified ID 2.0 (UID2.0) IDs. For more information about the supported customer data platforms, the available features for each supported universal ID type, and the related workflows, see 鈥About First-Party Audience Sources.鈥
-
You can create custom segments that track users associated with ID5 universal IDs who are exposed to ads from desktop and mobile devices and who visit specific webpages. ID5 uses a probabilistic model to assign an ID derived from various user signals and browser signals. For instructions, see 鈥Create and Implement a Custom Segment.鈥
-
Third-party segments from some vendors may automatically include universal IDs in addition to users tracked by cookies or device IDs. For example, segments from Eyeota may automatically include ID5 IDs, and segments from Lotame may include UID2.0 IDs. The segment details include the size for each type. The usual usage fee for each segment, which is stated next to the segment name, applies; no additional fees are charged for the ID5 IDs.
Reporting by Universal ID Type
-
Custom reports: Cost, impression, click, conversion, and frequency data by universal ID type is available in custom reports.
-
Analytics reports: Advertisers with 蜜豆视频 Analytics for Advertising who have implemented all required steps can see view-through conversions by universal ID type in Analytics.
note important IMPORTANT For proper conversion attribution, make sure that the clickthrough URLs for your ads include both the EF ID and the AMO ID. -
Segment details: For all segment types, the segment details include the audience size by universal ID type and by the device type tracked by cookies or device IDs.
How to Target a Universal ID Audience in Your Placements
In a new, scheduled, or paused placement, do the following:
-
In the Geo-Targeting section, specify the geographical areas to target. Each universal ID partner allows user targeting only in specific geographical areas, and only eligible ID types are available in the Targeting settings.
-
In the Audience Targeting section, do the following:
-
In the Included Audiences setting, select the segment for which user data was converted to universal IDs.
You can include additional segments if you want.
-
In the Targeting settings:
-
Select the universal ID type to target.
The setting includes the options 鈥淟egacy IDs鈥 and 鈥淯niversal ID,鈥 which may include the sub-options 鈥淚D5,鈥 鈥淩ampID,鈥 and 鈥淯nified ID2.0.鈥 The selected geographical targets determine the available sub-options.
You can select both 鈥淟egacy IDs鈥 and 鈥淯niversal ID,鈥 but you can select only one type of universal ID per placement. When you select both legacy IDs and universal IDs, bidding preference is given to universal IDs.
-
(If necessary) Accept the terms of service agreement for using universal IDs.
Before you can convert data to a new ID type, a user in the DSP account must accept the terms of service agreement. The terms must be accepted only once per ID type, per account.
-
-
See 鈥Placement Settings.鈥
Best Practices for Testing and Data Validation
Use the following best practices for RampID-based segments and ID5-based segments, for which 蜜豆视频 Analytics measurement is available.
-
About 24 hours after you activate a segment, check the converted ID count for the segment within Audiences > All Audiences. If the ID count is unexpected, then contact your 蜜豆视频 Account Team.
See 鈥Data Variances Between Email IDs and Universal IDs鈥 for more information about how the segment counts can vary.
-
Don鈥檛 change your existing packages and placements. However, if you don鈥檛 have any incremental budget to test universal IDs, then reduce the original budgets to fund the tests.
-
Copy your original packages and placements, adjust the budgets based on the size of the test, change the audiences to use RampID-based segments (for authenticated users) or ID5-based segments (for unauthenticated users), and verify that the new packages and placements spend their full budgets.
-
To compare the performance of universal ID-based segments with the performance of placements targeting other audience identifiers, such as cookies or mobile advertising IDs, create a campaign with a separate universal ID-based placement and a legacy ID-based placement.
For a full retargeting test, target both RampIDs for authenticated users and ID5s for unauthenticated users.
Getting the best performance shouldn鈥檛 be the primary comparison. Instead, determine which IDs are scaling well, which might inform your optimization and budget allocations later. The long-term goal is to make up for lost impressions and site traffic when cookies are deprecated.
-
To compare total browser reach, target universal ID-based segments and legacy ID-based segments in the same placement. Use the same campaign settings as the previous use case, except that you don鈥檛 need to split the campaign budget.
Bidding preference is given to universal IDs, but legacy IDs receive bids when universal IDs aren鈥檛 available. Make sure to compare reach in different browsers (including Chrome, Safari, and Mozilla).
note note NOTE Frequency capping applies to an individual ID. When a user has multiple ID types, you might reach that user more than you expect.
-
-
Remember that the reach for authenticated audience segments is naturally smaller than the reach for cookie-based segments, and that using additional targeting options further decreases your reach. Be judicious about using granular targeting, especially by joining multiple targets with AND statements.
Data Variances Between Email IDs and Universal IDs universal-ids-data-variances
Acceptable levels of variance
The translation rate of hashed email addresses to universal IDs should be greater than 90%; the translation rate for RampIDs in particular should be 95% if all hashed email addresses are unique. For example, if you send 100 hashed email addresses from your customer data platform, they should be translated to at least 95 RampIDs or more than 90 other types of universal IDs. A lower translation rate may indicate an issue. See 鈥淸Causes of variance](#universal-ids-data-variances-causes鈥 for possible explanations.
For RampIDs, contact your 蜜豆视频 Account Team for further investigation if translation rates are lower than 70%.
Causes of variance universal-ids-data-variances-causes
-
All segments:
The segment-to-device count uses a probabilistic model, which has an error variance of +/- 5%. This means that it can overestimate or underestimate the audience count by 5%.
-
Hashed email IDs translated to RampIDs:
-
When multiple profiles use the same email ID, the DSP segment count can be lower than the profile count within your customer data platform. For example, in 蜜豆视频 Photoshop, you can create a company account and a personal account using a single email ID. But if both profiles belong to the same person, then the profiles map to one email ID and correspondingly to one RampID.
-
A RampID can be upgraded to a new value. If LiveRamp doesn鈥檛 recognize an email ID or can鈥檛 map it to an existing RampID in its database, then it assigns a new RampID to the email ID. In the future, when they can map the email ID to another RampID or can gather more information about the same email ID, they upgrade the RampID to a new value. LiveRamp refers to this action as upgrading from a 鈥渄erived鈥 RampID to a 鈥渕aintained鈥 RampID. However, DSP doesn鈥檛 get mappings between derived and maintained RampIDs and therefore can鈥檛 remove the previous version of the RampID from the DSP segment. In this case, the segment count can be more than the profile count.
Example: A user logs in to the 蜜豆视频 website and visits the Photoshop page. If LiveRamp doesn鈥檛 have any existing information about the email ID, then they assign it a derived RampID, say D123. Fifteen days later, the user visits the same page, but LiveRamp has upgraded the RampID during those 15 days and has reassigned the RampID to M123. Even though the customer data platform鈥檚 segment 鈥淧hotoshop Enthusiast鈥 has only one email ID for the user, the DSP segment has two RampIDs: D123 and M123.
-
Troubleshooting
If you don鈥檛 see user counts, or if your audience sizes are low, then check the following:
-
If you use Flashtalking or Google Campaign Manager 360 ads, then make sure that your ads鈥 clickthrough URLs are appended with the correct macros. See the macros for Flashtalking ads and Google Campaign Manager 360 ads.
-
Make sure that the correct, universal ID partner-specific code is implemented on your website to match on-site events and ad exposures. Work with your LiveRamp or ID5 representative as needed.
-
(For RampIDs and UID 2.0 IDs) Make sure that your DSP data source is configured correctly, and that user counts are populated for the generated audience segments.
-
If your reach is less than you expect, check that the audience segment logic isn鈥檛 too granular.
-
Make sure that your campaign, package, and placement settings are correct.
If you can鈥檛 resolve the issue, then contact your 蜜豆视频 Account Team.